You are on page 1of 645

ELEMENTARY

DIFFERENTIAL EQUATIONS
SIXTH EDITION
EDWARDS & PENNEY
ABOUT THE COVER. This image illustrates
the trajectory of a moving point whose space
coordinates satisfy (as functions of time) the
Rossler system of differential equations that is discussed on page 553, and which
originated in studies of oscillations in chemical reactions. In its motion along its
trajectory the point may appear to spiral repeatedly around a set - the so-called
Rossler band - that somewhat resembles a (twisted) Mobius strip in space. To
portray the progress of the moving point, we can regard its trajectory as a necklace
string on which beads are placed to mark its successive positions at fixed increments
of time (so the point is moving fastest where the spacing between beads is greatest).
In order to aid the eye in following the moving point's progress, the color of the beads
changes continuously with the passage of time and motion along the trajectory. As the
point travels around and around the band, it may be observed to drift radially back and
forth across the band in an apparently unpredictable fashion. Two points that start
from nearby initial positions may loop around and around the band somewhat in
synchrony, while moving radially in quite different ways, so that their trajectories
diverge appreciably with the passage of time. This illustrates the phenomenon of
chaos, in which tiny differences in initial conditions can result in great differences in the
resulting situations some time later. Further discussion of chaos associated with
differential equations can be found in Section 7.6. Throughout this textbook
computer-generated graphics are used to portray numerical and symbolic solutions of
differential equations vividly and to provide additional insight.
PEARSON
Prentice
Hall
Upper Saddle River, NJ 07458
www.prenhall.com
m

z
ISBN-13: 978-0-13-239730-8
ISBN-10: 0-13-239730-7
9 780132397308
EDWARDS
&
PENNEY
m
r
m
s:
m
Z
>
::
-
C
-
-
-
m
::
m
Z
-
-

r
m
/
c

-
o
z

SIXTH
EDITION
ELE NT Y DIFFE E TI L
E UTI 0 N S
SIXTH EDITION
C. HENRY
EDWARDS
&
DAVID E.
PENNEY





,
.
.
LLLNL11
LrLC1FL
cCUA|ClS
cLLNt1/!`
!Lt1l7L
cO1IIo
5lxth ldltlon
C. tD
}
ldwurds
Duvd l. tDDt
]
1e|n:ve:s:jo(Geo:g:a
u:ththeass:staneeo(
Duvd Cuvs
8alu:n-wallaeecollege
UpperSadd|euiver, N]o
LibraryefCengressCataleging-in-FublicatienDataenhle.
r+.......o..-....c.,..-.s..--.-r-,.---..-,.-++..-.-+v..--..... Marci J. Horon
s--...r+....Holy Stark
r+.............-.Jennier Lonschein
s--...v.-.,.-,r+....Scott Dianno
e..+.....-r+....Irwin Zucker
..o..-.....-+c..-.o-..,--.Kenny Beck
..r+....Tomas Ben/atti
v.-........-,v.-.,-.Aln Fischer
v.-........-,s.,-.Lisa McDowel
s--...v...-..-,v.-.,-.Tm Galigan
200?, 2004, 2000, l 99-,e-....-r+......-|-.
e-....-r+......-|-.
u,,-.s.++.-s..-.-i-..-,0745?
....,-...-.-..-+.,......-..-....,--
.-,..+..-+.-.-,.....-,.-,-.-.
..-...,-......-. -....-,....--,.-...--.
TRADEMARK INFORMATION
v.ts. ...-,...-.-+...+-......--v..- w....|-.
r..v.ts,..+....-.......-,.-..-..-....
.--v..-w....|-.
,,.-a...o...-
.....v0l 70-209?us
.-. 50?-47-7000
r.50?-47-7 l 0l
r... .-..e..-......
w-- ..-......
v.,.-....-,...-.-+...+-.....w..-....v.,.-|-.
v..--.........-,...-.-+...+-.....w.....s-.-...-|-.
e..-.-+.-.--u-..-+s...-...-....
l 0 9 ? 7 5 4 2 l
lUN U1ddddU
e-....-r+......-t.o London
e-....-r+......-........e.,t...-+Sydney
e-....-r+......-s. -,.,..-e.-t.+
e-....-r+......-...-...t.+ Hong Kong
e-....-r+......-c.-.+.t.+ Toronto
e-....-r+......-+-v-...s +-cv
e-....-r+......-i.,.-Toko
e-....-r+......-v...,...e.-t.+

`|CIOCC vll
CHAPTER

CHAPTER

CHAPTER

First-Order Diferential Equations 1


J.J Di fferenti al Equati ons and Mathemati cal Model s
J .z I ntegral s as General and Parti cul ar Sol uti ons I
J . Sl ope Fi el ds and Sol uti on Curves I
J . Separabl e Equati ons and Appl i cati ons Z
J. Li near Fi rst-Order Equati ons 4
J. Substi tuti on Methods and Exact Equati ons
J.7 Popul ati on Model s /4
J. Accel erati on-Vel oci ty Model s d
Linear Equations of Higher Order 1
z.J
z.z
z.
z.
I ntroducti on: Second-Order Li near Equati ons I
General Sol uti ons of Li near Equati ons I I
Homogeneous Equati ons wi th Constant Coeffi ci ents I Z4
Mechani cal Vi brati ons I
z.
Nonhomogeneous Equati ons and U ndetermi ned Coeffici ents I 4d
z. Forced Osci l l ati ons and Resonance I Z
z.7 El ectri cal Ci rcuits I /
z. Endpoi nt Probl ems and Ei genval ues I d
Power Series Methods 19
.J I ntroducti on and Revi ew of Power Seri es I 4
.z Seri es Sol uti ons Near Ordi nary Poi nts Z/
. Regul ar Si ngul ar Poi nts Z I d
. Method of Frobeni us: The Excepti onal Cases Z
. Bessel ' s Equati on Z4d
. Appl i cati ons of Bessel Functi ons Z/
V
V| Contents
CHAPTER

CHAPTER

CHAPTER

CHAPTER

Laplace Transform Methods X


.J Lapl ace Transforms and I nverse Transforms Z
.z Transformati on of I ni ti al Val ue Probl ems Z//
. Transl ati on and Parti al Fracti ons Zd
. Deri vati ves, I ntegral s, and Products of Transforms Z/
. Peri odi c and Pi ecewi se Conti nuous I nput Functi ons 4
. I mpul ses and Del ta Functi ons 1
Linear Systems of Diferential Equations X
.J Fi rst-Order Systems and Appl i cati ons Z
.z The Method of El i mi nati on d
. Matri ces and Li near Systems 4/
. The Ei genval ue Method for Homogeneous Systems
. Second-Order Systems and Mechani cal Appl i cati ons d1
. Mul ti pl e Ei genval ue Sol uti ons
.7 Matrix Exponenti al s and Li near Systems 4/
. Nonhomogeneous Li near Systems 4Z
Numerical Methods 9
.J Numeri cal Approxi mati on: Eul er's Method 4
.z A Cl oser Look at the Eul er Method 44Z
. The Runge-Kutta Method 4
. Numeri cal Methods for Systems 44
Nonlinear Systems and Phenomena
7.J Equi l i bri um Sol uti ons and Stabi l i ty 4d
7.z Stabi l i ty and the Phase Pl ane 4dd
7. Li near and Al most Li near Systems
9
7. Ecol ogi cal Model s: Predators and Competi tors 1
7. Nonl i near Mechani cal Systems Z
7. Chaos in Dynami cal Systems 4Z
References for Further Study
Appendix: Existence and Uniqueness of Solutions ;
Answers to Selected Problems /
Index |1
1tf7'T
1
he evo|utlon ofthe presenttext ln successlve edltlons ls based onexperlence
teachlng the lntroductory dlfferentla| equatlons course wlth an emphasls on
conceptua| ldeas and the use ofapp|lcatlons and proects to lnvo|ve students ln
actlve prob|em-so|vlng experlences. At varlous polnts our approach reHects the
wldespreaduseoftechnlca|computlngenvlronments|lkeM.p/e,M./e-./...and
MA1IAu for the graphlca|, numerlca|, or symbo|lc so|utlon ofdlerentla| equa-
tlons. Neverthe|ess, wecontlnuetobe|levethatthetradltlona|e|ementaryana|ytlca|
methods ofso|utlonare lmportantforstudentsto |eam anduse. Onereasonlsthat
eectlve andre|lab|euse ofcomputermethods oftenrequlres pre|lmlnary ana|ysls
uslng standard symbo|lc technlques, the constructlon ofa rea|lstlc computatlona|
mode|oftenls basedonthestudyofaslmp|erana|ytlca|mode|.
-------------------------------------
Whl|ethesuccessfu|featuresofprecedlngedltlonshavebeenretalaed,theexposl-
tlonhasbeenslgnlncant|yenhancedlneverychapterandlnmostlndlvldua|sectlons
ofthe text. Bothnewgraphlcs andnewtexthavebeen lnsertedwhereneededfor
lmproved student understandlng of key concepts. However, the so|ld c|ass-tested
chapter and sectlon structure ofthe book ls unchanged, soc|ass notes andsy||abl
wl||notrequlrerevlslonforuseofthlsnewedltlon. Thefo||owlngexamp|esofthls
revlslon l||ustrate the way the |oca| structure ofthe texthas been augmentedand
po|lshedforthlsedltlon.
Chapter 1: NewFlgures1. 3. 9 and1. 3.10 showlngdlrectlonne|dsthatl||us-
tratefal|ureofexlstenceandunlquenessofso|utlons(page24); newProb|ems
34 and 35 showlngthatsma||changesln lnltla|condltlonscanmakeblgdlf-
ferencesln resu|ts,butblgchangesln lnltla|condltlonsmaysometlmesmake
on|ysma||dlerenceslnresu|ts(page30); newRemarks l and2 c|arlfylngthe
conceptoflmp|lcltso|utlons (page 35); new Remarkc|arlfylngthemeanlng
ofhomogeneltyfornrst-orderequatlons(page 61); addltlona|detal|s lnserted
ln the derlvatlonofthe rocketpropu|slonequatlon (page95), and new Prob-
|em5 lnserted to lnvestlgate the |lftopause ofa rocket on the |aunchpad
sometlmesobservedbeforeb|asto(page97) .
Chapter Z: New exp|anatlon ofslgns and dlrectlons oflntema| forces ln
mass-sprlng systems (page 101); new lntroductlon ofdlerentla| operators
and c|arlncatlon ofthe a|gebraofpo|ynomla| operators (page 127); new ln-
troductlon and l||ustratlon ofpo|ar exponentla| forms ofcomp|ex numbers
(page132); fu||erexp|anatlonofmethod ofundetermlnedcoemclentslnEx-
amp|es1 and3 (page149-150); newRemarks1 and2 lntroduclngshootlng
termlno|ogy,andnewFlgures2. 8. 1 and2. 8. 2 l||ustratlngwhy someendpolnt
V
VIII Preface
va|ueprob|emshavelnnnlte|ymanyso|utlons,whl|eothershavenoso|utlons
at a|| (page181); new Flgures 2. 8. 4 and 2. 8. 5 l||ustratlng dlerenttypes of
eigenfunctlons(pages183-184) .
Chapter 3: New Prob|em 35 on determlnatlon of radll ofconvergence of
power serles so|utlons ofdlerentla| equatlons (page 218); new Examp|e 3
ustbeforethesubsectlonon|ogarlthmlccaseslnthemethodofFrobenlus,to
l||ustratenrstthereductlon-of-orderformu|awlthaslmp|enon-serlesprob|em
(page239).
Chapter 4: Newdiscusslonc|arlfylngfunctlonsofexponentla|orderandex-
istence ofLap|ace transforms (page 273); new Remark dlscusslng the me-
chanics of partla|-nactlon decomposltlon (page 279); new much-expanded
dlscusslon ofthe proofofthe Lap|ace-transform exlstence theorem and lts
extenslon to lnc|ude theump dlscontlnultles that p|ay an lmportant ro|e ln
manypractlca|app|lcatlons(page286-287) .
Chapter 5: New Prob|ems 20-23 for student exp|oratlon ofthree-ral|way-
carssystemswlthdlerentlnltla|ve|ocltycondltlons(page392); newRemark
l||ustrating the re|atlon between matrlx exponentia| methods and the gener-
a|ized elgenva|ue methods dlscussedprevlous|y (page 416); new exposltlon
lnsertedatendofsectlontoexp|alntheconnectlonbetweenmatrlxvarlatlon
ofparametershereand(sca|ar)varlatlonofparametersforsecond-orderequa-
tlons dlscussedprevlous|ylnChapter3 (page427).
Chapter : New discusslon with new Flgures 6. 3.11 and 6. 3. 12 c|arlfylng
thedlerencebetweenrotatlngandnon-rotatlngcoordlnatesystemslnmoon-
earthorbltprob|ems(page473) .
Chapter 7: New remarks onphase p|ane portralts, autonomous systems,
and crltica| polnts (page 488-90); new lntroductlon of|lnearlzed systems
(page502); new 3-dimenslona|Flgures6. 5.18 and6. 5. 20 l||ustratlngLorenz
andRss|ertraectorles (page552-553).
Throughout the text, a|most 550 .o-pe ee.ea]es show students
vividplcturesofdlrectlonne|ds, so|utloncurves, andphasep|aneportraltsthatbrlng
symbo|lcso|utlonsofdlerentla|equatlonsto|lfe.
About15 .pp//../o-oa/esfo||owkey sectlonsthroughoutthetext. Thelr
puqoseistoaddconcreteapp|ledemphasl sandtoengagestudentsismoreexten-
slvelnvestlgatlonsthanaffordedbytyplca|exerclsesandprob|ems.
A so|ld-e/../e-p/.s/sprovldedwhereapproprlate(aslnChapter6 on
Numerlca|Methods)bythelnc|uslonofgenerlcnumerlca|a|gorlthmsanda|lmlted
numberofl||ustrativegraphlngca|cu|ator,BASIC,andMA1IAuroutlnes.
Organization and Content
Thetradltiona|organlzatlonofthlstextstl||accommodatesfreshnewmaterla|and
comblnatlonsoftoplcs. Forlnstance.

Thenna| two sectlonsofChapter1 (on popu|atlonsande|ementarymechan-


lcs)oeranear|ylntroductlontomathematlca|mode|lngwlthslgnlncantap-
p|lcatlons.

The nna| sectlon ofChapter 2 oers unusua||y ear|y exposure to endpolnt


Applications

Preface IX
prob|ems and elgenva|ues, wlth lnterestlng app|lcatlons to whlr|lng strlngs
andbuck|edbeams.
Chapter3 combinesacomp|eteandso|ldtreatmentoflnnnlteserles methods
wlthlnterestlngapp|lcatlonsofBesse|functlonslnlts nna|sectlon.
Chapter 4 combines a comp|ete and so|ld treatment of Lap|ace transform
methods wlth brlefcoverage of de|ta functlons and thelr app|lcatlons ln lts
nna|section.
Chapter 5 provldes an unusua||y Hexlb|e treatment of|lnear systems. Sec-
tlons 5.1 and 5. 2oeran ear|y, lntultlve lntroductlon to nrst-order systems
and mode|s. The chapter contlnues wlth a se|f-contalned treatment ofthe
necessary|lneara|gebra, and then presents the elgenva|ueapproachto|lnear
systems. Itlnc|udes an unusua| numberofapp|lcatlons (ranglng frombrlne
tanks toral|waycars)ofa||the varlous casesoftheelgenva|uemethod. The
coverageofexponentia|matrlces ln Sectlon5. 7l sexpandedfromear|leredl-
tlons.
Chapter 6 on numerlca| methods beglns ln Sectlon 6.1 wlth the e|ementary
Eu|er method for slng|e equatlons and ends ln Sectlon 6. 4wlth the Runge-
Kuttamethodforsystemsandapp|lcatlonstoorbitsofcomets andsate||ltes.
Chapter7 onnon|inearsystemsandphenomenarangesfromphasep|aneana|-
yslstoeco|oglca|andmechanica|systemstoanlnnovatlveconc|udlngsectlon
onchaos andblfurcatlon lndynamlca| systems. Sectlon7.6 presents an e|e-
mentary lntroductlontosuchcontemporarytoplcsasperlod-doub|lnglnblo-
|oglca|andmechanlca|systems, thepltchforkdlagram,andtheLorenzstrange
attractor(a||l||ustratedwlthvlvldcomputergraphlcs) .
Thl sbooklnc|udesadequatematerla|fordlerentlntroductorycoursesvary-
lngln |ength from a slng|e term to two quarters. The |ongerverslon, r/e-e.
n_ ee/./r,./os//soa. r./ei//e-s (0-13-600613-2), contalns
addltlona|chapters onFourlerserlesmethods andpartla|dlerentla|equatlons(ln-
c|udlngseparatlonofvarlab|esandboundaryva|ueprob|ems).
To samp|etherangeofapp|lcatlonslnthlstext, take a|ookatthefo||owlngques-
tlons.
Whatexp|alnsthecommon|yobserved|ag tlme between lndoorandoutdoor
dal|ytemperatureoscl||atlons!(Section1.5)
Whatmakesthedierencebetweendoomsdayandextlnctlonlna||lgatorpop-
u|atlons!(Section1.7)
Howdoaunlcyc|eandatwo-ax|ecarreactdlerent|ytoroadbumps! (Sec-
tlons2.6 and5. 5)
Whyareagpo|esho||owlnsteadofso|ld!(Sectlon3.6)
If a mass on a sprlng ls perlodlca||y struck with a hammer, how does the
behavlorofthe mass dependon the frequency ofthe hammer b|ows! (Sec-
tlon4.6)
Ifamovlngtralnhltstherearendofatralnofral|waycarsslttlngatrest,how
can lthappenthatustaslng|ecarls poppedothefrontendofthesecond
traln!(Section5. 5)
A Preface
Howcan youpredlctthetlmeofnextperlhe|lonpassageofanew|yobserved
comet!(Sectlon6. 4)
What determlnes whether two specles wl|| |lve harmonlous|y together, or
whethercompetltlonwl||resu|tln the extlnctlonofone ofthem andthe sur-
vlva|of theother! (Sectlon7. 4)
Why andwhen doesnon-|lnearlty|eadtochaoslnblo|oglca|andmechanlca|
systems!(Sectlon7. 6)
Applications and Solutions Manuals
Ackowledgments
Theanswersectlonhasbeenexpandedconslderab|ytolncreaseltsva|ueasa|em-
lngald. Itnowlnc|udestheanswers tomostodd-numberedprob|emsp|usagood
manyeven-numberedones. The 605-pageInstructor's Solutions Manual (0- I 3-
6006I 4-0) accompanylngthls book provldes worked-outso|utlonsformostofthe
prob|emslnthebook,andthe345-pageStudent Solutions Manual (0- I 3-6006I 5-
9) contalnsso|utlonsformostoftheodd-numberedprob|ems.
Theapproxlmate|y I5app|lcatlonmodu|eslnthetextcontalnaddltlona|prob-
|emandproectmaterla|deslgned|arge|ytoengage students lntheexp|oratlonand
app|lcatlonofcomputatlona|techno|ogy. Theselnvestlgatlonsareexpandedconsld-
erab|ylnthe320-page~pp//../osM../(0- I 3-600679-5)thataccompanlesthe
textandsupp|ementsltwlthabout30addltlona|app|lcatlonsmodu|es. Eachsectlon
lnthls manua|haspara||e| subsectlons UslngMap|e, UslngMathematlca,and
Uslng MA1IAuthat detal| the app|lcab|e methods and technlques ofeach sys-
tem,andwl||aord studentusers anopportunltytocomparethemerltsandsty|es
ofdlerentcomputatlona|systems.
Inpreparlngthlsrevlslonweprontedgreat|yfromtheadvlceandasslstanceofthe
fo||owlngverycapab|eandperceptlverevlewers.
RaymondA. C|aspad|e,
0/.es/-o)Me-p//s
SemlonGutman,
0/.es/-o)o//./o-.
Mlk|osBona,
0/.es/-o)i/o/a.
IrfanU|-Haq,
0/.es/-o)u/s.os/ i/.e.///e
Car|Lutzer,
ro./eseis/eo)e./o/oy
Slga|Gltt|leb,
0/.es/-o)M.ss../ses,n.-o/
It ls a p|easure to (once agaln) credlt Dennls K|etzlng and hls extraordlnary
Tpertlse forthe attractlvepresentatlonofboththe textandthe lnthls book.
Flna||y, but far from |east, I am especla||y happy to acknow|edge a new contrlb-
utorto thls eort, DavldCa|vls, who asslstedlnevery aspect ofthls revlslonand
contrlbutedtanglb|ytothelmprovementofeverychapterlnthebook.
C. H. E.
b.dwadS(m1ndSp1ng.COm
|



C1l/

cXump| e 1
First-Order
Differential Equations
1
he |aws ofthe unlverse are wrltten ln the |anguage ofmathematlcs. A|gebra
ls sufnclent to so|ve many statlc prob|ems, but the most lnterestlng natura|
phenomena lnvo|ve change and are descrlbed by equatlons that re|ate changlng
quantltles.
Because the derlvatlveax}a = ) , ; ofthefunctlon )l sthe rateatwhlch
the quantlty x = ), ; ls changlng wlth respect to the lndependent varlab|e , lt
ls natura| that equatlons lnvo|vlng derlvatlves are frequent|y used to descrlbe the
changlngunlverse. Anequatlonre|atlngan unknown functlonandoneormoreof
ltsderlvatlveslsca||edadiferential equation .
......
Thedlerentla|equatlon
ax
= x

a
lnvo|vesboththeunknownfunctlonx , ; andltsnrstderlvatlvex ,;=ax}a The
dlerentla|equatlon
a

y ay
ax

+
ax
+ y=+
lnvo|ves the unknown functlon y ofthe lndependent varlab|e x and the nrst two
derlvatlvesyandyofy
Thestudyofdlerentla|equatlonshasthreeprlnclpa|goa|s.
1. To dlscover the dlerentla| equatlon that descrlbes a speclned physlca|
sltuatlon.
Z. To nndelther exact|y or approxlmate|ythe approprlate so|utlon of that
equatlon.
J. To lnterprettheso|utlonthatlsfound.
\
Z Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| eZ
In a|gebra, wetyplca||y seek the unknown numbers that satlsfy an equatlon
suchasx

+7x
:
- 11x+4I = 0. Bycontrast,lnso|vlngadlerentla|equatlon,we
arecha||engedtonndtheunknownfnctions , = ,(x) forwhlchanldentltysuch
as,' (x) = 2x,(x)thatl s, thedlerentla|equatlon
d,
= 2x,
dx
ho|ds on some lnterva| ofrea| numbers. Ordlnarl|y, we wl|| want to nnd all
so|utlonsofthedlerentla|equatlon,lfposslb|e.
Ifcl saconstantand
then
( x2 ) x) = Ce ,
= c2xex2 ,= (2x) cex2 ,= 2x,.
( I )
Thus every functlon ,(x) ofthe form l nEq. ( I ) satisfes-and thus l s a so|utlon
ofthedlerentla|equatlon
d,
= 2x,
dx
(2)
fora|| x. In partlcu|ar, Eq. ( I ) dennes an infnite faml|y ofdlerent so|utlons of
thls dlerentla| equatlon, oneforeach cholce ofthe arbltrary constant c By the
methodofseparatlonofvarlab|es(Sectlon I . 4)ltcanbeshownthateveryso|utlon
ofthedlerentla|equatlonln(2)lsoftheformlnEq.( I ).
Diferential Equations and Mathematical Models
Thefo||owlngthreeexamp|esl||ustratetheprocessoftrans|atlngsclentlnc|awsand
prlnclp|es lnto dlerentla| equatlons. In each ofthese examp|es the lndependent
varlab|elstlmet, butwewl||seenumerousexamp|eslnwhlchsomequantltyother
thantlmelsthelndependentvarlab|e.
@.||

Newtn' s|a

ofoohn,

ay stted

lnthls way. The time rte c,chage(th


rateofchangewlth respectto tlme t) ofthe temperature T(t) ofabody ls propor-
tlona|tothedlerencebetweenandthetemperature~ofthesurroundlngmedlum
TcmpcmmA
Tcmpcm1
FIGU 1.1.1. Newton' s law of
cooling, Eq. (3), describes the
cooling of a hot rock in water.
cXump| e4
(Flg. I . I . I ) . Thatl s,
d
= -/, - ~; . (3)
where /l s a posltlve constant. Observe that lf ~ ~, then dI Jdt < 0, sothe
temperature ls adecreaslngfunctlonoft andthe body ls coo|lng. Butlf < ~,
thendI }dt ~ 0, s othatl slncreaslng.
Thusthephyslca||awl strans|atedlntoadlerentla|equatlon. Ifweareglven
theva|uesof/and ~,weshou|dbeab|etonndanexp|lcltformu|aforT(t), and
thenwlth the ald ofthls formu|awecan predlctthe future temperature ofthe
body.
......
Torr|ce|| l ' s|aw lmp|lesthatthetime rte c
)
change ofthe vo|ume \ ofwaterlna
dralnlngtank(Flg. I . I . 2)ls proportlona|tothesquarerootofthedepth , ofwater
lnthetank.
d\
= -/.
a
(4)
I .I Di fferenti al Equati ons and Mathemati cal Model s J
where/lsaconstant. Ifthetanklsacy|lnderwlthvertlca|sldesandcross-sectlona|
area ~,then =~y,soa, a=~

,ay, a ; Inthls caseEq.,:;takesthefom


whereh =/,~ls aconstant.
ay
=-h.,
a
,;

g
-

u|

wl

th

cons

an

th

te

sl

s,
FIGUR 1.1.2. Torricelli' s law
of draining, Eq. ( 4), describes the
draining of a water tank.
lnmanyslmp|ecases,proportlona|totheslzeofthepopu|atlon. Thatls,
ai
- = /i. ,:;
a
where/l stheconstantofproportlona|lty.
LetusdlscussExamp|efurther. Notenrstthateachfunctlonofthefom
i( ) = ce

,:;
ls aso|utlonofthedlerentla|equatlon
ai
- = /i
a
ln,:;. We verlfythlsassertlonasfo||ows.
i

,;= c/e

=/ce

= /i( )
for a||rea| numbers Becausesubstltutlonofeach functlonofthefom glvenln
,:;lntoEq. ,:;producesanldentlty,a||suchfunctlonsareso|utlonsof Eq.,:;.
Thus, even lfthe va|ue ofthe constant/ ls known, the dlerentla| equatlon
ai , a = /ihas/]/e/y-cydlerentso|utlonsoftheformi,;=ce

,onefor
eachcholceofthearbltrary constantc Thls l styplca| ofdlerentla|equatlons.
Itls a|sofortunate, because ltmay a||owus to useaddltlona|lnformatlonto se|ect
fromamonga||theseso|utlonsapartlcu|aronethatntsthesltuatlonunderstudy .
Bl
Sups tha

(t) ce
.

ls theu|atlonofaco|onyofbacterlattl

:that
thepopu|atlonattlme = c(hours, h) was l ccc,andthatthepopu|atlondoub|ed
afterlh. Thl saddltlona|lnfomatlonabout i, ; yle|dsthefo||owlngequatlons.
l ccc= i,c;=ce
,
=c.
zccc=i, i ; =ce

Itfo||owsthatc= l cccandthate

= z,so/=ln z c. : l::.Wlththlsva|ue
of/thedlerentla|equatlonln,:)l s
ai
a
=(|n 2)F ,c. : l::; i.
Substltutlonof/=l nzandc= l ccclnEq. ,:;yle|dsthepartlcu|arso|utlon
i, ; = l ccce

= ( ccc,e
'
:
;

= i ccc
.
z

(becausee
'
:
=z;
that satlsnes the glven condltlons. Wecan use thls partlcu|ar so|utlonto predlct
future popu|atlons ofthe bacterla co|ony. For lnstance, the predlcted number of
bacterlalnthepopu|atlonafterone andaha|fhours (when= i . )ls
i( i . ) = l cccz

:
zszs.
4 Chapter I Fi rst-Order Di fferenti al Equati ons
c-l z c-6 c-1

6
+
z
-z
-+
-6
-
z -l
c--l z
FIGUR 1.1.3. Graphs of
P(t) =

with k = In 2.
ThecondltlonF(0) = i ccclnExamp|e:l sca||edaninitial condition be-
cause we frequent|y wrlte dlerentla| equatlons for whlch = cls the startlng
tlme. Flgure I . I . 3 shows severa| dlerentgraphs oftheformF( ) = ce
,

wlth
k =lnz Thegraphsofa||thelnnnlte|ymanyso|utlonsofai,a =kFlnfactn||
theentlretwo-dlmenslona|p|ane, andnotwolntersect. Moreover,these|ectlonof
any onepolnt F, ontheF-axls amounts to adetermlnatlonofF(0). Becauseex-
act|yoneso|utlonpassesthrougheachsuchpolnt, weseelnthlscasethatanlnltla|
condltlonF(0)=F, determlnesaunlque so|utlonagreelngwlththeglvendata.
Mathematical Models
Ourbrlefdlscusslonofpopu|atlongrowthlnExamp|es5and:l||ustratesthecrucla|
processof-c/e-c/.c/-oae//(Flg. I . l . 4),whlchlnvo|vesthefo||owlng.
1. Theformu|atlon ofarea|-wor|dprob|emlnmathematlca| terms, thatls, the
constructlonofamathematlca|mode| .
Z. Theana|yslsorso|utlonoftheresu|tlngmathematlca|prob|em.
J. The lnterpretatlon ofthe mathematlca| resu|ts lnthecontext oftheorlglna|
rea|-wor|dsltuatlonforexamp|e, answerlngthequestlonorlglna||yposed.
FIGUR 1. 1.4. The process of mathematical modeling.
Inthepopu|atlonexamp|e,therea|-wor|dprob|em ls thatofdetermlnlngthe
popu|atlonatsomefuture tlme. A mathematical model conslstsofa |lstofvarl-
ab|es(Fand;thatdescrlbetheglvensltuatlon,togetherwlthoneormoreequatlons
re|atlngthesevarlab|es,ai,a=kF, F(0) =F,)thatareknownorareassumedto
ho|d. Themathematlca|ana|yslsconslstsofso|vlngtheseequatlons(here,forF as
afunctlonof ; Flna||y, weapp|y thesemathematlca|resu|tstoattempttoanswer
theorlglna|rea|-wor|dquestlon.
As an examp|e ofthls process, thlnk ofnrstformu|atlng themathematlca|
mode| conslstlngoftheequatlonsai, a=kF, F(0) = l ccc,descrlblngthebac-
terlapopu|atlonofExamp|e : Thenourmathematlca| ana|yslsthereconslstedof
so|vlngfortheso|utlonfunctlon i, = l ccce

= l cccz

asourmathemat-
lca| resu|t. For an lnterpretatlon ln terms ofour rea|-wor|d sltuatlonthe actua|
bacterla popu|atlonwe substltuted = I . 5 toobtalnthe predlcted popu|atlonof
F( I . 5) zszsbacterlaafter I . 5 hours. If,forlnstance,thebacterlapopu|atlonls
growlngunderldea|condltlonsofun|lmltedspaceandfoodsupp|y, ourpredlctlon
maybequlte accurate, lnwhlch case weconc|udethatthemathematlca|mode| ls
qulteadequateforstudylngthl spartlcu|arpopu|atlon.
Ontheotherhand,ltmaytumoutthatnoso|utlonofthese|ecteddlerentla|
equatlon accurate|y nts the actua|popu|atlonwe're studylng. Forlnstance,foro
cholceoftheconstantscandkdoestheso|utlonF,;=ce
,
' lnEq. ,:;accurate|y
cXump| e
I .I Di fferenti al Equati ons and Mathemati cal Model s 5
descrlbethe actua| growthofthe human popu|atlonofthewor|doverthe pastfew
centurles. We mustconc|ude that the dlerentla| equatlon ai,a = /ils lnad-
equate formode|lng the wor|d popu|atlon-whlch ln recent decades has |eve|ed
o as compared wlth the steep|y c|lmblng graphs ln the upper ha|f ,i > c;of
Flg. l l Wlth sufnclentlnslght, we mlghtformu|ateanew mathematlca|mode|
lnc|udlngaperhapsmorecomp|lcateddlerentla|equatlon,onethatthattakeslnto
accountsuchfactorsasa|lmltedfoodsupp|yandtheeectoflncreasedpopu|atlon
onblrthanddeathrates. Wlththeformu|atlonofthls newmathematlca|mode|,we
mayattempttotraverseonce agalnthedlagramofFlg. l i :lnacounterc|ockwlse
manner. Ifwecan so|vethe new dlerentla| equatlon, wegetnew so|utlon func-
tlons to compare wlth the rea|-wor|dpopu|atlon. Indeed, a successfu| popu|atlon
ana|yslsmayrequlrerennlngthemathematlca|mode|stl||furtherasltlsrepeated|y
measuredagalnstrea|-wor|dexperlence.
ButlnExamp|e:weslmp|ylgnoredanycomp|lcatlngfactorsthatmlghtaf-
fect our bacterla popu|atlon. Thls made the mathematlca| ana|ysls qulte slmp|e,
perhapsunrea|lstlca||yso. Asatlsfactorymathematlca|mode|lssubecttotwocon-
tradlctoryrequlrements. Itmustbesufnclent|ydetal|edtorepresenttherea|-wor|d
sltuatlonwlthre|atlveaccuracy, yetltmustbesumclent|yslmp|etomakethemath-
ematlca| ana|yslspractlca| . Ifthe mode| ls so detal|ed that ltm||y represents the
physlca|sltuatlon,then themathematlca|ana|yslsmaybetoodlmcu|ttocarryout.
Ifthemode| ls too slmp|e, theresu|tsmay be so lnaccurateastobeuse|ess. Thus
therelsanlnevltab|etradeobetweenwhatlsphyslca||yrea|lstlcandwhatlsmath-
ematlca||yposslb|e. The constructlonofamode| thatadequate|y brldges thls gap
between rea|lsm and feaslbl|lty ls therefore the most crucla| and de|lcate step ln
the process. Ways must be found to slmp|lfy the mode| mathematlca||y wlthout
sacrlnclngessentla|featuresoftherea|-wor|dsltuatlon.
Mathematlca|mode|sare dlscussedthroughoutthl sbook. Theremalnderof
thl slntroductorysectlonls devotedtoslmp|eexamp|esandtostandardtermlno|ogy
usedlndlscusslngdlerentla|equatlonsandthelrso|utlons.
Examples and Terminolog
Ifclsaconstantandy,x; = i,,c- x; , then
lfx = c Thus
ay i
:
= = y
ax ,c- x;
:
l
y,x; =
c- x
dennesaso|utlonofthedlerentla|equatlon
ay
:
- = y
ax
(8)
(9)
onany lnterva|ofrea| numbers not contalnlngthepolntx = c Actua||y, Eq. (8)
dennesaoe pcm-ee)c-//yofso|utlonsofay,ax = y
:
,oneforeachva|ueof
thearbltrary constantorparameterc Wlthc= iwegetthepartlcu|arso|utlon
i
y,x; =

l - x
thatsatlsnesthelnltla|condltlony,c;= l AslndlcatedlnFlg. l i , thlsso|utlon
lscontlnuousonthelnterva| (c, i ; buthasavertlca|asymptoteatx=i

6 Chapter I Fi rst-Order Di fferenti al Equati ons


cXump| ed
:x
:
y+y=c , i c;
fora||x ~ 0.
Sol uti on Flrstwecomputethederlvatlves
ThensubstltutlonlntoEq. , i c;yle|ds
lfxl sposltlve,sothedlerentla|equatlonl ssatlsnedfora||x ~ 0.
Thefactthatwecan wrlte adlerentla| equatlon ls not enoughtoguarantee
thatlthasaso|utlon. Forexamp|e,ltlsc|earthatthedlerentla|equatlon
( i l )
has o (rea|-va|ued)so|utlon, becausethe sum ofnonnegatlvenumberscannotbe
negatlve. Foravarlatlononthlstheme,notethattheequatlon
( l z)
obvlous|yhas on|ythe(rea|-va|ued) so|utlony,x; = c Inourprevlousexamp|es
anydlerentla|equatlonhavlngat|eastoneso|utlonlndeedhadlnnnlte|ymany.
The order ofadlerentla|equatlonls theorderofthehlghestderlvatlvethat
appears ln lt. The dlerentla| equatlon ofExamp|e s ls ofsecondorder, those ln
Examp|eszthrough:arenrst-orderequatlons,and
ls a fourth-order equatlon. The most genera| form of an nth-order dlerentla|
equatlon wlth lndependentvarlab|exandunknownfunctlonordependentvarlab|e
y=y,x; l s
i,

c x. y. y . y . , y = . , ( ;
whereil saspeclncrea|-va|uedfunctlonof+zvarlab|es.
Ouruseofthe wordso//ohasbeenuntl|nowsomewhatlnforma|. To be
preclse,wesaythatthecontlnuousfunctlon=,x;l sasolution ofthedlerentla|
equatlonln, i ;on the interval iprovldedthatthederlvatlves

,, . . . ,

exlst
oniand
i,

c x, , , , . =
for a|| x ln i For the sake ofbrevlty, we may say that = ,x; satisfes the
dlerentla|equatlonln, l ; oni
Hemarkt Reca||frome|ementaryca|cu|usthatadlerentlab|efunctlonon
an open lnterva| ls necessarl|y contlnuous there. Thls ls why on|y a contlnuous
functlon can qua|lfy as a (dlerentlab|e) so|utlon ofa dlerentla| equatlon onan
lnterva| .
cXump| e
COlI|lUCO
cXump| e
:
y= ll(~x)

-:
-: :
x
FIGURE 1. 1.5. The solution of
y' = y
2
defned by
y(x) = lf,l - x) .
:

x
FIGURE 1.1.6. The three
solutions ][(x) = 3 cos 3x,
Y
2
(
X) = 2 sin 3x, and
Y3(X) = -3 cos 3x + 2 sin 3x of
the diferential equation
y" + 9y = 0o
I .I Di fferenti al Equati ons and Mathemati cal Model s 7
______ __ ........ ..... ..... ... . . .... . . . ... . ... ....... . ... ...
Flgure i l showsthetwo 'connectedbranchesofthegraphy= IJ(I - x) The
|eft-handbranchls thegraphofa(contlnuous)so|utlonofthedlerentla| equatlon
y=y
:
thatls dennedonthelnterva| ( -cg l ; Therlght-handbranchlsthegraph
ofaa_ eeso|utlonofthe dlerentla| equatlonthat ls denned(and contlnuous)
onthedlerentlnterva| , l . ) . So the slng|eformu|ay(x) = IJ(I - x) actua||y
dennes two dlfferent so|utlons (wlth dlerent domalns ofdennltlon) ofthe same
dlerentla|equatlony=y
:


IfA andB areconstantsand
y(x) =Acos3x +B sln3x,
then twosuccesslvedlfferentlatlonsyle|d
y (x) =3A sln 3x + 3B cos 3x,
y
(x) =9A cos 3x 9B sln 3x =-y(x)
(I4)
for a||x Consequent|y, Eq , l :;dennes what ltls natura| to ca|| amo pcc-ee
)c-//y ofso|utlonsofthesecond-orderdlerentla|equatlon
y

+y=c (I5)
on the who|e rea| number |lne Flgure I I 6 shows the graphs ofsevera| such
so|utlons
A|thoughthedlerentla|equatlonsln(I I ) and(I 2)areexceptlonstothegen-
era| ru|e, we wl|| see that an nth-order dlfferentla| equatlon ordlnarl|y has an n-
parameterfaml|yofso|utlonsonelnvo|vlngdlerentarbltraryconstantsorpa-
rameters
InbothEqs(I I ) and, l z;, theappearanceofyasanlmp|lclt|ydennedfunc-
tloncausescomp|lcatlons Forthlsreason, wewl|| ordlnarl|y assumethatanydlf-
ferentla|equatlonunderstudycanbeso|vedexp|lclt|yforthehlghestderlvatlvethat
appears, thatl s, thattheequatloncanbewrlttenlntheso-ca||edoc/)o-
(
,
)

G
(
,
~[
)
(I6) y

x. y. y . y . . y ,
where G ls area|-va|uedfunctlonof+ I varlab|es Inaddltlon, wewl||a|ways
seekon|yrea|-va|uedso|utlonsun|esswewmthereaderotherwlse
A|| thedlerentla|equatlonswehavementlonedsofarare ordinary dler-
entla|equations,meanlngthatthe unknownfunctlon(dependentvarlab|e)depends
onon|y a s//e lndependentvarlab|e Ifthe dependent varlab|e ls afunctlon of
twoormorelndependentvarlab|es,thenpartla|derlvatlvesare|lke|ytobelnvo|ved,
lftheyare, theequatlon ls ca||eda partial dlerentla|equatlon Forexamp|e,the
temperature=,x. ; ofa|ongthlnunlformrod atthe polntx attlme satlsnes
(underapproprlateslmp|econdltlons)thepartla|dlerentla|equatlon
-
=/
-
:

- -x
:

where / ls a constant (ca||ed the /ec/a_ s/./- ofthe rod) In Chapters I


through :wewl||beconcemedon|ywlthoa/c dlerentla|equatlonsandwl||
refertothemslmp|yasdlerentla|equatlons
Inthl schapterweconcentrateon]s oae dlerentla|equatlonsoftheform
ay
ax
=),x. y; (I :)
Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| e 1
We a|sowl|| samp|e thewlderange ofapp|lcatlons ofsuchequatlons. A typlca|
mathematlca|mode|ofanapp|ledsltuatlonwl||bean initial value problem, con-
slstlngofadlerentla|equatlonoftheformln(I7)togetherwlthan initial condi
tion y,x,; = y, Notethatweca||y,x,; = y,an lnltla|condltlon whether ornot
x,= cTo solve thelnltla|va|ueprob|em
ay
ax
=),x,y; . y,x,;=y, , i s;
means to nnd a dlerentlab|e functlon y = y,x; that satlsnes both condltlons ln
Eq., i s;onsomelnterva|contalnlngx,
Clven the so|utlon y,x; = IJ(C - x; ofthe dlerentla| equatlon ay}ax = y

dlscussedlnExamp|e7, so|vethelnltla|va|ueprob|em
ay
:
- =y , y, l ; =z
ax
Sol uti on We needon|ynndava|ueofC sothattheso|utlony,x; = IJ( C - x; satlsnesthe
lnltla|condltlony, l ; = z Substltutlonoftheva|uesx = I andy=zlntheglven
so|utlonyle|ds
:
|
|
y=2/|3 2)
|
|
\
( l ,

.
.-3/2
.
~:
~:
,
X

|
|
|
:
FIGUR 1.1.7. The solutions of
y' = y
-
defned by
y(x) = 2/ (3 - 2x).
I
z=y, i ; =
C - I
`
so 2C - z = i , and hence C = . Wlth thls va|ue ofC weobtaln the deslred
so|utlon
I z
y,x;

=
z
_ - x - x
Flgure I . I . 7 showsthetwobranchesofthe graph y = z},- zx; The|eft-hand
branch ls the graphon (-c, ,ofthe so|utlonofthe glvenlnltla| va|ueprob|em
y= y
:
, y,;= z Therlght-handbranchpassesthroughthepolnt ,z,-z;andl s
thereforethegraphon ,,c)ofthe so|utlonofthedlerentlnltla| va|ueprob|em
y= y
:
,y,z;= -z
Thecentra|questlonofgreatestlmmedlatelnteresttouslsthls. Ifweareglven
adlerentla|equatlonkuowntohaveaso|utlon satlsfylngaglvenlnltla|condltlon,
how do we actua||y]a or .o-pe that so|utlon! And, once found, what can
we do wlth lt! We wl|| see that a re|atlve|yfew slmp|e technlquesseparatlon
of varlab|es (Sectlon i .:;, so|utlon of|lnear equatlons (Sectlon I . 5), e|ementary
substltutlon methods (Sectlon I . 6)are enough to enab|e usto so|ve a varlety of
nrst-orderequatlonshavlnglmpresslveapp|lcatlons.

h|
^

-.
In Prblems 1 thrugh I2, veri by substitution that each
given fnction is a solution of the given diferential equation.
Thrughout these prblems, primes denote derivatives with re
spect to x.
1. y' = x
-
; y = x
3
+ 7
2. y' + 2y = 0; ] = 8
-
`
3. y" + 4y = 0; ]
) = cos 2x, ]
-
= sin 2x
4. y
"
=
9
y; ])
= e
3
` .]
-
= e
3
`
5. y' = y + 2e` ; y = eX - e-X
6. y
" + 4y' + 4y = 0; ]
J
= e
-
` , ]
-
= xe
-
-
`
7. ] - 2y' + 2y = 0; ]
J
= eX cosx, ]
-
= eX sinx
8. y
"+y = cos 2x, ]| = cosx-cos 2x, ]
-
= sinx-cos 2x
l
9. y' + 2xy
2
= 0; ] =
I +x
-
l
10. x2y" + xy' - ] = lnx; ] = x - lnx, ]
-
= - - lnx
x
I .I Di fferenti al Equati ons and Mathemati cal Model s 9
1 lnx
11. x2y" +5xy' +4y = 0; ] = . ] = -
2 X X
12. x2y" - xy' + 2y = 0; ] =x cos(lnx) , ] =x sin (In x)
In Prblems IJthrugh 16, substitute y = erx into the given
diferential equation to determine all values of the constant r
for which y = erx is a solution of the equation.
13. 3y' = 2y 14. 4y" = y
1
5. y" + y' - 2y = 0 16. 3y" + 3y' - 4y = 0
In Prblems I7thrugh 2, frst veri that y(x) satisfes the
given diferential equation. Then deterine a value of the con
stant C so that y(x) satisfes the given initial condition. Use a
computer or graphing calculator (desired) to sketch several
tpical solutions of the given dif erential equation, and high
light the one that satisfes the given initial condition.
17. y' + y = 0; y(x) = Ce-X , y(O) = 2
18. y' = 2y; y(x) = Ce2x, y(O) = 3
19. y' = y + 1 ; y(x) = Cex - I, y(O) = 5
20. y' =x - y; y(x) = Ce-X + x - I, y(O) = 1 0
21. y' + 3x2y = 0; y(x) = Ce-x
3
, y(O) = 7
22. eYy' = 1 ; y(x) = In(x + C), y(O) = 0
dy
23. x
dx
+ 3y = 2x5; y(x) = x5 + Cx-
3
, y(2) = 1
24. xy' - 3y =x
3
; y(x) =x
3
(C + lnx), y(l) = 1 7
25. y' = 3x2 (y2
+ 1 ) ; y(x) = tan(x
3
+ C), y(O) = 1
26. y' + y tanx = cosx; y(x) = (x + C)cosx, y(r) = 0
In Prblems 27thrugh JI, a fnction y = g (x) is described
by some geometric prpert of its graph. Write a dif erential
equation of the for dyjdx = f(x, y) having the fnction g
as its solution (or as one of its solutions).
27. The slope of the graph of g at the point (x, y) is the sum
of x and y.
28. The line tangent to the graph of g at the point (x, y) inter
sects the x-axis at the point (xj, 0).
29. Every straight line normal to the graph of g passes through
the point (0, 1 ) . Can you guess what the graph of such a
function g might look like?
30. The graph of g is normal to every curve of the form
y =x2
+ k (k is a constant) where they meet.
31. The line tangent to the graph of g at (x, y) passes through
the point ( -y, x).
In Prblems 32 thrugh J, write-in the manner of Eqs. (J)
thrugh () of this section-a dif erential equation that is a
mathematical model of the situation described.
32. The time rate of change of a population P is proportional
to the square root of P.
33. The time rate of change of the velocity v of a coasting
motorboat is proportional'to the square of v.
34. The acceleration dvjdt of a Lamborghini is proportional
to the diference between 250 K and the velocity of the
car.
35. In a city having a fxed population of P persons, the time
rate of change of the number of those persons who have
heard a certain rumor is proportional to the number of
those who have not yet heard the rumor.
36. In a city with a fxed population of P persons, the time rate
of change of the number of those persons infected with
a certain contagious disease is proportional to the product
of the number who have the disease and the number who
do not.
In Prblems 37 thrugh 42, deterine by inspection at least
one solution of the given dif erential equation. That is, use
your knowledge of derivatives to make an intelligent guess.
Then test your hypothesis.
37. y" = 0 38. y' = y
39. xy' + y = 3x2 40. (y')2
+ y2 = 1
41. y' + y = eX 42. y" + y = 0
43. (a) If k is a constant, show that a general (one-parameter)
solution of the diferential equation
dx
= kx
2
dt
is given by x(t ) =lj (C -kt ) , where C is an arbitrary
constant.
(b) Determine by inspection a solution of the initial value
problem x' = kx2, x (0) =O.
44. (a) Continuing Problem 43, assume that k is positive, and
then sketch graphs of solutions of x' = kx2 with sev
eral typical positive values of x(O).
(b) How would these solutions difer if the constant k
were negative?
45. Suppose a population P of rodents satisfes the diferen
tial equation dPjdt = kp2. Initially, there P (O) = 2
rodents, and their number is increasing at the rate of
dPjdt = 1 rodent per month when there are P = 1 0 ro
dents. How long will it take for this population to grow
to a hundred rodents? To a thousand? What's happening
here?
46. Suppose the velocity v of a motorboat coasting in water
satisfes the diferential equation dvjdt = kv2. The initial
speed of the motorboat is v(O) = 10 meters per second
(ms), and v is decreasing at the rate of 1 ms
2
when v =5
ms. How long does it take for the velocity of the boat u
decrease to 1 ms? To ms? When does the boat come
to a stop?
47. In Example 7 we saw that y(x) = lj (C - x) defnes a
one-parameter family of solutions of the diferential equa
tion dyjdx = y2. (a) Determine a value of C so that
y(lO) = 1 0. ..Is there a value of C such that y(O) =O?
Can you nevertheless fnd by inspection a solution of
dyjdx = y2 such that y(O) = O? (c) Figure 1 . 1 . 8 shows
typical graphs of solutions of the form y(x) =lj (C -x).
Does it appear that these solution curves fll the entire xy
plane? Can you conclude that, given any point (a, b) in
the plane, the diferential equation dyjdx = y2 has ex
actly one solution y(x) satisfying the condition y(a) = b?
\ 0 Chapter I Fi rst-Order Di fferenti al Equati ons
c--l c--I c-0 c-I c-l c-1
1
l
c-+
.aa
sa
-a
a
la
~ 0*

- l0
- +0
-a
sa
X
FIGURE 1. 1.S. Graphs of solutions of the
equation dyjdx = y2.
FIGURE 1. 1.9. The graph y = CX
4
for
various values of C.
48. (a) Show that y(x) = CX
4
defnes a one-parameter fam
ily of differentiable solutions of the diferential equation
xy' = 4y (Fig. ..Show that
y(x) =
_x
4
if x 0,
X
4
if x 0
defnes a diferentiable solution of xy' = 4y for all x, but
is not of the form y(x) = CX
4
. (c) Given any two real
numbers u and b, explain why-in contrast to the situa
tion in part (c) of Problem 47-there exist infnitely many
diferentiable solutions of xy' = 4y that all satisfy the
condition y(u) = b.
The nrst-orderequatlon d,Jdx = ](x , ,) takes an especla||y slmp|eform lfthe
rlght-hand-sldefunctlon]doesnotactua||yl nvo|vethedependentvarlab|e,, so
d,
dx
=](x) .
I nthl sspecla|caseweneedon|ylntegratebothsldesofEq. (I )toobtaln
,(x) =](x) dx+C.
(I )
(2)
Thlsls ageneral solution ofEq.(I ), meanlngthatltlnvo|vesanarbltraryconstant
C, andforeverycholceofC ltls a so|utlonofthedlerentla| equatlon ln(I ) . If
G(x) ls apartlcu|arantlderlvatlveof]thatls, lfG' (x) = ](x)then
,(x) =G(x) +C. ,;
The graphs of any two such so|utlons ), (x) = G(x) +C and )
:
(x) =
G(x)+ C
:
onthesamelnterva|iepara||e|lnthesensel||ustratedbyFlgs. I . 2. I
and I . 2. 2. There we see thatthe constant C ls geometrlca||ythevertlca|dlstance
betweenthetwocurves, (x) =G(x) and,(x) =G(x) +C.
To satlsfyanlnltla|condltlon,(x,) =),, weneedon|ysubstltutex =x, and
, =), lnto Eq. ,;toobtaln), = G(x,) +C, sothatC =), - G(x,) . Wlththls
cholceofC,weobtalntheparticular solution of Eq.(I )satlsfylngthelnltla|va|ue
prob|em
d
d)
x
=](x) , ( ) , x, =),

cXump| e 1
I . Z I ntegral s as General and Parti cul ar Sol uti ons \ \
:
z

.
z
:
-
+ : z . . z : +
X
c~.
c-z
+
z
z
-+
0 z -
X
FIGURE 1.2.1. Graphs of FIGURE 1.2.2. Graphs of
y = x
2
+ C for various values of C.
y = sin x + C for various values of
Wewl||seethatthl sl sthetyplca|pattemforso|utlonsofnrst-orderdlerentla|
equatlons. Ordlnarl|y, we wl|| nrst nnd a eem/so//olnvo|vlng an arbltrary
constantC. Wecanthenattempttoobtaln, byapproprlatecholceofC, apc/./c
so//osatlsfylngaglvenlnltla|condltlony,x,;=y,
Hemarkt As thetermlsusedl ntheprevlousparagraph,aee/so//o
ofanrst-orderdlerentla|equatlon ls slmp|yaone-parameterfaml|yofso|utlons.
A natura| questlon ls whether a glven genera| so|utlon contalns e.epartlcu|m
so|utlon of the dlerentla| equatlon. When thls ls known to be true, we ca|| lt
the genera| so|utlon of the dlerentla| equatlon. For examp|e, because any two
antlderlvatlves ofthe samefunctlon ](x) can dler on|yby aconstant, ltfo||ows
thateveryso|utlonofEq.(I ) lsoftheformln(2) Thus Eq. (2)servestodennethe
genera|so|utlonof(I )

So|vethelnltla| va|ueprob|em
d,
- =2x +3, ,(I ) =2
dx
Sol ution
Integratlonofbothsldesofthedlerentla|equatlonaslnEq. (2)lmmedlate|yyle|ds
+
z

-+
-
s

- -+ z z +
X
FIGURE 1.2.3. Solution curves
for the diferential equation in
Example
thegenera|so|utlon
y(x) =(2x+ 3)dx
-
x
:
+3x+C.
Flgure I 2 3 shows the graph y = x
:
+ 3x + C forvarlousva|ues of C. The
partlcu|arso|utlon we seekcorrespondsto the curve that passes through thepolnt
(I , 2), therebysatlsfylngthelnltla|condltlon
y, i ; =(I )
:
+3

(I ) +C =2
Itfo||owsthatC =2, sothedeslredpartlcu|arso|utlonl s
y(x) =x
:
+3x 2

\ 2 Chapter I Fi rst-Order Di fferenti al Equati ons


Second-order equatons. The observatlon that the specla| nrst-order equatlon
ay, ax=),x; lsreadl|yso|vab|e(provldedthatanantlderlvatlveof)canbefound)
extendstosecond-orderdlerentla|equatlonsofthespecla|form
,:;
ln whlchthe functlonon the rlght-handsldelnvo|vesneltherthedependentvarl-
ab|eynorltsderlvatlveay, ax. Weslmp|ylntegrateoncetoobtaln
=y
,x; ax=,x; ax=o,x; +c.
where o ls an antlderlvatlve of and c, ls an arbltrary constant. Then another
lntegratlonyle|ds
y,x; =y

,x; ax= o,x; +cax=o,x; ax+c, x+c


:
g
where C
:
ls a second arbltrary constant. In eect, the second-order dlerentla|
equatlon ln ,:; ls one that can be so|ved by so|vlng successlve|y the]s oae
equatlons
ac
ax
=,x; and
Velocity and Acceleration
ay
- =c ,x;
ax
Dlrectlntegratlonls sumclenttoa||owus toso|veanumberoflmportantprob|ems
concemlng the motlon ofa partlc|e (or-.sspo/;lnterms oftheforces actlng
onlt. The motlonofapartlc|ea|ongastralght|lne (thex-axls)lsdescrlbedbylts
position function
x=), ;
glvlngltsx-coordlnateattlme Thevelocity ofthepartlc|el sdennedt obe
c, ; =)

,; , thatls,
ax
c = .
a
Itsacceleration c ,;l sc,;=c

,;=x ,; ,lnLelbnlznotatlon,
ac a
:
x
. = =
a a
:
(5)
(6)
(7)
Equatlon(6)l ssometlmesapp|ledeltherlnthelndennltelntegra|formx,;=
,c ,; aorlnthedennltelntegra|form
x ,; =x,,;+,

c ,s; .

whlchyou shou|drecognlzeasastatementofthefundamenta|theoremofca|cu|us
(preclse|ybecauseax, a=c;
cXump| eZ
I .2 I ntegral s as General and Parti cul ar Sol uti ons 1 J
Newton' sse.oa/.o)-o/osaysthatlfaforce i,; actsonthepartlc|e
andlsdlrecteda|ong lts|lneofmotlon,then
-. ,;=i,; , thatls, i=-., ,s;
where - ls the mass ofthepartlc|e. Iftheforce i l sknown, thentheequatlon
x,) = i,;}-canbelntegrated twlceto nnd theposltlonfunctlonx(r) lnterms
oftwo constantsoflntegratlon. Thesetwo arbltrary constantsarefrequent|ydeter-
mlnedby the initial position x, = x(0) and the initial velocity :, = c,c;ofthe
partlc|e.
Constant acceleratin. Forlnstance, suppose thattheforce i, andtherefore the
acce|eratlon.=i }-,are.os.Thenwebeglnwlththeequatlon
ac
- =. ,.ls aconstant)
a
andlntegratebothsldestoobtaln
c ,; =. a = . + c,
,;
Weknow that c = :, when = 0, and substltutlon ofthls lnformatlon lntothe
precedlngequatlonyle|dsthefactthatc, =:,So
Asecondlntegratlonglves
ax
c ,; = - =.+:,
a
x ,; =c , ; a=,.+c,;a= ,.
:
+c,+c
:
,
andthesubstltutlon=0,x =x,glvesc
:
=x,Therefore,
x ,; = ,.
:
+c,+x,
(I0)
(I I )
Thus, wlthEq. (I 0)wecannndtheve|oclty,andwlthEq.(I I ) theposltlon,of
thepartlc|eat any tlmeln termsoflts.os.acce|eratlon., ltslnltla|ve|oclty
c,,andltslnltla|posltlonx,
A |unar |ander ls fa||lng free|ytoward the surface ofthe moon at a speed of450
meters per second (mJs). Its retrorockets, when nred, provlde a constant dece|-
eratlon of2. 5meterspersecondpersecond (mJs
:
) (thegravltatlona|acce|eratlon
producedbythemoonlsassumedtobelnc|udedlntheglvendece|eratlon). Atwhat
helghtabovethe|unarsurfaceshou|dtheretrorocketsbeactlvatedtoensureasoft
touchdown ,c=0atlmpact) !
1 4 Chapter I Fi rst-Order Di fferenti al Equati ons
Sol ution We denote byx ,; thehelght ofthe|unar |ander abovethesurface, as lndlcated
Lunmsuace
FIGUR 1.2.4. The lunar lander
of Example 2.
ln Flg. I . 2.4. We |et = 0 denote the tlme at whlch the retrorockets shou|dbe
nred. Then :, = -450(mJs, negatlve becausethehelghtx,; ls decreaslng), and
=+2. 5, becauseanupwardthrustlncreasestheve|ocltyc(a|thoughltdecreases
thespeea c ; ThenEqs.( I 0)and( I I ) become
c ,; =2. 5- 450 ( I 2)
and
x,; = I . 25r
:
- 450t+x, . ( I 3)
wherex,ls thehelghtofthe|anderabovethe|unarsurfaceatthetlme =0when
theretrorocketsshou|dbeactlvated.
FromEq.( I 2)weseethatc=0(softtouchdown)occurswhen=450J2.5=
I S0s (thatls, 3 mlnutes) , thensubstltutlonof= I S0,x=0lntoEq.( I 3)yle|ds
x,=0- ( I . 25) ( I S0)
:
+450( I S0) =40, 500
metersthatls, x, =40. 5k 25 ml|es. Thustheretrorocketsshou|dbeactl-
vatedwhenthe|unar|anderls40. 5kl|ometersabovethesurfaceofthemoon,andlt
wl||touchdownsoft|yonthe|unarsurfaceafter3mlnutesofdece|eratlngdescent.

Physical Units
Numerlca| work requlresunltsforthe measurementofphyslca| quantltles such as
dlstance and tlme. We sometlmesusead hoc unltssuch asdlstance ln ml|esor
kl|ometersandtlmelnhourslnspecla|sltuatlons(suchaslnaprob|emlnvo|vlng
an auto trlp). However, the foot-pound-second (fps) and meter-kl|ogram-second
(mks) unltsystemsareusedmoregenera||ylnsclentlncandenglneerlngprob|ems.
In fact, fps unlts are common|y used on|y lnthe Unlted States (and afewother
countrles), whl|emksunltsconstltutethestandardlntematlona|systemofsclentlnc
unlts.
Force pound (lb) newton (N)
Mass slug kilogram (kg)
Di stance foot (h) meter (m)
Time second (s) second (s)
g 32 ftls
2
rs
2
The |ast|lneofthl stab|e glves va|uesforthe gravltatlona| acce|eratlon _ at
the surface ofthe earth. A|though these approxlmate va|ues wl|| sumcefor most
examp|esandprob|ems,moreprecl seva|uesare9. 7S05 mJs
:
and32. 0SSftJs
:
(at
sea|eve|attheequator).
Bothsystemsarecompatlb|ewlthNewton' ssecond|awi=. ThusI Nls
(bydennltlon)theforcerequlredtol mpartanacce|eratlonofI mJs
:
toamassofI
kg. Slml|ar|y, I s|ug ls (by dennltlon)themassthatexperlencesan acce|eratlonof
I ftJs
:
underaforceofI |b. (Wewl||usemks unltsln a||prob|emsrequlrlngmass
unltsandthuswl||re|yneeds|ugstomeasuremass.)
I .2 I ntegral s as General and Parti cul ar Sol uti ons 1 5
Inchesandcentlmeters (aswe||asml|esandkl|ometers)a|so8 common|y
used lndescrlblngdlstances. Forconverslonsbetweenfpsandmksunltslthe|psto
rememberthat
Forlnstance,
I ln.=2.54cm(exact|y) and I lb 4. 44SN.
I ft = I2ln. Z 2. 54
m
=30. 4Scm,
rn.
andltfo||owsthat
I ml=52S0ftZ 30.4S
cm
= I 60934. 4cm I . 609k.
ft
ThusapostedU. S. speed|lmltof50mlJhmeansthatlnlntematlona|termsthe
|ega|speed|lmltls about 50 Z I . 609 S0. 45kmJh.
Vertical Motion with Gravitational Acceleration
The weight uofabodyl stheforceexertedonthebodybygravlty. Substltutlon
of.=,andi= ulnNewton' ssecond|awi=-.glves
u = -, (I4)
forthewelghtuofthemass-atthesurfaceoftheearth(where, 32ftJs
:
9. S
mJs
:
).Forlnstance, amassof-=20kghasawelghtofu
-
(20kg)(9. SmJs
:
) -
I 96N.Slml|ar|y,amass-welghlng I 00poundshas mkswelght
soltsmassl s
u=( I 00|b)(4.44SNJ|b)=444. SN,
u 444. S N
-= = 45.4kg.
, 9. SmJs
:
Todlscussvertlca|motlonlt ls natura| tochoosethe,-axlsas thecoordlnate
systemforposltlon,frequent|y wlth , =0correspondlngto 'ground|eve|. Ifwe
choosethep.adlrectlonastheposltlvedlrectlon,thentheeectofgravltyona
vertlca||ymovlngbody lstodecreaseltshelghtanda|sotodecreaseltsve|ocltyu -
ay}a Consequent|y,lfwelgnorealrreslstance, thentheacce|eratlon.=ac}aof
thebody lsglvenby
ac
- = -,
a
( I 5)
Thls acce|eratlon equatlon provldes a startlng polnt ln many prob|ems lnvo|vlng
vertlca|motlon. Successlvelntegratlons(aslnEqs. ( I 0)and( I I ) yle|dtheve|oclty
andhelghtformu|as
c ,; = -,+:, ( I 6)
and
( I 7)
Here, y,denotesthelnltla|,=0)helghtofthebodyand:,ltslnltla|ve|oclty.
1 6 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| e
)-axis

-axis

FIGUR 1.2.5. A swimmer's


problem (Example 4).
cXump| e4
(a) Supposethataba|| ls thrown stralght upward from the ground ,
y, = 0)wlth
lnltla|ve|oclty:,= 96 (ftJs, soweuse,= 32ftJs
:
ln qs unlts). Then ltreaches
ltsmaxlmumhelghtwhenltsve|oclty(Eq. ( I 6) lszero,
u (r) =-32t +96=0,
andthuswhenr =3 s. Hencethemaxlmumhelghtthattheba||attalnsls
,(3) = -|
.
32
.
3
:
+96
.
3 +0= I 44(ft)
(wlththe aldofEq. ( I 7) .
(b) Ifanarrowlsshotstralghtupwardfromthegroundwlthlnltla|ve|oclty:,=49
(mJs, soweuse,=9. SmJs
:
lnmksunlts),thenltretumstothegroundwhen
, (r)=-|
.
(9. S) r
:
+49r =(4. 9) r (-r +I 0) =0,
andthusafter I 0sl nthealr.
A Swmmer
'
s Problem
Flgure I . 2. 5showsanorthward-owlngrlverofwldthd =2a. The|lnesx =Ja
representthebanksoftherlverandthe,-axlsltscenter. Supposethattheve|oclty
u,atwhlchthewaterows lncreasesasoneapproachesthecenteroftherlver,and
lndeedlsglvenlntermsofdlstancex fromthecenterby
( I S)
You can use Eq. ( I S) to verlfy that the water does ow the fastest at the center,
where :,=:,,andthat:,=0ateachrlverbank.
Supposethataswlmmerstartsatthepolnt(-a, 0)onthewestbankandswlms
dueeast(re|atlvetothewater)wlthconstantspeedu . AslndlcatedlnFlg. I . 2. 5, hls
ve|ocltyvector(re|atlvetotherlverbed) hashorlzonta|component :
s
andvertlca|
component:,Hencetheswlmmer' sdlrectlonang|eo ls glvenby
:,
tan o = -
:
Becausetano =d,Jdx, substltutlonuslng( I S)glvesthedlerentla|equatlon
fortheswlmmer' straectory, =,(x) ashecrossestherlver.
( I 9)
.- ..~~. .. ......~... .. . . ....~.. ... .~- .
Supposethattherlverls I ml|ewldeandthatltsmldstreamve|ocltyls :,=9mlJh.
If theswlmmer' sve|ocltyl s: =3mlJh,thenEq.( I 9)takestheform
Integratlonyle|ds
d,
:
- =3 ( I - 4x ) .
dx
,(x) =(3- I 2x
:
) dx =3x- 4x

+C
I . Z I ntegral s as General and Parti cul ar Sol uti ons 1 7
fortheswlmmer' straectory. Thelnltla|condltlon,-,
-
0yle|dsc=I , so
,(x) =3x- 4x

+I .
Then
, |) =3 |) 4 |)

+I =2,
sotheswlmmerdrlfts2 ml|esdownstreamwhl|eheswlms I ml|eacrosstherlver

Problems
..I ..,.I0,........ ...,
.,.,.., ..,.......,....
...
..
1. . . .
..
2.
..
. .
..
..
3. ,.
..
..
4.
..

.
.
..
5.

.
..
..
..

6.
..
...
..
7. .
.. .
..
8.
..
cos ..
..
9. .
..
..
10. . .
..
..II..,.I,...,.. .....
.,,....,..... ..,.
... .....
11. . .
12. . .
13. . . .
14. . .
15. . . .

16. .

, .
.

17. . , .

18. . .. .
..I9..,.22, .,......,....
...,..............
,.,.......,.I. 2. ..,.I. 2. 9. ...,.,.
....,,......,,
19.
I O
&
6
>
(:. :

4
2
O
O 2 4 6
FIGURE 1.2.6. Graph of the
velocity function of Problem
20.
IO*
&
6
6 & | O
FIGURE 1.2.7. Graph of the
velocity function of Problem
21. I O
&
6
'
:. :)
FIGURE 1.2.8. Graph of the
velocity function of Problem
1 8 Chapter I Fi rst-Order Di fferenti al Equati ons
22.
|O
&
6
|
3

5)
2 4
|7

6 & I O
FIGUR 1.2.9. Graph of the
velocity function v(t ) of Problem 22.
23. What is the maximum height attained by the arrow of part
(b) of Example 3?
24. A ball is dropped from the top of a building 400 ft high.
How long does it take to reach the ground? With what
speed does the ball strike the ground?
25. The brakes of a car are applied when it is moving at 1 00
km/h and provide a constant deceleration of 1 0 meters per
second per second (m/s
2
) . How far does the car travel be
fore coming to a stop?
26. A projectile is fred straight upward with an initial veloc
ity of 1 00 m/s from the top of a building 20 m high and
falls to the ground at the base of the building. Find (a) its
maximum height above the ground; (b) when it passes the
top of the building; (c) its total time in the air.
27. A ball is thrown straight downward from the top of a tall
building. The initial speed of the ball is 10 m/s. It strikes
the ground with a speed of 60 m/s. How tall is the build
ing?
28. A baseball is thrown straight downward with an initial
speed of 40 f/s from the top of the Washington Monu
ment (555 f high). How long does it take to reach the
ground, and with what speed does the baseball strike the
ground?
29. A diesel car gradually speeds up so that for the frst 1 0 s
its acceleration is given by
.
.
= (0. 1 2) t
2
(0. 6)t (ft/s
2
) .
I f the car starts from rest (Xq = 0, 0q = 0) , fnd the dis
tance it has traveled at the end of the frst 10 s and its
velocity at that time.
30. A car traveling at 60 mi/h (f/s) skids 1 76 f after its
brakes are suddenly applied. Under the assumption that
the braking system provides constant deceleration, what
is that deceleration? For how long does the skid continue?
31. The skid marks made by an automobile indicated that its
brakes were fully applied for a distance of 75 m before it
came to a stop. The car in question is known to have a con
stant deceleration of 20 m/s
2
under these conditions. How
fast-in km/h-was the car traveling when the brakes
were frst applied?
32. Suppose that a car skids 15 m if it is moving at 50 km/h
when the brakes are applied. Assuming that the car has
the same constant deceleration, how far will it skid if it is
moving at 1 00 km/h when the brakes are applied?
33. On the planet Gzyx, a ball dropped from a height of 20 f
hits the ground in 2 s. If a ball is dropped from the top of
a 200-ft-tall building on Gzyx, how long will it take to hit
the ground? With what speed will it hit?
34. A person can throw a ball straight upward from the sur
face of the earth to a maximum height of l ++ ft. How
high could thi s person throw the ball on the planet Gzyx
of Problem 29?
35. A stone is dropped from rest at an initial height h above
the surface of the earth. Show that the speed with which it
strikes the ground is U = J2gh.
36. Suppose a woman has enough "spring" in her legs to jump
(on earth) from the ground to a height of 2. 25 feet. If
she jumps straight upward with the same initial velocity
on the moon-where the surface gravitational acceleration
is (approximately) 5. 3 fts
2
-how high above the surface
will she rise?
37. At noon a car starts from rest at point .and proceeds at
constant acceleration along a straight road toward point
B. If the car reaches B at 1 2: 50 . V. with a velocity of
60 mi/h, what is the distance from .to B?
38. At noon a car starts from rest at point .and proceeds with
constant acceleration along a straight road toward point C,
35 miles away. If the constantly accelerated car arrives at
C with a velocity of 60 mi/h, at what time does it arrive
at C?
39. If u = 0. 5 mi and 0q = 9 mi/h as in Example +, what
must the swimmer' s speed 0 be in order that he drifs
only l mile downstream as he crosses the river?
40. Suppose that u = 0. 5 mi , 0q = 9 mi/h, and 0 = 3 mi/h
as in Example +,but that the velocity of the river is given
by the fourth-degree function
rather than the quadratic function in Eq. ( l ). Now fnd
how far downstream the swimmer drifts as he crosses the
river.
41. A bomb is dropped from a helicopter hovering at an alti
tude of 800 feet above the ground. From the ground di
rectly beneath the helicopter, a projectile i s fred straight
upward toward the bomb, exactly 2 seconds after the bomb
is released. With what initial velocity should the projectile
be fred, in order to hit the bomb at an altitude of exactly
400 feet?
42. A spacecraft is in free fall toward the surface of the moon
at a speed of 1 000 mph (mi/h). Its retrorockets, when
fred, provide a constant deceleration of 20,000 mi/h
2
. At
what height above the lunar surface should the astronauts
fre the retrorockets to insure a soft touchdown? (As in
Example 2, ignore the moon
'
s gravitational feld. )
.3 Sl ope Fi el ds and Sol uti on Cures 1 9
43. Arthur Clarke' s ..WndJm .Sun ( 1 963) describes
Diana, a spacecraf propelled by the solar wind. Its alu
minized sail provides it with a constant acceleration of
,= 0. 0098 m/s
2

Suppose this spacecraft starts
from rest at time = 0 and simultaneously fres a pro
jectile (straight ahead in the same direction) that travels at
one-tenth of the speed c = 3 ? 1 0
8
m/s of light. How long
will it take the spacecraft to catch up with the projectile,
and how far will it have traveled by then?
44. A driver involved in an accident claims he was going only
25 mph. When police tested his car, they found that when
its brakes were applied at 25 mph, the car skdded only
45 feet before coming to a stop. But the driver's skd
marks at the accident scene measured 21 0 feet. Assum
ing the same (constant) deceleration, determine the speed
he was actually traveling just prior to the accident.
Slpe Fields and Solution Cures
Y
- x
FIGUR 1.3. 1. A solution curve
for the diferential equation
y
'
= X - y together with tangent
lines having
slope ml = Xl - Yl at the
point (Xl . Yl ) ;
slope m
2
= X
2
- Y
2
at the
point (X
2
, Y
2
) ; and
slope m
3
= X
3
- Y
3
at the
point (X3 , Y3) .
cXump| e 1
Conslderadlerentla|equatlonoftheform
ay
ax
=),x, ,) ( I )
wheretherlght-handfunctlon),x, ,)lnvo|vesboththelndependentvarlab|exand
thedependentvarlab|e,. Wemlghtthlnkoflntegratlngboth sldesln( I ) wlthre-
specttox, and hencewrltey,x; =,),x, y,xax+ C. However, thlsapproach
doesnot|eadtoaso|utlonofthedlerentla|equatlon,becausethelndlcatedlntegra|
lnvo|vesthe/ofunctlony,x; ltse|f,andthereforecannotbeeva|uatedexp|lc-
lt|y. Actua||y,thereexlstsostralghtforwardprocedurebywhlchagenera|dleren-
tla|equatloncanbeso|vedexp|lclt|y. Indeed,theso|utlonsofsuchaslmp|e-|ooklng
dlerentla|equatlonas,' =x
:
+,
:
cannotbeexpressedlntermsoftheordln
e|ementaryfunctlonsstudledlnca|cu|ustextbooks. Neverthe|ess,thegraphlca|and
numerlca|methodsofthls and |ater sectlonscanbeusedtoconstruct.ppax/-.e
so|utlonsofdlerentla|equatlonsthatsumceformanypractlca|purposes.
Slope Fields and Graphical Solutions
There l s a slmp|e geometrlc way to thlnkaboutso|utlons ofa glven dlerentla|
equatlon,' = ),x, ,) . At each polnt ,x,,) ofthe\-p|ane, the va|ue of),x, ,)
determlnesas|ope- =),x,,) . Aso|utlonofthedlerentla|equatlonls slmp|ya
dlerentlab|efunctlonwhosegraph, =y,x; has thls 'corrects|opeateachpolnt
,x,y,xthrough whlch ltpasses~that ls, y ,x; = ),x, y,x; Thus a solution
curve ofthe dlerentla| equatlon ,'
= ),x,,)~the graph ofa so|utlon ofthe
equatlon~ls slmp|yacurve lnthe\-p|anewhosetangent|lneateachpolnt ,x,,)
has s|ope - = ),x, ,) . For lnstance, Flg. I . 3. I shows a so|utlon curve ofthe
dlerentla| equatlon ,' = x - , together wlth lts tangent |lnes at three typlca|
polnts.
Thls geometrlc vlewpolnt suggests a,.p//../-e/oaforconstructlng.p
pax/-.eso|utlons of the dlerentla| equatlon ,' = ),x, ,) . Througheachofa
representatlveco||ectlonofpolnts ,x,,) lnthep|anewedrawashort|lne segment
havlngthe proper s|ope- = ),x, ,) . A|| these|lne segments constltuteaslope
feld (oradirection feld) fortheequatlon,' =),x, ,) .
. ..... .......... ... . ...... ..... . . . .... . . . . . . ..... . . ... .... ......... . ............ .................
Flgures I . 3. 2(a)(d) shows|opene|dsandso|utloncurvesforthedlerentla|equa-
tlon
ay
= /y
ax
(2)
wlththeva|ues/=2, 0. 5, -I , and3ofthepameter/lnEq.(2). Notethateach
s|ope ne|d yle|ds lmportant qua|ltatlve lnformatlon about the set of a|| so|utlons
20 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| eZ
~
:
z
~

z
X
FIGURE 1.3.2(a) Slope feld
and solution curves for . 2y.
:
z

~ 0

z
:
X
FIGURE 1.3.2(c) Slope feld
and solution curves for y' ~ .
-
:
z
~ 0

z
X
FIGURE 1.3.2(b) Slope feld
and solution curves for
y' .

:
z

z
:
X
FIGURE 1.3.2(d) Slope feld
and solution curves for y' -3y.
ofthe dlerentla|equatlon. Forlnstance, Flgs. I . 3. 2(a) and (b) suggestthateach
so|utlon ,(x) approaches Jc asx +c lfk > 0, whereasFlgs. I . 3. 2(c) and
(d) suggestthat ,(x) 0 asx +c lfk < 0. Moreover, a|though the slgn
ofk determlnesthea/e./ooflncreaseordecreaseof,(x) , ltsabso|uteva|ue k,
appears to determlne the .eo)./.,eof,(x) . A|| thls ls apparent from s|ope
ne|ds |lke those ln Flg. I . 3. 2, even wlthout knowlng that the genera| so|utlon of
Eq. (2)ls glvenexp|lclt|yby,(x) = Ce

A s|ope ne|d suggests vlsua||y the genera| shapes ofso|utlon curves ofthe
dlerentla| equatlon. Through each polnta so|utloncurveshou|dproceed ln such
adlrectlonthatlts tangent |lne ls near|ypara||e|tothenearby |lne segmentsofthe
s|opene|d. Startlngatanylnltla|polnt,.,/; , wecanattempttosketchfreehandan
approxlmate so|utlon curve that threads lts way throughthe s|opene|d, fo||owlng
thevlslb|e|lne segments asc|ose|yasposslb|e.
Constructas|opene|dforthedlerentla|equatlon,' =x - , anduselttosketch
anapproxlmateso|utloncurvethatpassesthroughthepolnt(-4, 4) .
Sol ution So|utlonFlg. I . 3. 3showsatab|eofs|opesfortheglvenequatlon. Thenumerlca|
s|ope m = x - , appears atthe lntersectlon ofthe horlzonta|x-row and the ver-
tlca| ,-co|umnofthetab|e. Ifyou lnspectthepattem ofupper-|eftto |ower-rlght
dlagona|slnthls tab|e,youcanseethatltwas easl|yandqulck|yconstructed. (Of
4
3
I
O
-I
-4 -3 -2 -I O I 2 3 4
x
FIGURE 1.3.6. Slope feld and
typical solution curves for
y
'
= X - y.
x \
y -4 -3 -2
-4 0 -I -2
-3 I 0 -I
-2 2 I 0
-I 3 2 I
0 4 3 2
I 5 4 3
2 6 5 4
3 7 6 5
4 S 7 6
I . 3 Sl ope Fi el ds and Sol uti on Cures 21
-I 0 I 2 3 4
-3 -4 -5 -6 -7 -S
-2 -3 -4 -5 -6 -7
-I -2 -3 -4 -5 -6
0 -I -2 -3 -4 -5
I 0 -I -2 -3 -4
2 I 0 -I -2 -3
3 2 I 0 -I -2
4 3 2 I 0 -I
5 4 3 2 I 0
FIGURE 1.3.3. Values of the slope y
'
= X - y for -4 _ X, y _ +.
5
\ \ \ \ ` -
\ \ \ \ ` ~ Z
\ \ \ \ ` ~ Z /
\ \ \ \ ~ Z / /
O
\ \ ` / / / |
\ ` ~ Z
/ / | |
Z / / | | |
~ Z
/ / | | | |
-5
-5
,
X
FIGURE 1.3.4. Slope feld for
y
'
= X - y corresponding to the
table of slopes in Fig. 1 . 3. 3.
5
5 r
4
3
2
I
O
-I
-2
-3
` Z
-4 Z / /
-5
-5 O
X
FIGURE 1.3.5. The solution
curve through ,-+, +) .
5
course,amorecomp|lcatedfunctlon),x. y;ontherlght-handsldeof thedleren-
tla|equatlonwou|dnecessltatemorecomp|lcatedca|cu|atlons.) Flgure I . 3.4shows
thecorrespondlngs|opene|d,andFlg. I . 3. 5 shows anapproxlmate so|utlon curve
sketchedthrough the polnt (-4, 4) so as tofo||ow asthls s|opene|dasc|ose|yas
posslb|e.Ateachpolntltappearstoproceedlnthedlrectlonlndlcatedbythenearby
|lnesegmentsofthes|opene|d.
A|though a spreadsheetprogram (for lnstance) readl|y constructs atab|e of
s|opes as ln Flg. I . 3. 3, lt can bequltetedloustop|otby hand asumclentnumber
ofs|ope segments as ln Flg. I . 3.4. However, mostcomputera|gebra systemsln-
c|udecommandsforqulckandreadyconstructlonofs|opene|dswlthasmany|lne
segments asdeslred, such commandsarel||ustratedlntheapp|lcatlonmaterla|for
thlssectlon. Themore|lnesegments are constructed, themoreaccurate|yso|utlon
curves canbevlsua|lzedandsketched. Flgure I . 3. 6shows anners|opene|dfor
the dlerentla| equatlon y = x - yof Examp|e2, togetherwlthtyplca| so|utlon
curvestreadlngthroughthls s|opene|d.
Ifyou |ook c|ose|yatFlg. I . 3. 6, you may spot aso|utloncurvethatappears
tobe a stralght|lne' Indeed, you can verlfy that the|lnearfunctlony
-
x - i ls
aso|utlonoftheequatlon y =x- y. and lt appears|lke|y that theotherso|utlon
curves approach thls stralght |lne as an asymptote as x - +c. Thls lnference
l||ustratesthefactthatas|opene|dcansuggesttanglb|elnformatlonaboutso|utlons
thatlsnotata||evldentfromthedlerentla|equatlonltse|f. Canyou,bytraclngthe
22 Chapter I First-Order Di fferenti al Equati ons
cXump| e
| :
a
:
FIGUR 1.3.7. Slope feld and
typical solution curves for
U = 32
approprlateso|utloncurvelnthlsngure,lnferthat,(3) 2fortheso|utlon,(x) of
thelnltla|va|ueprob|em,' = x - ,, ,( -4) = 4!
Applications of Slope fields
Thenexttwoexamp|esl||ustratetheuseofs|opene|dstog|eanusefu|lnformatlon
lnphyslca|sltuatlonsthataremode|edbydlerentla|equatlons. Examp|e3lsbased
onthefactthatabaseba||movlngthroughthealratamoderatespeedu (|essthan
about300ft/s)encounters alrreslstancethatls approxlmate|yproportlona| to u. If
the baseba|| ls thrown stralghtdownwardfrom thetopofa ta|| bul|dlngorfrom a
hoverlnghe|lcopter, then ltexperlencesboth thedownwardacce|eratlonofgravlty
and an upwd acce|eratlon ofalrreslstance. Ifthe ,-axls ls dlrectedao.a,
thentheba|| ' sve|ocltyu = ay}aandltsgravltatlona|acce|eratlon,=32fJs
:
are
bothposltlve,whl|elts acce|eratlonduetoalrreslstancels negatlve. Henceltstota|
acce|eratlonlsoftheform
au
= , - /u.
a
Atyplca|va|ueofthealrreslstanceproportlona|ltyconstantmlghtbe/= 0. I 6.
(3)
Supposeyouthrow abaseba||stralghtdownwardfrom ahe|lcopterhoverlngatan
a|tltudeof3000feet. You wonderwhethersomeonestandlngonthegroundbe|ow
cou|d concelvab|ycatchlt. Inordertoestlmatethe speedwlthwhlch theba||wl||
|and, youcanuseyour|aptop' scomputera|gebrasystemtoconstructas|opene|d
forthedlerentla|equatlon
au
= 32 - 0. I 6u.
a
(4)
Theresu|tls shown lnFlg. I . 3. 7, togetherwlth a numberofso|utloncurves
correspondlngtodlerentva|uesofthelnltla|ve|oclty u (0) wlth whlch youmlght
throwthebaseba||downwd. Notethata||theseso|utloncurvesappeartoapproach
the horlzonta| |lne u = 200 as an asymptote. Thls lmp|les thathowever you
throw ltthe baseba|| shou|d approach the //-//,.e/o./(u = 200fJs lnstead
ofacce|eratlnglndennlte|y (asltwou|dlnthe absenceofany alrreslstance). The
handyfactthat60ml/h=SSfJsyle|ds
ft 60ml/h ml
u = 200- Z I 36. 36 -.
s SS fJs h
Perhaps a catcher accustomed to I 00 ml/h fastba||s wou|d have some chance of
ne|dlngthlsspeedlngba||.
Lomment Iftheba| | ' slnltla| ve|oclty ls u(0) = 200, then Eq. (4)glves
u' (0) = 32 - (0. I 6) (200) = 0, sotheba||experlencesolnltla|acce|eratlon. Its
ve|ocltythereforeremalnsunchanged,andhence u(t)= 200lsaconstantequl|lb-
rlumso|utlonofthedlerentla|equatlon. Ifthelnltla|ve|ocltyl sgreaterthan200,
thenthelnltla|acce|eratlonglvenbyEq. (4)lsnegatlve,sotheba||s|owsdownaslt
fa||s. Butlfthelnltla|ve|ocltyls|essthan200,thenthelnltla|acce|eratlonglvenby
(4) ls posltlve, sotheba|| speedsupasltfa||s. Itthereforeseems qultereasonab|e
that,becauseofalrreslstance,thebaseba||wl||approacha|lmltlngve|ocltyof200
fJswhateverlnltla| ve|oclty ltstarts wlth. Youmlght |lketoverlfythatlnthe
absenceofalrreslstancethlsba||wou|dhltthegroundatover300ml/h.
cXump| e4
I .3 Sl ope Fi el ds and Sol uti on Cures 23
InSectlon i :wewl||dlscusslndetal|the|oglstlcdlerentla|equatlon
ai
= /i(M - i)
a
,
thatoften l s used to mode| a popu|atlon i, ; that lnhablts an envlronment wlth
.c/,.cpc./-M ThlsmeansthatMlsthemaxlmumpopu|atlonthatthlsenvl-
ronmentcansustaInona|ong-termbasls(lntermsofthemaxlmumaval|ab|efood,
forlnstance) .
Ifwetake/= c ccc:andM= i c, thenthe|oglstlcequatlonln,;takestheforn
ai

= c ccc:i, i c- i; =c c:i- 0. 0004P .


a
,:
Theposltlvetermc c:iontherlghtln,:;correspondsto natura|growthata6
annua|rate (wlthtlmemeasuredlnyears). Thenegatlveterm -c ccc:i

repre-
sentsthelnhlbltlonofgrowth dueto|lmltedresourceslntheenvlronment.
Flgurei sshowsas|opene|dforEq. ,:;, togetherwlthanumberofso|utlon
curves correspondlng to posslb|e dlerent va|ues ofthe lnltla| popu|atlon P(0) .

Note that a|| theseso|utloncurves appeartoapproachthehorlzonta| |lne i i c


asanasymptote. Thl slmp|lesthatwhateverthelnltla|popu|atlonthepopu|atlon
/
i, ; approachesthe//-//,pop/c/oi= i cas-- 0.

FIGUR 1.3.8. Slope feld and


typical solution curves for
=

cXump| e
Lomment Ifthelnltla|popu|atlonls P(0) = i c,thenEq. ,:;glves
P
'
(0) = c ccc:, i c; , i c- i c;= c,
sothepopu|atlonexperlencesolnltla|(lnstantaneous)change. Itthereforeremalns
unchanged, andhence i, ; = i cls aconstant'equl|lbrlumso|utlonofthedlf-
ferentla|equatlon. Ifthelnltla|popu|atlonlsgreaterthan i c,thenthelnltla|rateof
changeglvenby ,:;ls negatlve, so the popu|atlonlmmedlate|ybeglnstodecrease.
Butlfthelnltla|popu|atlonls |essthan i c,thenthelnltla|rateofchangeglvenby
,:;lsposltlve,sothepopu|atlonlmmedlate|ybeglnstolncrease. Itthereforeseems
qulte reasonab|etoconc|udethatthepopu|atlon wl|| approach a |lmltlngva|ue of
I 50whateverthe(posltlve)lnltla|popu|atlon.
Existence and Uniqueness of Solutions
Before onespendsmuchtlmeattemptlngtoso|veaglvendlerentla| equatlon, lt
l swlsetoknowthatso|utlons actua||y e/s We may a|sowanttoknow whether
there ls on|y one so|utlonoftheequatlonsatlsfylngaglvenlnltla|condltlonthat
l s, whetherlts so|utlonsare/,e
(a) Fal|ureofexlstence] Thelnltla|va|ueprob|em
, i
, = - ,(0) =c
x
,:
has oso|utlon,becausenoso|utlony,x; = ], i}x; ax= ln x +C ofthedler-
entla| equatlon ls denned atx = 0. We see thls graphlca||y ln Flg. i ,whlch
showsadlrectlon ne|d and sometyplca|so|utloncurvesforthe equatlon ,' = i}x
Itls apparent thatthelndlcateddlrectlon ne|d 'forces a|| so|utloncurvesnearthe
,-axlstop|ungedownwardso thatnonecanpassthroughthepolnt,c,c;
24 Chapter I Fi rst-Order Di fferenti al Equati ons
,
/
l
u x
FIGURE 1.3. 11. The rectangle
.and x-interval of Theorem
and the solution curve y = y(x)
through the point .
FIGURE 1.3.9. Direction feld
and typical solution curves for
the equation y' = l/x.

/ / /
/ / /
/ / /
/ / /
/ / /
Z Z Z
Z Z Z
/ / / /
/ / / /
/ / / /
/ / / /
,y
,
(x)= x
Z Z Z Z
x
/
Z Z
Z Z
FIGURE 1.3.10. Direction feld and two
diferent solution curves for the initial value
problem y' = 2., y(O) = O.
(b) Fal|ureofunlqueness] Ontheotherhand,youcanreadl|yverlfythatthelnltla|
va|ueprob|em
,' =2J, ,(0) =c ,s;
hasthemodlerentso|utlons
y, ,x; =x

andy

,x; = c(seeProb|em27). Flgure


i i cshows adlrectlonne|dandthesetwodlerent so|utlon curvesforthelnltla|
va|ueprob|emln,s; Weseethatthecurvey, (x) =x

threadsltswaythroughthe
lndlcateddlrectlonne|d,whereasthedlerentla|equatlon,' =2. speclness|ope
,' = ca|ongthex- axlsy

,x;=0
Examp|e l||ustrates the fact that, before we can speak of the so|utlon of
an lnltla| va|ue prob|em, we need to know that lt has oe cao/yoeso|utlon.
Questlons of exlstence and unlqueness of so|utlons a|so bear on the process of
mathematlca| mode|lng. Supposethatwearestudylngaphyslca|systemwhosebe-
havlorlscomp|ete|ydetermlnedbycertalnlnltla|condltlons,butthatourproposed
mathematlca| mode| lnvo|ves a dlerentla| equatlon ohavlng a unlque so|utlon
satlsfylng those condltlons. Thls ralses an lmmedlate questlon as to whetherthe
mathematlca|mode|adequate|yrepresentsthephyslca|system.
Thetheoremstatedbe|owlmp|lesthatthelnltla|va|ueprob|em,' =](x, ,) ,
y,c;=/hasoneandon|yoneso|utlondennednearthepolntx =conthex-axls,
provlded thatboththefunctlon ] andlts partla| derlvatlve ]J, are contlnuous
nearthepolnt,c,/;lnthe\-p|ane. Methodsofprovlngexlstenceandunlqueness
theoremsaredlscussedlntheAppendlx.
THEOREM 1 Existence and Uniqueness of Sol utions
Supposethatboththe functlon ](x, ,) andltspartla|derlvatlve D
,
](x, ,) 8
contlnuous on somerectang|e r ln the x,-p|ane that contalns the polnt ,c,/;
ln lts lnterlor. Then, for some open lnterva| I contalnlngthepolntc,thelnltla|
va|ueprob|em
ay
ax
= ](x, ,) , ,(a)= / ,;
hasoneandon|yoneso|utlonthatlsdennedonthelnterva| I. (Asl||ustratedln
Flg.i i i , theso|utlonlnterva| Imaynotbeaswldeastheorlglna|rectang|e
rofcontlnulty, seeRemarkbe|ow. )
6
4
:
g
-:
-4
y t/( t

_ . . . .
/
| | R
|
(O, t,
1 ~
-:
g

|
:
FIGUR 1.3.12. The solution
curve through the initial point
(0, 1 ) leaves the rectangle R
before it reaches the right side of
R.
cXump| e
4
I .3 Sl ope Fiel ds and Sol uti on Cures 25
Hemark 1 : Inthecaseofthedlerentla|equatlond,Jdx = -,ofExam-
p|e I and Flg. I . 3. 2(c), boththefunctlon ](x, ,) = -, andthepartla|derlvatlve
]J3, = -I arecontlnuouseverywhere, soTheorem I lmp|lestheexlstenceofa
unlqueso|utlonforanylnltla|data ,c./; A|thoughthetheoremensuresexlstence
on|y on some open lnterva|contalnlngx = c,eachso|utlon,(x) = ce

actua||y
ls dennedfora||x.
Hemark Zt In the case ofthe dlerentla| equatlon d,Jdx = -2_of
Examp|e 5(b) and Eq. ,s;, thefunctlon ](x, ,) = -2, ls contlnuouswherever
, > 0, butthepartla|derlvatlve ]J3, = IJ_ls dlscontlnuouswhen, =0,and
henceatthepolnt (0,0) . Thls ls why lt ls posslb|efortheretoexlsttwodlerent
so|utlons )
t
(x) = x
:
and )
:
(x) = 0, each ofwhlch satlsnesthelnltla| condltlon
,(0) = 0.
Hemark t InExamp|e:ofSectlon I . I weexamlnedtheespecla||yslm-
p|edlerentla|equatlond,Jdx = ,
:
. Herewehave](x, ,) = ,
:
and]J, =2,.
Both ofthesefunctlons e contlnuouseverywhere ln thex,-p|ane, and ln partlc-
u|ar onthe rectang|e -2 < x < 2, 0 < , < 2. Becausethe polnt (0, I ) |les ln
thelnterlorofthlsrectang|e,Theorem I guarantees aunlqueso|utlonnecessarl|y
acontlnuousfunctlonofthelnltla|va|ueprob|em
d,
:
= , , ,(0) = I
dx
onsomeopenx-lnterva|contalnlngc= 0. Indeedthlslstheso|utlon
I
,(x) = =
I - x
( I 0)
thatwedlscussedln Examp|e: But,(x) = IJ( I - x) ls dlscontlnuousatx = I ,
s oourunlquecontlnuousso|utlondoesnotexlstontheentlrelnterva|-2 < x < 2.
Thus the so|utlonlnterva| i ofTheorem I maynotbe as wldeas the rectang|e K
where ] and ]J, e contlnuous. Geometrlca||y, thereason ls thatthe so|utlon
curve provlded bythe theoremmay |eavethe rectang|ewhereln so|utlons ofthe
dlerentla|equatlonareguaranteedtoexlstbeforeltreachestheoneorbothends
of thelnterva| (seeFlg. I . 3 . I 2) .
The fo||owlng examp|e shows that, lfthe functlon ](x,,) and/orltspartla|
derlvatlve 3]J3, fal| to satlsfy the contlnulty hypothesls ofTheorem I , then the
lnltla| va|ue prob|em ln (9)mayhave e//eno so|utlonomany-~ven lnnnlte|y
manyso|utlons.
Conslderthenrst-orderdlerentla|equatlon
d,
x
dx
= 2,. ( I I )
App|ylng Theorem I wlth ](x, ,) = 2,Jx and ]J, = 2Jx, we conc|ude that
Eq. ( I I ) musthaveaunlqueso|utlonnearanypolntln the\-p|anewherex = 0.
Indeed,weseelmmedlate|ybysubstltutlonln( I I ) that
,(x) = cx
:
( I 2)
26 Chapter I Fi rst-Order Di fferenti al Equati ons
|
O./) |
O. O)
X
FIGURE 1.3.13. There are
infnitely many solution curves
through the point but no
solution curves through the point
b) if b ;
X
FIGURE 1.3.14. There are
infnitely many solution curves
through the point ( l , - 1 ) .
satlsnesEq. ( I I ) forany va|ueoftheconstantcandfora||va|uesofthevarlab|e
x. Inptlcu|ar,thelnltla|va|ueprob|em
d,
x
dx
= 2,, ,(0) = 0 ( I 3)
haslnnnlte|ymany dlerentso|utlons,whoseso|utloncurvesaretheparabo|as, =
Cx
:
l||ustratedlnFlg. I . 3 . I 3 . (In casec = ctheparabo|a ls actua||ythex-axls
, = c ;
Observe that a|| theseparabo|aspassthrough the orlgln ,c, c; , but none of
thempassesthrough any other polntonthe ,-axls. Itfo||ows that thelnltla|va|ue
prob|emln( I 3)haslnnnlte|ymanyso|utlons,butthelnltla|va|ueprob|em
d,
x
dx
= 2,, ,(0) = b ( I 4)
has noso|utlonlfb =0.
Flna||y,notethatthroughanypolntothe,-axlstherepasseson|yoneofthe
parabo|as, = Cx
:
Hence,lfa =c,thenthelnltla|va|ueprob|em
d,
x
dx
= 2,, ,(a) =b ( I 5)
hasaunlqueso|utlononanylnterva|thatcontalnsthepolntx =abutnottheorlgln
x = cInsummary,thelnltla|va|ueprob|emln( I 5)has
aunlqueso|utlonnear (a,b) lfa =c.
noso|utlonlfa = cbutb =c,
lnnnlte|ymany so|utlonslfa = b = 0.
Stl|| more can be sald about the lnltla| va|ue prob|em ln ( I 5) . Conslder a
typlca| lnltla|polntothe,-axlsforlnstancethepolnt( -I , I ) lndlcatedlnFlg.
I . 3 . I 4. Thenforanyva|ueof theconstantcthefunctlondennedby
,(x) =

x
:
lfx _ c,
Cx
:
lfx > c
ls contlnuousandsatlsnesthelnltla| va|ueprob|em
d,
x - = 2,, ,( -I ) = I .
dx
( I 6)
( I 7)
Forapartlcu|ar va|ue ofc, the so|utlon curve dennedby ( I 6) conslstsofthe |eft
ha|foftheparabo|a, = x
:
and the rlght ha|foftheparabo|a, = Cx
:
. Thus the
unlque so|utlon curve near ( -I , I ) branches at the orlgln lnto the lnnnlte|y many
so|utloncurvesl||ustratedlnFlg. I . 3. I 4.
We therefore see thatTheorem I ( l fltshypotheses are satlsned) guarantees
unlquenessoftheso|utlonnearthelnltla|polnt (a,b), butaso|utloncurvethrough
(a,b) mayeventua||ybranche|sewheresothatunlquenessls |ost. Thus aso|utlon
mayexlstona|gerlnterva|thanoneonwhlchtheso|utlonlsunlque. Forlnstance,
the so|utlon,(x) = x
:
ofthelnltla| va|ue prob|em ln ( I 7) exlstsonthe who|ex-
axls, butthlsso|utlonl sunlqueon|yonthenegatlvex-axls-0 < x < 0.
Problems
..I ..,.I0, ...,....,.
......, ..,........
.....................
...,.........,.
1
..

. ...
..
~
:
z
0
' ' ' '
` ` ' ' '
' ` ` ` ` ` ` ` '
\g\
`
@ ` \ \
` ` ` `

` ` `
/ / 1 1 / / /
-l
.
\
v
gv v
\ \ v v \ \
` ' ' ' ' ' ' ' '
` ` ' ' ' ' ' ' `
` ` ` ' ' ' ` `
1 1 1 1 1 1 1 1 / / / / " " " " / /
- z
-
/ / / / / / / / / 1 1 1 / / / 1 / 1
/ / / / / / / / / I 1 1 1 1 1 1 1 1
: -z -l 0 z
FIGURE 1.3.15.
.
2.
..
..
- - / / / / 1 1 1
- - - / / / / / /
z
_
X
/ / / / / / |
I / / / / / /
/ / q / q/
I I / / / / / /
I I I / / / / /
/ I I I / / / / /
/ / 1 1 1 1 1 1 1
/ / 1 1 1 1 1 1 1
/ / / / / 1 1 1 1
:
~ 0
-l
-z
FIGURE 1.3.16.
3
..

...
..
z
/ / / / / / / /
/ / / / / / I I
I / / / / I
I I / / / / I I I
I I I I I I I I I
x
I I I I 1 I 1 1 1
1 1 / / / / 1 1 1
I / ,/ / /. / I
/ / / / / / / / /
/ I I I I I I / /
/ 1 1 1 1 1 1 / /
/ / / I / / / /
.
-l
` ` `
-z
`

' ` ` ` ` ` ` ' '


' ' ` ` ` ` ' ' '
: -z -l 0
X
FIGURE 1.3.17.
z
:
:
I . 3 Sl ope Fi el ds and Sol uti on Cures 27
.
4. .
..
z
-l
-z ` `
- - - / / / / 1 1
- - / / / / 1 1 1
: -z - l
FIGURE 1.3.18.
.
5. ] .
..
|
|
z
|
|
/ /

/ /
~ 0
/ /
0 z
x
` ` ` ` '

-
` ` ` ` ' '
- l
- z
- l
FIGURE 1.3.19.
.
6. .
..
z
\ \ \
' '
'
t \
'
\
\ g\ v
' '
_\
' ' ' \ ' ' ` `

' '
\ ' ' ` `

' ' ' ' ` `

`
0
x
`
`
\v
' \
` ` ` ` ' ' \ '
` ` ` ' ' ' ' '

` ` ' ' ' ' ' '


\

v
\ \ \ \ \ \ \
\
'
v \ \
z
' ' ` ` ` `
' ' ` ` ` `
~
'
\
-l
. . w .
- - - / / / 1 1 1
-z
/ / / / 1 1 1 1 /
/ / I I I / / /
: -z -l 0
X
FIGURE 1.3.20.
:
:
28 Chapter I First-Order Di fferenti al Equati ons
7
..

... ...
..
z
-
.
-z
-z -l
FIGURE 1.3.21.
..
8. .
..
9.
1
z
- 0
.
-z
FIGURE 1.3.22.
..
. .
..
1
z
- 0
/ l / -~ \
/ / / - -
.
| / / ' - ~
I / l / / - ~ ~
| | / / / - - ~
-z
| | f I I / .. --
/ / P ~
| | f I / / - -
| | | I / I 1 / ,.
-1
-1 -z .
FIGURE 1.3.23.
` . . . ..
0 z 1
X
X
0 z 1
X

..

1 . . ...
..
-
.
-z
FIGURE 1.3.24.
X
..11 ..,.20, ........1 ..
...,...........,...
..,...,...........
..1 .....,.....,.......
..
11.
..

..
..
.
..
. 12. . . .
..
..
13. , .
..
..
14.
..
, .0
15.
..
. ,. .
..
..
16.
..
,. . .
17.
..
.
.
..
.
18. . . . 0
..
19.
..
0

.
..
20.
..
.

.
.

..
..2I ...22, .........,2
.....,..,.., ..,...
..........,...,.,..
..................
...........
21. . . . .
0;
.
22. .. . ..0; . .
.2J ...24 ...2I ...22, ...
...,..,....,...,....,
..,.., ..,...............
........................
,.,...,.
23.
24.
25.
. = .+ .- 1 , .O) = .= ?
1
.= .+ _ . = .= ?
You bail out of the helicopter of Example and pull the
ripcord of your parachute. Now .= 1 . 6 in Eq. so
your downward velocity satisfes the initial value problem
.v
.
= - 1 . 6v, v(O) = O.
In order to investigate your chances of survival, construct
a slope feld for this diferential equation and sketch the
appropriate solution curve. What will your limiting veloc
ity be? Will a strategically located haystack do any good?
How long will it take you to reach of your limiting
velocity?
26. Suppose the deer population .a small forest satisfes
the logistic equation
.
.
=

Construct a slope feld and appropriate solution curve to


answer the following questions: If there are deer at
time = and is measured in months, how long will
it take the number of deer to double? What will be the
limiting deer population?
......,...........,.
....I .............,
. = . ... = ..........
...................
27. (a) Verify that if is a constant, then the function defned
piecewise by
.. =
, .
for .
for . ~
satisfes the diferential equation .= for all .(in
cluding the point .= Construct a fgure illustrating the
fact that the initial value problem .= .= has
infnitely many diferent solutions. (b) For what values of
does the initial value problem . = .= have
(i) no solution, (ii) a unique solution that is defned for all
x?
28. Verify that i f .i s a constant, then the function .. = ..
satisfes the diferential equation ..= .for all .Con
strct a slope feld and several of these straight line so
lution curves. Then determine (in terms of .and how
many diferent solutions the initial value problem ..= .
.. = has-ne, none, or infnitely many.
I . Sl ope Fi el ds and Sol uti on Curves 29
29. Verify that if is a constant, then the function defned
piecewise by
.. =
,

.
for .
for . ~
satisfes the diferential equation .= .

for all .Can


you also use the "lef half" of the cubic .= .

in
piecing together a solution curve of the diferential equa
tion? (See Fig. Sketch a variety of such solution
curves. Is there a point .of the xy-plane such that
the initial value problem .= .

..= has either


no solution or a unique solution that is defned for all .
Reconcile your answer with Theorem 1 .
)
C
FIGUR 1.3.25. A suggestion for Problem
30. Verify that if is a constant, then the function defned
piecewise by

+1
.. = cos (x
-1
if .
.
if . +
satisfes the diferential equation .= -for all .
(Perhaps a preliminary sketch with = will be helpful . )
Sketch a variety of such solution curves. Then deterine
(in terms of .and how many diferent solutions the ini
tial value problem . = ..= has.
31. Carry out an investigation similar to that in Problem
except with the diferential equation . = +.
Does it suffce simply to replace cos (x with sin .-L)
in piecing together a solution that is defned for all .
32. Verify that i f ~ then the function defned piecewise
by
,
O f
( )
1 X
..=
. if .~
satisfes the diferential equation .= ..for all .
Sketch a variety of such solution curves for diferent val
ues of Then determine (in terms of .and how many
diferent solutions the initial value problem .= ..
..= has.
30 Chapter I First-Order Di fferenti al Equati ons
33. If .verify that the function defned by .. =
.. 1 ) (with graph illustrated in Fig. 1 . 3. 26) satisfes
the diferential equation .. .=

if ..1/c. Sketch
a variety of such solution curves for diferent values of
Also, note the constant-valued function .. =

that
does not result fom any choice of the constant Finally,
determine (in terms of u and how many diferent solu
tions the initial value problem .. .= .. =
has.
` ` \
` \
` \

`


'
` \ '
` \ \ '
` ` \ \ '
|
'
'
'
'
X
+
,
. -


.\ \ `
.
\ \ `
.

\ \ ` `
FIGURE 1.3.26. Slope feld for .. .=

and graph of a solution .. = .. 1 ) .


1 . | COl O
34. (a) Use the direction feld of Problem to estimate the
values at .= 1 of the two solutions of the difer
ential equation .= . .1 with initial values
. = -1 . 2 and . =
(b) Use a computer algebra system to estimate the val
ues at .= 3 of the two solutions of this diferen
tial equation with initial values . -3) = -3. 01 and
.-3) = -2. 99.
The lesson of this problem is that small changes in initial
conditions can make big diferences in results.
35. (a) Use the direction feld of Problem 6 to estimate the
values at .= 2 of the two solutions of the difer
ential equation .= . .1 with initial values
.-3) = and .-3) =
(b) Use a computer algebra system to estimate the val
ues at .= 3 of the two solutions of this diferen
tial equation with initial values .-3) = and
y(-3) =
The lesson of this problem is that big changes i n initial
conditions may make only small differences in results.
Wlde|yaval|ab|ecomputera|gebrasystemsandtechnlca|computlngenvlronments
lnc|udefacl|ltlestoautomatetheconstructlonofs|opene|dsandso|utloncurves, as
do somegraphlngca|cu|ators(seeFlg. I . 3. 27).
FIGURE 1.3.27. Slope feld and solution curves for the diferential
equation
..
.
..
= ... .
with initial points = -3, -1 , -2, 0, 2, 4 and window
. . on a TI- 89 graphing calculator.
The app|lcatlons manua| accompanylngthls textbook lnc|udes dlscusslonof
Mcp/e

.Mc/e-c/.c

. and MATLAB
T
V
resources for the lnvestlgatlonofdlf-
ferentla|equatlons. Forlnstance,theMcp/ecommand
with ( DEtools ) :
DEplot ( diff ( y( x ) , x ) =sin ( x-y ( x , y ( x ) , x=-5 5 , y=-5
5 ) J
5
4
l
2


-2
0 2 l 4 5
X
FIGUR 1.3.28. Computer
generated slope feld and solution
curves for the diferential equation
y' = sin(x - y) .
andtheMc/e-c/.ccommand
I . Sl ope Fi el ds and Sol uti on Cures 3\
* Graphics\PlotField . m
PlotVectorField [ { l , Sin [ x-
y] } , { x , -
5 , 5 } , {y, -
5 , 5 } ]
produce s|ope ne|ds slml|ar to the one shown ln Flg. I . 3 . 2S. Flgure I . 3. 2S lt-
se|fwas generated wlth the MA1LAu program dfield John Po|klng andDavld
Amo|d, oa/cn_ ee/c/i,c/os0s/,MA1LAu, 2ndedltlon,UpperSad-
d|e Rlver, NJ. Prentlce Ha||, I 999] that ls free|y aval|ab|e for educatlona| use
(math.rice.edul"dfeld). When a dlerentla| equatlon ls entered lnthe dfield
setupmenu(Flg. I . 3. 29), youcan(wlthmousebuttonc|lcks)p|otbothas|opene|d
andthe so|utloncurve (orcurves)through any deslred polnt (orpolnts). Another
free|y aval|ab|e and user-frlend|y MA1LAu-based ODE package wlth lmpresslve
graphlca|capabl|ltlesls lode (www.math.uiuc.eduiode).
FIGURE 1.3.29. NA1LA dfield setup to construct slope feld and solution curves
for y' = sin(x - y) .
Useagraphlngca|cu|atororcomputersystemlnthefo||owlnglnvestlgatlons.
You mlght warm up by generatlng the s|ope ne|ds and some so|utlon curves for
Prob|emsI through I 0 lnthl ssectlon.
INVESTIGATION A. P|otas|opene|dandtyplca|so|utloncurvesforthedleren-
tla|equatlona,Jax = sln(x- ,) , butwltha|argerwlndowthanthatofFlg. I . 3. 2S.
Wlth-I 0 x I 0, - I 0 , I 0, forlnstance, anumberofapparentstralght
|lneso|utloncurvesshou|dbevlslb|e.
(a) Substltute, =cx+/lnthedlerentla|equatlontodetermlnewhatthecoefn-
clentscand/mustbeln ordertogetaso|utlon.
(b) Acomputera|gebrasystemglvesthegenera|so|utlon
,(x) = x - 2 tan
|

x - 2 - C
x - C
P|otthl sso|utlonwlthse|ectedva|uesoftheconstantCtocomparetheresu|tlng
so|utloncurveswlththoselndlcatedlnFlg. I . 3. 2S. Canyouseethatova|ueof
C yle|dsthe|lnearso|utlon, = x - J2correspondlngtothelnltla|condltlon
,(J2) = 0! Arethereany va|ues ofC forwhlch thecorrespondlng so|utlon
curves|lec|osetothlsstralght|lne so|utloncurve!
32 Chapter I Fi rst-Order Di fferenti al Equati ons
INVESTIGATION B: For yourown persona| lnvestlgatlon,| etn bethe sma//est
dlgltlnyour studentO numberthatls greaterthan I , andconslderthedlerentla|
equatlon
d, I
= cos (x - n,) .
dx n
(a) Flrstlnvestlgate(as lnpart(a)ofInvestlgatlonA)theposslbl|ltyofstralght|lne
so|utlons.
(b) Thengenerateas|opene|dforthlsdlerentla|equatlon,wlththevlewlngwln-
dowchosensothatyoucanplcturesomeofthesestralght|lnes, p|usasumclent
numberofnon|lnearso|utloncurves thatyoucanformu|ateaconectureabout
whathappensto,(x) asx - +c. State yourlnferenceasp|aln|yasyoucan.
Glven the lnltla|va|ue ,(0) =
topredlct(perhapslntermsof
how
,(x) behavesasx - +c.
(c) Acomputera|gebrasystemglvesthegenera|so|utlon
Can you make a connectlon between thls symbo|lc so|utlon and yourgraphl-
ca||ygeneratedso|utloncurves (stralght|lnesorotherwlse)!
Thenrst-orderdlerentla|equatlon
d,
= H(x, ,)
dx
(I)
lsca||edseparable provldedthatH(x,,) canbewrlttenastheproductofafunctlon
ofx andafunctlonof,.
d, g(x)
dx
= g(x) h(,) =
](,)

whereh(,)= IJ ](,) . Inthlscasethevarlab|esxand,canbeseartedlso|ated


onopposltesldesofanequatlonbywrltlnglnforma||ytheequatlon
](,)d,= g(x) dx,
whlchweunderstandtobeconclsenotatlonforthedlerentla|equatlon
d,
](,)
dx
= g(x) . (2)
It ls easy to so|ve thls specla| type ofdlerentla| equatlon slmp|y by lntegratlng
bothsldeswlthrespecttox.
](, (x) dx =g(x) dx+ c
.
cXump| e 1
I . 4 Separabl e Equati ons and Appl i cati ons 33
equlva|ent|y,
),y;ay= ,,x; ax+c (3)
A||thatl srequlredlsthattheantlderlvatlves
i,y;= ),y;ay and o,x;= ,,x; ax
canbefound.ToseethatEqs. (2) and ,;areequlva|ent, notethefo||owlngconse-
quenceofthechalnru|e.
/ /
ay
n

i,y,x; ; } = i,y,x; ; y,x;= ),y; - = ,,x; =n

o,x; } .
ax
whlchlntumlsequlva|entto
i,y,x; ; = o,x; +c, ,:;
because two functlons have the same derlvatlve on an lnterva| lfand on|y lfthey
dlerbyaconstantonthatlnterva|.
So|vethelnltla| va|ueprob|em
ay
= -:xy, y,c;=:
ax
Sol ution Informa||y,wedlvldebothsldesofthedlerentla|equatlonbyyandmu|tlp|yeach
sldebyaxtoget

|
6
|
|
4
|
|
:
l
- OP

:
-4
-6
-

FIGUR 1.4. 1. Slope feld and


solution curves for y' = -6xy in
Example 1 .
Hence
ay
= -:x ax
y

a
= ,-:x; ax,
l n

y= -x
:
+c
We seefromthelnltla|condltlony,c; = :thaty,x; l s posltlvenearx =c,sowe
mayde|etetheabso|uteva|uesymbo|s.
andhence
|n , = -x
:
+ c,
y,x; = e

= e

= ~e

,
where ~=e

Thecondltlony,c;=:yle|ds~=:,sothedeslredso|utlonls
y,x; = :e
-

Thlsls theupperemphaslzedso|utloncurveshownlnFlg. i : i
J4 Chapter I First-Order Di fferenti al Equati ons
cXump| eZ
Memarkt Suppose,lnstead,thatthelnltla|condltlonlnExamp|eIhadbeen
y,c; = -4. Then ltwou|d fo||ow that , (x) ls e,c/.enear x = c We shou|d
therefore rep|ace

y wlth -y ln the lntegrated equatlon ln, y = -x


:
+ c to
obtaln
l n(-y;=-x
:
+c
Thelnltl a|condltlonthenyle|dsc=ln 4, so|a,-y; =-x
:
+ln 4, andhence
Thlslsthe|oweremphaslzedso|utloncurvelnFlg. I .4. I .
So|vethedlerentla|equatlon
ay 4- 2x
ax
=
y
:
-

,;
Sol ution Whenweseparatethevarlab|esandlntegratebothsldes, weget
. . . . . . . =' = = = - -
+
. . w . ~ = . ( I , 3) - - -
:

= = ~ ~ ~ ~ ~ ~ ~ ~ ~
-
s
-
s

: O : + s

x
FIGU 1.4.2. Slope feld and
solution curves for
.= = ..= in
Example
(3,- ) ay=(4- zx; ax,
, - y=4x- x
:
+c
Thlsequatlonl snotreadl|yso|vedforyasanexp|lcltmnctlonofx
,:;

AsExamp|e2 l||ustrates, ltmay ormaynotbeposslb|eorpractlca|to so|ve


Eq. (4) exp|lclt|yforylnterms ofx Ifnot, then we ca|| (4) an /-p//./so//o
ofthe dlerentla| equatlon ln (2). Thus Eq. ,:; glves an lmp|lclt so|utlon ofthe
dlerentla|equatlonln ,; A|thoughlt ls notconvenlenttoso|veEq.,:;exp|lclt|y
lntermsofx,weseethateachso|utloncurvey=y(x) |lesonacontour(or|eve|)
curvewherethefunctlon
(x, y;=x
:
- 4x+,- y
lsconstant.Flgure I . 4. 2shows severa| ofthesecontourcurves.

oso|

th

va

em


ay 4- 2x
- = , y, i ; =3 ,
ax y
:
-
,:;
wesubstltutex= I andy=3 lnEq. ,:;andgetc=9. Thusthedeslredpartlcu|ar
so|utlon,(x) lsdennedlmp|lclt|ybytheequatlon
, - y=4x - x
:
+9. (S)
The conespondlng so|utlon curve y = , (x) |les on the upper contourcurveln
Flg. I . 4. 2the onepasslngthrough ( I , 3) . Because the graph ofa dlerentlab|e
so|utlon cannot have a vertlca| tangent |lne anywhere, lt appears nom the ngure
thatthls partlcu|arso|utlonls dennedonthelnterva| ( -I , ;butnotonthelnterva|
, -, :;
z0
l 5
l 0
5

0
5
l 0
l 5
z0
- z 0 z
,
FIGUR 1.4.3. Graph of
.= .

FIGUR 1.4.4. Slope feld and


solution curves for .= ..
-
I . 4 Separabl e Equati ons and Appl i cati ons J5
Memark 1 : When a speclncva|ueofx lssubstltutedlnEq. ,s;.wecan
attempttoso|venumerlca||yfor,. Forlnstance,x =:yle|dstheequatlon
](,) =, - y- =0.
Flgure I .4. 3showsthegraphof]. Wlthagraphlngca|cu|atorwecanso|veforthe
slng|erea| root , z sz Thlsyle|dstheva|ue,(4) z szofthepartlcu|ar
so|utlonlnExamp|e3.
Memark Zt I fthe lnltla| condltlon l n,:;lsrep|aced wlth the condltlon
,( I ) = c,then the resu|tlng partlcu|ar so|utlon ofthe dlerentla| equatlon ln (5)
|lesonthe|owerha|foftheova|contourcurvelnFlg. i : z Itappearsthatthls
partlcu|ar so|utlon through , i . c; ls denned on the lnterva| ,c. :; but not on the
lnterva|( I , ) . On theotherhand,wlththelnltla|condltlon,( I ) -zwegetthe
|owercontourcurvelnFlg. I . 4. 2. Thlspartlcu|arso|utlonls dennedfora||x. Thus
the lnltla| condltlon can determlne whethera partlcu|arso|utlon ls dennedon the
who|erea| |lneoron|yonsomeboundedlnterva|. Wlthacomputera|gebrasysten
onecanreadl|yca|cu|ateatab|eofva|uesofthe,-so|utlonsofEq. ,s;forx-va|ues
atdeslredlncrementsfromx = -I tox = (forlnstance). Such atab|eofva|ues
serveseectlve|yasan
_
merlca|so|utlonofthelnltla|va|ueprob|enln,:;
Implicit, General, and Singular Solutions
TheequatlonK(x,,) = clscommon|yca||edanimplicit solution ofadlerentla|
equatlon lf lt ls satlsned (on some lnterva|) by some so|utlon , = ,(x) of the
dlerentla|equatlon. Butnote thatapartlcu|arso|utlon, = ,(x) ofK(x,,) - o
may or may not satlsfy a glven lnltla| condltlon. For examp|e, dlerentlatlon of
x
:
+,
:
=:yle|ds
d,
x + , - = c.
dx
sox
:
+,
:
=:ls anlmp|lcltso|utlonofthedlerentla|equatlonx+,,' oBut
on|ythenrstofthetwoexp|lcltso|utlons
,(x) =+
,
4- x
:
and ,(x) =-
,
:- x
:
satlsnesthelnltla|condltlon,(0) =z(Flg. i : :;
Hemark 1 : You shou|dnotassumethateveryposslb|ea|gebralcso|utlon
, = ,(x) ofan lmp|lclt so|utlon satlsnes the same dlerentla| equatlon. Forln-
stance, lfwemu|tlp|ythelmp|lcltso|utlonx
:
+,
:
- :=cbythefactor(,- 2x) ,
thenwegetthenewlmp|lcltso|utlon
(,- 2x) (x
:
+,- :;=c
that yle|ds (or 'contalns) not on|y the prevlous|y noted exp|lclt so|utlons , -
+

:- x
:
and, = -

:- x
:
ofthedlerentla|equatlonx+,,' =c,buta|sothe
addltlona|functlon, =zxthatdoes osatlsfy thls dlfferentla|equatlon.
Hemark Zt Slml|ar|y, so|utlons ofa glven dlerentla|equatlon can be
elther galned or |ost when lt ls mu|tlp|led or dlvlded by an a|gebralcfactor. For
lnstance,conslderthe dlerentla|equatlon
d,
(,- 2x) ,- =-x(,- zx;
dx
,;
36 Chapter I Fi rst-Order Di fferenti al Equati ons
-I 5 -I O -5 O 5 I O I 5
X
FIGUR 1.4.5. The general
solution curves y = (x - C)
2
and
the singular solution curve y =
of the diferential equation
(y')
2
= 4y.
cXump| e4
havlngthe obvlous so|utlon y = zx But lfwedlvldeboth sldesbythecommon
factor,y- zx; , thenwegettheprevlous|ydlscusseddlerentla|equatlon
ay
y- =-x.
ax
ay
or x + y - = c,
ax
, i c;
ofwhlchy = zxls oaso|utlon. Thus we|osetheso|utlony = zxofEq. (9)
uponltsdlvlslonbythefactor ,y- zx; , a|tematlve|y, wegalnthlsnewso|utlon
whenwemu|tlp|yEq. , i c;by ,y- zx; Suche|ementarya|gebralcoperatlonsto
slmp|lfy aglvendlerentla| equatlonbeforeattemptlngto so|veltarecommonln
practlce, buttheposslbl|ltyof|ossorgalnofsuchextraneousso|utlonsshou|dbe
keptlnmlnd.
Aso|utlonofadlerentla|equatlonthatcontalnsanarbltraryconstant(|lke
theconstantcln theso|utlonofExamp|es i andz;ls common|yca||edageneral
solution ofthedlerentla|equatlon, anypartlcu|archolceofaspeclncva|ueforc
yle|dsaslng|epartlcu|arso|utlonoftheequatlon.
TheargumentprecedlngExamp|eiactua||ysumcestoshowthate.epartlc-
u|arso|utlonofthedlerentla|equatlon],y;y= _,x;ln,z;satlsnestheequatlon
i, y,x; ; =o,x; +cln,:; Consequent|y,ltlsapproprlatetoca||,:;notmere|ya
genera|so|utlonof,z;, butthe genera|so|utlonof,z;
InSectlon I . 5 wesha||seethateverypartlcu|arso|utlonofa//ecnrst-order
dlerentla| equatlon ls contalned ln lts genera| so|utlon. By contrast, lt ls com-
mon for a non|lnear nrst-order dlfferentla| equatlon to have both a genera| so|u-
tlon lnvo|vlng an arbltrary constant c and one orsevera|partlcu|arso|utlons that
cannot be obtalned by se|ectlng a va|ue for c These exceptlona| so|utlons are
frequent|yca||edsingular solutions. In Prob|em )cwe askyou to show that the
genera|so|utlonofthedlerentla|equatlon,y;
:
=:yyle|dsthefaml|yofparabo-
|as y = ,x- c;
:
l||ustrated ln Flg. I . 4. 5, andto observe thattheconstant-va|ued
functlon y,x; = clsa slngu|ar so|utlon thatcannotbe obtalnednom thegenera|
so|utlonbyanycholceofthearbltraryconstantc
Flnda||so|utlonsofthedlerentla|equatlon
ay
- =:x,y- i ;
:

ax
Sol ution Separatlonofvarlab|esglves
X
FIGUR 1.4.6. General and
singular solution curves for
y' 6x(y - 1 )
2/3 .
!
i
:

ay= zx ax,
,y- i ;
, y- i ;
.

=x
:
+c,
y,x; = i+,x
:
+c;

Posltlve va|ues ofthe arbltrary constant c glve the so|utlon curves ln Flg. i : :
that |le above the |lne y = i , whereas negatlveva|ues yle|dthose that dlp be|ow
lt. The va|ue c = cglves the so|utlon y,x; = i +x

, but ova|ue ofcglves


the slngu|ar so|utlon y,x; = i that was |ost when the varlab|es were separated.
Note that the two dlerent so|utlons y,x; = i and y,x; = i + ,x
:
- i ;

both
satlsfy the lnltla| condltlon y, i ; = i Indeed, the who|e slngu|ar so|utlon curve
y = i conslstsofpolnts where the so|utlon ls not unlque and where themnctlon
),x, y;= :x,y- i ;
:

lsnotdlerentlab|e.
I . 4 Separabl e Equati ons and Appl i cati ons 37
Natural Growth and Decay
Thedlerentla|equatlon
ax
= /x ,/aconstant)
a
, i i ;
servesasamathematlca|mode|foraremarkab|ywlderangeofnatura|phenomena
anylnvo|vlngaquantltywhosetlmerateofchangelsproportlona|toltscurrentslze.
Herearesomeexamp|es.
POPULATION GROWTH: Suppose that r,; ls the number oflndlvldua|s ln a
popu|atlon(ofhumans, orlnsects,orbacterla)havlng.os.blrthanddeathrates
; and (lnblrths ordeaths per lndlvldua| per unltoftlme) . Then, durlnga short
tlmelnterva|c , approxlmate|y;i,; c blrths and i( )c deathsoccur,sothe
changeln r,;ls glvenapproxlmate|yby
and therefore
where/= ; .
ci , ; ) i()c ,
ai

ci
- = hm -= /i,
a
^-0 c
, i z;
COMPOUND INTEREST: Let~,;bethenumberofdo||arslnasavlngsaccount
at tlme (ln years) , and suppose that the lnterest ls .o-poaea.o/os/vat
an annua| lnterestrate (Note that i c annua| lnterest means that c i c ;
Contlnuouscompoundlngmeansthatdurlngashorttlmelnterva|c, theamountof
lnterestaddedtothe accountls approxlmate|yc~= ~,;c , sothat
a~ c~
- = | l m -= ~
a
/ -0 c
, i );
RADIOACTIVE DECAY: Conslderasamp|eofmaterla|thatcontalnsx,;atons
ofacertalnradloactlvelsotopeattlme Ithasbeenobservedthataconstantfractlon
ofthoseradloactlveatomswl||spontaneous|ydecay(lntoatomsofanothere|ement
orlntoanotherlsotopeofthesamee|ement)durlngeachunltoftlme. Consequent|y,
thesamp|ebehavesexact|y|lkeapopu|atlonwlthaconstantdeathrateandnoblrths.
To wrlte a mode|for x, ; , weuseEq. , i z;wlth xln p|aceof i, wlth / > oln
p|aceof, andwlth;= cWethusgetthedlerentla|equatlon
ax
-= /x.
a
Theva|ueof/dependsonthepartlcu|arradloactlvelsotope.
, i :;
The key to the method ofma/o../o a./, ls that a constant proportlon
ofthe carbon atoms ln any |lvlng creature ls made up ofthe radloactlve lsotope
| +
C ofcarbon. Thls proportlon remalnsconstantbecausethefractlonof
| +
Clnthe
atmosphere remalns a|mostconstant, and |lvlng matter ls contlnuous|y taklng up
carbonfromthealrorlsconsumlngother|lvlngmattercontalnlngthesameconstant
ratloof
| +
Catomstoordlnary
| :
Catoms. Thlssameratlopermeatesa|||lfe,because
organlcprocessesseemtomakenodlstlnctlonbetweenthetwolsotopes.
38 Chapter I Fi rst-Order Di fferenti al Equati ons
Theratloof
| +
Ctonorma|carbonremalnsconstantlntheatmospherebecacse,
a|though
| +
C lsradloactlveand s|ow|ydecays, theamountlscontlnuous|yreplea
lshedthroughtheconverslonof
| +
N(ordlnarynltrogen)to
. +
Cbycosmlcraysboa
bardlngthe upper atmosphere. Overthe |ong hlstory ofthep|anet, thls decayaad
rep|enlshmentprocesshas comelntonear|ysteadystate.
Ofcourse, when a |lvlng organlsm dles, ltceases lts metabo|lsm ofcarboa
and theprocessofradloactlvedecaybeglnstodep|ete lts
| +
C content. There is ao
rep|enlshmentofthls
| +
C, andconsequent|ytheratloof
| +
Ctonorma|carbonbegias
todrop. By measurlng thlsratlo, theamountoftlme e|apsed slnce thedeathofthe
organlsmcan beestlmated. Forsuchpurposeslt ls necessarytomeasurethedecay
constant / For
| +
C, ltlsknownthat/ o ooo i z l : lfls neasuredlnyears.
(Mattersarenotasslmp|easwehavemadethemappear. lnapp|ylngthetech
nlqueofradlocarbondatlng, extremecaremustbetakentoavoldcontanlnatlng|he
samp|ewlthorganlcmatterorevenwlth ordlnary freshalr. Inaddltlon,thecosa|c
ray|eve|sapparent|yhavenotbeenconstant, sotheratlo of
| +
C lntheatmosphe
has varled overthe past centurles. By uslnglndependentmethods ofdatlng saa
p|es, researcherslnthlsareahave compl|ed tab|esofcorrectlonfactors toenhaace
theaccuracyofthlsprocess. )
DRUG ELIMINATION: Inmany cases the amount A(t ) ofa certaindrcgi nthe
b|oodstream,measuredby theexcessoverthenatura||eve|ofthedrug,wl||decliae
atarateproportlona|tothecurrentexcessamount. Thatl s,
a~
- =-~.
a
where > cTheparameterlsca||edthe elimination constant ofthedrcg.
The Natural Growth Equation
, i ,
The prototype dlerentla| equatlon ax}a = /x wlthx, ; > o and /a constaa|
(elthernegatlve orposltlve) ls readl|y so|ved by separatlng the varlab|es and iate
gratlng.
ax=/ a ,
|a x = / + c
Then weso|veforx
Becausecl s aconstant, sol s ~ = Itl s a|soc|earthat~ =x,o)
-
Xg, sothe
partlcu|arso|utlonofEq. , l l ) wlththe lnltla|condltlonx,o)=Xg ls slnp|y
x, ; =

, i ,
Becauseofthepresenceofthenatura|exponentlalfunctionlnlts sol ctlon,the
dlerentla|equatlon
ax
~ = /x
a
, l 7)
ls oftenca||edthe exponential or natural growth equation. Flgure l . 4. 7showsa
typlca|graphofx , ; lnthecase /> o,thecase/< ols l||ustratedlnFlg. i

: s
cXump| e
I . 4 Separabl e Equati ons and Appl i cati ons 39
X
X = x

(k >

FIGURE 1.4.7. Natural growth.


X
FIGURE 1.4.8. Natural decay.
A::e:a|a,:eaa:ai|s:eaa:www.census.gov, :aewe:ia s:e:aieaia:|ea:ea:a:a
: ||ii|eae:seas |a m|a i , aaawas :aea |a::eas|a,a::ae:a:eeia|ea:zi z
:aeasaaae:seasea:aaay Assam|a,:aa:aa:a:aieaia:|ea,:ew:aa::a|s:a::
:ea:|aaes, wewaa::eaaswe::aeseaes:|eas
(a) waa:|s:aeaaaaai,:ew:a:a:e/:
(b) waa:w|ii|e:aewe:iaeaia:|eaa::aem|aaieei:aezis::ea:a:y:
(c) uewiea,w|ii|::a|e:aewe:iaeaia:|ea:e|a::ease:eaieia-:ae::|y::a:a
|a,:ae:c||ii|ea:aa:semeaeme,:aae:s|ei|eve:e|e:aemas|mamie:wa|:a:a:
iaae::aa:ev|aeaaeaa:eieeasai|es:
Sol ution (a) we measa:e:aewe:iaeaia:|ea
_,
|a||ii|easaaameasa:e:|m:|ay:a:s
we:a|e
,
-
c:e:e::eseaa:e,m|a) i , sei,
-
:1aeia:::aa:i|s|a:::as|a
|yzi z,ccc,e:c ccczi z ||ii|ea,e:sease:aaya::|me
,
-
cmeaas:aa:
i ,c)= ,c ccczi z) ,: z)

c c:
||ii|eae:yea: t:em:aeaa:a:ai,:ew:aeaa:|ea
_
,
-
/i w|:a
,
-
cw:aew
e|:a|a
/
-
i ,c)

c c:

c ci z
i,c) :
1aas:aewe:iaeaia:|eawas,:ew|a,a::ae:a:eeia|ea:i zaaaaaiiy|ai
1a|svaiaeei/,|ves:aewe:iaeaia:|eaiaa::|ea
_,
-

(b) w|:a
,
-
i wee|:a|a:ae:ea|::|ea
i, i )
-


i i s,||ii|ea)
ie::aewe:iaeaia:|ea|am|azcc,se:aeeaia:|eaw|iiaimes:aav:aea|i:a
|a:ae]as:eve:aaaii:ea:a:ys|a:ei )
(c) 1aewe:iaeaia:|easaeaia:ea:a:c||ii|eawaea
:c=

aaa:aas|a:aeyea:z i
i ai c
:aa:|s, waea
,
=
c ci z

i s,

40 Chapter I Fi rst-Order Di fferenti al Equati ons


cXump| e
Dotet A::aaiiy,:ae:a:eei,:ew:|ei:aewe:iaeaia:|ea|sexe::ea:a
siewsemewaa:aa:|a,:aeaex:|aii:ea:a:y,aaa:|e|es::a::ea::ea|::|eaie::a:
zcceaia:|ea |s eaiy l ||ii|ea A s|miema:aema:|:ai meaei :aaae:o:
ese::ea:em|::e::e:|seiy:ae:emies|:yei:|e:eaiwe:ia
1aeae:ay:eas:aa:eia:aa|ea::|ve|se:ee|seuease:|aea|a:e:mseiaa
e:|e:em|:|:ai:eas:aa:,:|e/.(/,eei:ae|se:ee,|e:aase:a|sa:ame:e:|sma::
:eavea|ea:1ae half-life reia:aa|ea::|ve|se:ee|s:ae:|me:ea|:eaie:/.(ai
|::eae:ay 1eaaa:ae:eia:|eas|||e:weea/ aaa r, wese: = r aaaV = N
|a:|eeaa:|eaV,;= x,e

,se:aa:V, x,e

waeaweseiveie:r, weaaa
:aa:
ia z
r =

, l s)
le:esamie,:|eaaiii|ieei

|s r ,ia z)},c cccl zl :) , a:ex|ma:eiyco


yea:s
Ase:|meaei:aa::eaiieaaaa:s:eaeaea,e:ansea::e:ea:a|a:asma:a
asasamieei:esea:aay:aa::eaieieaaimass waa:|s:aea,eei:aesamie:
Sol uti on we :aie = cas:ae:|meei:aeaea:aei:ae::ee i:em wa|:a:ae s:eaeaea:
:aa::eaiwasmaaeaaaV,as:aeaam|e:eia:ems:aa::aes:eaeaea,esami:
:ea:a|aea:aea wea:e ,|vea :aa:V = ,c :) V, aew, seweseive:aeeaa:|aa
,c :) V,= x,e
-

w|:a:aevaiae/= c cccl zl : 1|asweaaa:aa:


cXump| e
ia,c :)
=

c cccl zl :
scc,yea:s)
1|as:aesamie|sa|ea:sccyea:seia i i | : |asaay:eaae::|eaw|:|:ae|a|iae:sai
s:eae|ea,e,ea::ema:a:|eassa,,es::aa::a|se|se:va:e:y,meaamea:,e::emi:-
w||:|eve:|:may|e-aa:esi:eml sccB. C. e:ea:i|e:
Cooling and Heating
A::e:a|a,:eNew:ea siaw ei:eei|a, , ,)eise::|ea l l ), :ae :|me:a::ai
:aaa,eei:ae:eme:a:a:e, ; eia|eay|mme:sea|aamea|amei:eas:aa:::m
e:a:a:e~|s:ee::|eaai:e:aea|iie:ea:e~ 1aa:|s,
a
- = /,~

) ,
a
, l )
wae:e/|saes|:|ve:eas:aa:1a|s|saa|as:aa:eei:aei|aea:a:s:e:ae:a|ne::a:|ai
eaa:|eaw|:a:eas:aa::eeu:|ea:s
ax
a
= .x+/ ,zo)
i:|a:iaaes:aeeseaea:|aieaa:|eaasase:|ai:ase ,/ = c)aaa|s aiseeasy:a
seive|ysea:a:|eaeiva:|a|ies
A:i|:eas:, |a|:|aiiya:cl, |s ia:ea|aa levea a: cc e v Aue:
m|aa:es|:|sieaaa:aa::ae:eme:a:a:e,;ei:|e:eas:|s l z t waeaw|ii:a:
:eas:|el cl,mea|am:a:e) :
I . 4 Separabl e Equati ons and Appl i cati ons 41
Sol ution we:aie:|me|am|aa:es,w|:a=c:e::eseaa|a,:e ccl. N. we aiseassam:
,semewaa:aa:eai| s:|:aiiy):aa:a:aay|as:aa::ae:eme:a:a:e, ) ei:a::aas:|s
aa|ie:m:a:ea,aea: weaave , ) < ~ = , ,c) = c, aaa,) i z
uea:e
..
=/,- ) ,

i
..= .
-

- ia,- ) =+c,
- =s,

New ,c) = c|mi|es:aa:s = z, se= - z,

w:aise|aaw
:aa: = i z waea= sa|s:|:a:|eaei:aesevaiaes|a:ae:e:ea|aeaa:|aa
y|eias
=-ia y c cc
uea:eweaaaiiyseive:aeeaa:|ea
i c=- z,

'

ie:= -ia,zz}z)[},c cc) i c ,m|a), :ae:e:ai :ee||a,:|me:ea|::a


se:aase:ae:eas:wasia:ea|a:aeeveaa: ccl. N. , |:saeaia|e:emeveaa:a|aa:
: :l. N.
TorriceIIi' s Law
saese:aa:awa:e::aa|aasaaeiew|:aa:ea.a:|:s|e::em,i:emwa|:awa:::|s
iea||a, Deae:e|yy, ) :aeae:|eiwa:e:|a:ae:aa|a::|me aaa|y .:a:
veiameeiwa:e:|a:|e:aa|:aeai:|siaas||ie-aaa::ae,aaae:|aeai:aaa|:|eas-
:aa::aeveie:|:yeiwa:e:es|:|a,:a:ea,|:aeaeie|s
u =,z,y, ,zi )
wa|:a| s :aeveie:|:yaa:eeiwa:e:weaiaa:a|:e|aiaii|a,i:eeiyi:em:aesa:ia::
ei:aewa:e::e:ae|eie,seer:e|iemeise::|eai z) Oae:aaae:|ve:a|sia:maia
|e,|aa|a,w|:a:aeassam:|ea:aa::aesamei:ae||ae:|:aaae:ea:|ai:ae:,yai:a:
sys:em:ema|as:eas:aa:Uaae::eai:eaa|:|eas,:a||a,|a:ea::eaa::ae:eas::|::|aa
eiawa:e:]e:i:emaae:|a:e,u =.,z,y,wae:e.|saaem|:|:ai:eas:aa:|e:w::a
0 aaai ,asaaiiya|ea:c :ie:asmaii:ea:|aaeass::eameiwa:e:) te:s|mi|:|:y
we:a|e.= i |a:aeieiiew|a,a|s:ass|ea
Asa:easeaea:eei ,zi ),weaave
ea|vaiea:iy,
..
=-.u =-.,z,y,
.
..
= wae:e =.,
.
,zza)
,zz|)
1a|s|sas:a:emea:ei1e::|:eii| siawie:aa:a|a|a,:aa|Le:~,y;aeae:::aeae:|
zea:ai::essse::|eaaia:eaei:ae:aa|a:ae|,a:y1aea,ai|ea:ea:a|aae:|zaa:ai
42 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| ed
s|lce of water at helght, wlth area ~,,; and thlckness a,, the lntegra| ca|cu|us
methodofcrosssectlonsglves
r,y;=
~,,;a

The fundamenta| theorem of ca|cu|us therefore lmp|les that ar }ay


hencethat
ar

ar

ay
= ~, ;
ay

a ay a
y
a
FromEqs. ,zz;and,z;wenna||yobtaln
ay

~,y; - = -, z,y

/
$
a
ana|tematlveformofTorrlce||l

s|aw.
~,y;and
,z;
,z:;

Ahemlspherlca|bow|has topradlus:uandattlme cls fu||ofwater. Atthat


momentaclrcu|arho|ewlthdlameterlln. lsopenedlnthebottomofthetank. How
|ongwl||lttakefora||thewatertodralnfromthetank!
Sol ution Fromtherlghttrlang|elnFlg. l : , weseethat
FIGURE 1.4.9. Draining a
hemispherical tank.
~,y;= -

= - l : - ,:- y;

] -,sy- ,;
Wlth,= zftJs
:
,Eq.,z:;becomes
Nowy,o;= :, so
:
ay

.
)
:
-,sy-

;
a
= -- ,z zy,
,,sy
,

- ,

;ay= - ,a ,
_
y

:
.
,
y

.
+c

.
,

448

The tank ls empty wheny c,thus when


= :z
.

8
zi c(s) ,
that l s, about 35 mln c s. So lt takes s|lght|y |ess than 36 mln for the tank to
draln.
Problems
........,.....,.
..., ..,......I ....I.
.........,.
.
1.
..
.= 0
.

3.
..
= sm.
5. 2.
.

..
.
7. - = (4.)
J

..
9. (l - .

)
.
=
..
11. = .

.
13.

= (
4
1 ) cos .
..
15.
.
=
.

.. .


17. = . .
side. )
.
2. .

0
..
.
4. ( l . .
..
.
6.
..
= 3
.
.
8.
..
= .sec
10. ( l .

.
= ( l )

..
12. = .

1 )
14.
. 1 + .
.. 1 ,
16. .

l ) (tan = .
.....Factor the right-hand
18. .

= 1 .

...,,............,.
.I9....2.
.
19.
..
= =
.
20.
..
= 3.

1 ) , = 1
. .
21. = (5) = 2
..
1
'
.
22.
..
= ..

= -3
.
23. 1 = = 1
..
.
24. (tan .
..
=
_i
=

.
25. . 2.

, = 1
..
.
26.
..
= 2.

3.

, = -1
27. =

= 0
.
28. 2. = cos
2
.= .
..
29. (a) Find a general solution of the diferential equation
...=

(b) Find a singular solution that is not in


cluded in the general solution. (c) Inspect a sketch of typi
cal solution curves to determine the points .for which
the initial value problem =

, .has a unique
solution.
I . 4 Separabl e Equati ons and Appl i cati ons 43
30. Solve the differential equation ...

= .to verify the


general solution curves and singular solution curve that
are illustrated in Fig. 1 . 4. 5. Then determine the points
.in the plane for which the initial value problem
( )

. .= has (a) no solution, (b) infnitely


many solutions that are defned for all .(c) on some
neighborhood of the point .= .only fnitely many solu
tions.
31. Discuss the diference between the diferential equations
...

= .and ...= 2,. Do they have the


same solution curves? Why or why not? Determine the
points .in the plane for which the initial value prob
lem = 2,, .= has (a) no solution, (b) a unique
solution, (c) infnitely many solutions.
32. Find a general solution and any singular solutions of the
diferential equation .......= Deter
mine the points .in the plane for which the initial
value problem = .= has (a) no solu
tion, (b) a unique solution, (c) infnitely many solutions.
33. (Population growth) A certain city had a population of
25000 in 1 960 and a population of 30000 in 1 970. Assume
that its population will continue to grow exponentially at a
constant rate. What population can its city planners expect
in the year 2000?
34. (Population growth) In a certain culture of bacteria, the
number of bacteria increased sixfold in 1 0 h. How long
did it take for the population to double?
35. (Radiocarbon dating) Carbon extracted from an ancient
skull contained only one-sixth as much 1
4
C as carbon ex
tracted from present-day bone. How old is the skull?
36. (Radiocarbon dating) Carbon taken from a purported relic
of the time of Christ contained 4. 6 X 1 01 0 atoms of 1 4C
per gram. Carbon extracted from a present-day specimen
of the same substance contained 5. 0 X 1 01 0 atoms of 1 4C
per gram. Compute the approximate age of the relic. What
is your opinion as to its authenticity?
37. (Continuously compounded interest) Upon the birth of
their frst child, a couple deposited $5000 in an account
that pays 8% interest compounded continuously. The in
terest payments are allowed to accumulate. How much
will the account contain on the child' s eighteenth birth
day?
38. (Continuously compounded interest) Suppose that you
discover in your attic an overdue library book on which
your grandfather owed a fne of 30 cents 1 00 years ago. If
an overdue fne grows exponentially at a 5% annual rate
compounded continuously, how much would you have to
pay if you retured the book today?
39. (Drug elimination) Suppose that sodium pentobarbital is
used to anesthetize a dog. The dog is anesthetized when
its bloodstream contains at least 45 milligrams (mg) of
sodium pentobarbitol per kilogram of the dog' s body
44 Chapter I Fi rst-Order Di fferenti al Equati ons
weight. Suppose also that sodium pentobarbitol is elim
inated exponentially from the dog' s bloodstream, with a
half-life of 5 h. What single dose should be administered
in order to anesthetize a 50-kg dog for 1 h?
40. The half-life of radioactive cobalt is 5. 27 years. Suppose
that a nuclear accident has lef the level of cobalt radia
tion in a certain region at 1 00 times the level acceptable
for human habitation. How long will it be until the region
is again habitable? (Ignore the probable presence of other
radioactive isotopes. )
41. Suppose that a mineral body formed in an ancient
cataclysm-perhaps the formation of the earth itself
originally contained the uranium isotope
23
8U (which has
a half-life of 4. 5 1 X 1 0
9
years) but no lead, the end product
of the radioactive decay of
23
8U. If today the ratio of
23
8U
atoms to lead atoms in the mineral body is 0. 9, when did
the cataclysm occur?
42. A certain moon rock was found to contain equal numbers
of potassium and argon atoms. Assume that all the argon
is the result of radioactive decay of potassium (its half-life
is about 1 . 28 X 1 0
9
years) and that one of every nine potas
sium atom disintegrations yields an argon atom. What is
the age of the rock, measured from the time it contained
only potassium?
43. A pitcher of buttermilk initially at 25 C is to be cooled
by setting it on the front porch, where the temperature is
0 C. Suppose that the temperature of the buttermilk has
dropped to 1 5 C after 20 min. When will it be at 5 C?
44. When sugar is dissolved in water, the amount .that re
mains undissolved afer minutes satisfes 'the differential
equation ...= ..(.~ 0). If 25% of the sugar dis
solves after 1 min, how long does it take for half of the
sugar to dissolve?
45. The intensity of light at a depth of .meters below
the surface of a lake sati sfes the diferential equation
...= ( -1 . 4) (a) At what depth is the intensity half
the intensity
0
at the surface (where .= O)? (b) What
is the intensity at a depth of 10 m (as a fraction of lo)?
(c) At what depth wi l l the intensity be 1 % of that at the
surface?
46. The barometric pressure ,(in inches of mercury) at an
altitude .miles above sea level satisfes the initial value
problem .,..= ( -0. 2) ,,= 29. 92. (a) Calculate
the barometric pressure at 1 0,000 ft and again at 30, 000
ft. (b) Without prior conditioning, few people can sur
vive when the pressure drops to less than 1 5 in. of mer
cury. How high is that?
47. A certain piece of dubious information about phenylethy
lamine in the drinking water began to spread one day in a
city with a population of 1 00,000. Within a week, 1 0,000
people had heard this rumor. Assume that the rate of in
crease of the number who have heard the rumor is propor
tional to the number who have not yet heard it. How long
will it be until half the population of the city has heard the
rumor?
48. According to one cosmological theory, there were equal
amounts of the two uranium isotopes
235
U and
23
8U at the
creation of the universe in the "big bang." At present there
are 1 37. 7 atoms of
23
8 U for each atom of
235
U. Using the
half-lives 4. 5 1 X 1 0
9
years for
23
8U and 7. 1 0 X 1 08 years
for
235
U, calculate the age of the universe.
49. A cake is removed from an oven at 21 0 F and left to cool
at room temperature, which is 70 F. Afer 30 min the
temperature of the cake is 1 40 F. When will it be 1 00 F?
50. The amount .of atmospheric pollutants in a certain
mountain valley grows naturally and is tripling every 7. 5
years.
(a) If the initial amount is 1 0 pu (pollutant units), write
a formula for .giving the amount (in pu) present
afer years.
(b) What will be the amount (in pu) of pollutants present
in the valley atmosphere afer 5 years?
(c) If it will be dangerous to stay in the valley when the
amount of pollutants reaches 1 00 pu, how long will
this take?
51. An accident at a nuclear power plant has lef the surround
ing area polluted with radioactive material that decays nat
urally. The initial amount of radioactive material present
is 15 su (safe units), and 5 months later it is still 10 suo
(a) Write a formula giving the amount .of radioactive
material (in su) remaining after months.
(b) What amount of radioactive material will remain after
8 months?
(c) How long-total number of months or fraction
thereof-will it be until .= 1 su, so it is safe for
people to return to the area?
52. There are now about 3300 diferent human "language fam
ilies" in the whole world. Assume that all these are de
rived from a single original language, and that a language
family develops into 1 . 5 language families every 6 thou
sand years. About how long ago was the single original
human language spoken?
53. Thousands of years ago ancestors of the Native Americans
crossed the Bering Strait from Asia and entered the west
ern hemisphere. Since then, they have fanned out across
North and South America. The single language that the
original Native Americans spoke has since split into many
Indian "language families." Assume (as in Problem 52)
that the number of these language families has been mul
tiplied by 1 . 5 every 6000 years. There are now 1 50 Native
American language families in the western hemi sphere.
About when did the ancestors of today' s Native Ameri
cans arrive?
54. A tank is shaped like a vertical cylinder; it initially con
tains water to a depth of 9 f, and a bottom plug is removed
at time = 0 (hours). Afer 1 h the depth of the water has
dropped to 4 ft. How long does it take for all the water to
drain from the tank?
55. Suppose that the tank of Problem 48 has a radius of 3 f
and that its bottom hole is circular with radius 1 in. How
long will it take the water (initially 9 f deep) to drain com
pletely?
56. At time = 0 the bottom plug (at the vertex) of a full con
ical water tank 16 f high is removed. After 1 h the water
in the tank is 9 f deep. When will the tank be empty?
57. Suppose that a cylindrical tank initially containing .gal
lons of water drains (through a bottom hole) in .minutes.
Use Torricelli' s law to show that the volume of water in
the tank after .minutes is .= .[ 1 - (.) ]

.
58. A water tank has the shape obtained by revolving the curve
y = .

around the y-axis. A plug at the bottom is re


moved at 1 2 noon, when the depth of water in the tank is
1 2 ft. At 1 . V. the depth of the water is 6 ft. When will
the tank be empty?
59. A water tank has the shape obtained by revolving the
parabola .

= by around the y-axis. The water depth is


4 f at 1 2 noon, when a circular plug in the bottom of the
tank is removed. At 1 . V. the depth of the water is 1 ft.
(a) Find the depth y(t) of water remaining afer hours.
(b) When will the tank be empty? (c) If the initial radius
of the top surface of the water is 2 f, what is the radius of
the circular hole in the bottom?
60. A cylindrical tank with length 5 f and radius 3 ft is sit
uated with its axis horizontal . If a circular bottom hole
with a radius of 1 in. is opened and the tank is initially
half full of xylene, how long will it take for the liquid to
drain completely?
61. A spherical tank of radius 4 f is full of gasoline when a
circular bottom hole with radius 1 in. is opened. How long
will be required for all the gasoline to drain from the tank?
62. Suppose that an initially full hemi spherical water tank of
radius 1 m has its fat side as its bottom. It has a bottom
hole of radius 1 cm. If this bottom hole is opened at 1 . V. ,
when will the tank be empty?
63. Consider the initially full hemispherical water tank of Ex
ample 8, except that the radius of its circular bottom hole
is now unknown. At 1 . V. the bottom hole is opened and
at 1 : 30 . V. the depth of water in the tank is 2 ft. (a) Use
Torricelli ' s law in the form .. .(0. 6)

J2gy
(taking constriction into account) to determine when the
tank will be empty. (b) What is the radius of the bottom
hole?
64. (The ,...or water clock) A 1 2-h water clock is to
be designed with the dimensions shown in Fig. 1 . 4. 1 0,
shaped like the surface obtained by revolving the curve
y = . around the y-axis. What should be this curve,
and what should be the radius of the circular bottom hole,
in order that the water level will fall at the ....rate of
4 inches per hour (in. /h)?
I . 4 Separabl e Equati ons and Appl i cati ons 45
4 f
) -]()
)
ot
) y())
)
FIGUR 1.4.10. The clepsydra.
65. Just before midday the body of an apparent homicide vic
tim is found in a room that is kept at a constant tempera
ture of 70 F. At 1 2 noon the temperature of the body is
80 F and at 1 . V. it is 75 F. Assume that the temperature
of the body at the time of death was 98. 6 F and that it has
cooled in accord with Newton' s law. What was the time
of death?
66. Early one morning it began to snow at a constant rate. At
7 A. M. a snowplow set off to clear a road. By 8 A. M. it
had traveled 2 miles, but it took two more hours (until
1 0 A. M. ) for the snowplow to go an additional 2 miles.
(a) Let = 0 when it began to snow and let A denote the
di stance traveled by the snowplow at time Assuming
that the snowplow clears snow from the road at a constant
rate (in cubic feet per hour, say), show that
.. 1
.
.
where .is a constant. (b) What time did it start snowing?
... 6 A. M. )
67. A snowplow sets of at 7 A. V. as i n Problem 66. Suppose
now that by 8 A. M. it had traveled 4 miles and that by
9 A. M. it had moved an additional 3 miles. What time did
it start snowing? This is a more diffcult snowplow prob
lem because now a transcendental equation must be solved
numerically to fnd the value of ...4: 27 A. V. )
68. Figure 1 . 4. 1 1 shows a bead sliding down a frictionless
wire from point to point Q. The .....,
asks what shape the wire should be in order to min
imize the bead' s time of descent from to Q. In June
of 1 696, John Bernoulli proposed this problem as a pub
lic challenge, with a 6-month deadline (later extended to
Easter 1 697 at George Leibniz' s request). Isaac Newton,
then retired from academic life and serving as Warden
of the Mint in London, received Bernoulli 's challenge on
January 29, 1 697. The very next day he communicated
his own solution-the curve of minimal descent time is an
46 Chapter I Fi rst-Order Di fferenti al Equati ons
arc of an inverted cycloid-to the Royal Society of Lon
don. For a modem derivation of this result, suppose the
bead starts from rest at the origin and let y = y(.) be
the equation of the desired curve in a coordinate system
with the y- axis pointing downward. Then a mechanical
analogue of Snel l ' s law in optics implies that
sin a
= constant,

(i)
where a denotes the angle of defection (from the verti
cal) of the tangent line to the curve-so cot a = y' .
(why?)-and = ,.is the bead' s velocity when it has
descended a di stance y vertically (from =

.= -PE) .
F

FIGUR 1. 4. 11. A bead sliding down a


wire-the brachistochrone problem.
(a) First derive from Eq. (i) the differential equation
.

..
-
y
-
where .is an appropriate positive constant.
(ii)
(b) Substitute y = 2. sin

.= 4. sin cos .in (ii)


to derive the solution
.= .- sin y = . - cos (iii)
for which = Y = 0 when .= O. Finally, the sub
stitution of 0 = .in (iii) yields the standard para
metric equations .= .(0 - sin 0) , y = . - cos 0)
Linear First-Order E9ation
69.
of the cycloid that is generated by a point on the rim
of a circular wheel of radius .as it rolls along the .
axi s. [See Example 5 i n Section 9. 4 of Edwards and
Penney, ...............7th edition
(Upper Saddle River, NJ: Prentice Hall, 2008). ]
Suppose a uniform fexible cable is suspended between
two points ..at equal heights located symmetri
cally on either side of the x- axis (Fig. 1 . 4. 1 2). Principles
of physics can be used to show that the shape y = . of
the hanging cable satisfes the diferential equation
.

1
..

where the constant .= ..is the ratio of the cable' s


tension .at its lowest point .= 0 (where y' (0) = 0
) and its (constant) linear density . If we substitute
= ..........= .

..

in this second
order diferential equation, we get the frst-order equation
.

. = 1

..
Solve this diferential equation for . = .
sinh(x/a) . Then integrate t o get the shape function
. = .cosh
of the hanging cable. This curve is called a ...from
the Latin word for ...
"
|| ||
"c
)
FIGUR 1.4. 12. The catenary.
InSectlon l . :wesawhowto so|ve a separab|e dlerentla| equatlon bylntegratlng
.)emu|tlp|ylng both sldes by an approprlate factor. For lnstance, to so|ve the
equatlon
ay
zxy ,y~ c),
ax
wemu|tlp|ybothsldesbythefactor i}ytoget
i ay
zx, thatl s,
)
(lny;
)
x
:
)
) ax
, l )
,z;
Becauseeachsldeoftheequatlonln,z;lsrecognlzab|easaae/../.e(wlthrespect
tothelndependentvarlab|ex; . a||thatremalnsaretwo slmp|e lntegratlons, whlch
I . Li near Fi rst-Order Equati ons 47
yle|dlny x

c Forthlsreason, thefunctlon,,y; i}ylsca||edan/e,-.


/,]..ofor the orlglna|equatlonln , i ) . An integrating factor foradlerentla|
equatlonlsafunctlon,,x.y;suchthatthemu|tlp|lcatlonofeachsldeofthedler-
entla|equatlonby,,x.y;yle|dsanequatlonlnwhlcheach sldelsrecognlzab|eas
aderlvatlve.
Wlththealdoftheapproprlatelntegratlngfactor,therelsastandardtechnlque
forso|vlngthelinear frst-order equation
ay
- +i,x; y _,x;
ax
(3)
onanlnterva|onwhlchthecoemclentfunctlonsi,x; and_,x; arecontlnuous. We
mu|tlp|yeachsldelnEq.(3)bythelntegratlngfactor
Theresu|tls
Because
,,x; e
!

e
!

ay

i,x; e
!

_,x; e
!

ax
n

i,x; ax|
i,x; .
,:)
(5)
the|eft-handsldelsthederlvatlveofthepa.y,x;

e
!

.soEq(5)lsequlv-
a|entto
n

y,x;

e
!

_,x;e
!

Integratlonofboth sldesofthlsequatlonglves
y,x; e
!

_,x; e
!


,ax

c
Flna||y,so|vlngfory. weobtalnthegenera|so|utlonofthe|lnearnrst-orderequatlcn
ln(3).
y,x; e
- !

_,x; e
!

,ax

(6)
Thlsformu|ashou|d not bememorlzed Inaspeclncprob|em ltgenera||y ls
slmp|erto use the-e/oaby whlch we deve|oped the formu|a That ls, ln crder
to so|vean equatlonthatcanbewrltten lntheform lnEq (3) wlth thecoemclent
functlons i,x; and _,x; dlsp|ayedexp|lclt|y, you shou|d attempttocarryoutthe
fo||owlngsteps
METHOD: SOLUTIONOFFIRST-ORDEREQUATIONS
1 . Beglnbyca|cu|atlngthelntegratlngfactor,,x;

e
!


2. Thenmu|tlp|ybothsldesof thedlerentla|equatlonby,,x;
J. Next, recognlzethe |e-hand sldeoftheresu|tlngequatlonas the derlvatlve
ofaproduct.
n

,,x; y,x; }

,,x; _,x;
48 Chapter I First-Order Di fferenti al Equati ons
cXump| e 1
4. Flna||y, lntegratethlsequatlon,
,,x; y,x; ,,,x; _,x; ax+c,
thenso|veforytoobtalnthegenera|so|utlonoftheorlglna|dlerentla|equa-
tlon.
Hemark 1 : Glven an lnltla| condltlon y,x,; y,. you can (as usua|)
substltutex x,andy
y,lntothegenera|so|utlonandso|vefortheva|ueofc
yle|dlngthepartlcu|arso|utlonthatsatlsnesthlslnltla|condltlon
Hemark Zt Youneednotsupp|yexp|lclt|yaconstantoflntegratlonwhen
younndthelntegratlngfactor,,x; Forlfwerep|ace
lnEq ,:), theresu|tls
,i,x; ax wlth ,i,x; ax+K
,,x; c
,
-

c
,
c

Buttheconstantfactorc
,
does not aectmaterla||ytheresu|tofmu|tlp|ylngboth
sldes ofthedlerentla| equatlon ln(3)by ,,x; . sowemlghtas we|| take K o
You may therefore choose for ,i,x; ax .yconvenlent antlderlvatlve of i,x; ,
wlthoutbotherlngt oaddaconstantoflntegratlon
So|vethelnltla|va|ueprob|em
ay
. .

ax
- y c g y,o; -i .
Sol uti on Herewehave i,x; = -i and _,x;
.
_
,
c

,sothelntegratlngfactorls
,,x; c

.
c

Mu|tlp|lcatlonofboth sldesoftheglvenequatlonbyc

yle|ds
whlchwerecognlzeas

ay

. .

c - c y c ,
ax
a
~
c

L
_
,
c

ax
Hencelntegratlonwlthrespecttoxglves
c

y ,
_
c

_
x
.

+c,
andmu|tlp|lcatlonbyc

glvesthegenera|so|utlon
y,x; cc

,)
,s)
Substltutlon ofx cand y -i nowglves c

.
;
. sothedeslredptlcu|ar
so|utlonls

' -1

-4
y* exp(-xI3)
-1 0 4 5
x
FIGURE 1. 5. 1. Slope feld and
solution curves for
y
'
= y + e-
x
/3
cXump| eZ
I . Li near First-Order Equati ons 49
Hemarkt Flgure i i showsas|opene|dandtyplca| so|utloncurvesfor
Eq. (7), lnc|udlngtheonepasslngthroughthepolnt,c. -i ; Notethatsomeso|u-
tlonsgrowrapld|ylntheposltlvedlrectlonasxlncreases,whl|eothersgrowrapld|y
ln thenegatlvedlrectlon. The behavlorofaglvenso|utloncurve ls determlnedby
ltsl nltla| condltlon, (0) )g. The two typesofbehavlorare separatedbythepar-
tlcu|arso|utlony ,x;
.
e
-

forwhlchc clnEq. ,s), so)g -forthe


so|utlon curve thatls dashedln Flg. i i If)g > -, then c > clnEq. ,s),so
theteme
eventua||ydomlnatesthebehavlorofy,x; , andhencey,x;-- + as
x -- +c. But lf)g < -, then c < c, so both tems ln y,x;are negatlveand
thereforey,x; -- -casx-- +c. Thusthelnltla|condltlon)g -ls.//../
ln the sensethat so|utlons that start above - onthe ,-axlsgrow lntheposltlve
dlrectlon,whl|eso|utlonsthatstart|owerthan-growlnthenegatlvedlrectlonas
x-- +c. Thelnterpretatlonofamathematlca|mode|oftenhlngesonnndlngsuch
acrltlca|condltlonthatseparates oneklndofbehavlorofaso|utlonfromadlfferent
klndofbehavlor.
Flnd agenera|so|utlonof
ay
,x

+ i ; - + xy= :x
ax
,)
Sol uti on
Aerdlvlslonofbothsldesoftheequatlonbyx
:
+ i, werecognlzetheresu|t
X
FIGURE 1.5.2. Slope feld and
solution curves for the differential
equation in Eq. (9).
ay x :x
-+ ,

ax x

+ i

x
:
+ i
asanrst-order|lnearequatlonwlth i,x; x},x

+ i ; and _,x;= :x},x

+ i ;
Mu|tlp|lcatlonby
,,x; exp
x

i
ax,
expin,x

+ i ;) ,x

+ i ;

yle|ds
andthus
Integratlonthenyle|ds
Mu|tlp|lcatlonofbothsldesby,x

+ i;
-

glvesthegenera|so|utlon
y,x; z+c,x

+ i ;
-

( ! 0)

Hemarkt Flgure i zshows a s|opene|dand typlca| so|utloncurvesfor


Eq. ,) Note that, as x -- +c, a|| other so|utlon curves approach the constant
so|utlon curve , (x; = z that corresponds to c cln Eq. , i c; Thls constant
50 Chapter I Fi rst-Order Di fferenti al Equati ons
so|utloncanbedescrlbedasane,/////-so//oofthedlerentla|equatlon,be-
causey,o; 2 lmp|les thaty,x; 2 for a||x(andthustheva|ueofthe so|utlon
remalnsforeverwhere ltstarts) . Moregenera||y, theword'equl|lbrlumconnotes
'unchanglng, so by an equl|lbrlum so|utlon of a dlfferentla| equatlon ls meant a
constantso|utlony,x; = ..forwhlchltfo||owsthaty ,x; = 0 Note thatsubstl-
tutlonofy clnthedlfferentla|equatlon,)yle|dsxy 6x, soltfo||owsthat
y 2lfx /0. Henceweseethaty,x; = 2lstheon|yequl|lbrlumso|utlonofthls
dlerentla|equatlon,asseemsvlsua||yobvlouslnFlg I 5 2

A Closer Look at the Method


Theprecedlngderlvatlonoftheso|utlonlnEq (6)ofthe|lnearnrst-orderequatlon
y+iy _bearsc|oserexamlnatlon Supposethatthecoefnclentfunctlonst,x;
and _,x; are contlnuous on the (posslb|y unbounded) open lnterva| l. Then the
antlderlvatlves
i,x;ax and

_,x; e
!

ax
exlstonl . OurderlvatlonofEq (6) shows that,y y,x; lsaso|utlonofEq.(3)
on l, /ey,x; lsglvenbytheformu|alnEq (6)for somecholceofthe constant
cConverse|y,youmayverlfybydlrectsubstltutlon(Prob|em3 I ) thatthefunctlon
y,x; glven lnEq (6) satlsnesEq (3) Flna||y, glvenapolntx,ofl andany num-
bery,, there lsas prevlous|ynoteda unlque va|ueofc such that y,x,;
-
y,
Consequent|y,we have provedthefo||owlngexl stence-unlquenesstheorem
THEOREM 1 The Linear First-Order Equation
Ifthefunctlonsi,x;and Q,x;arecontlnuousontheopenlnterva| l contalnlng
thepolntx,,thenthelnltla|va|ueprob|em
ay
ax
+ i,x; y _,x; , y,x,;
y, ( I I )
hasaunlqueso|utlony,x; onl , glvenbythefomu|al nEq(6)wlthanappro-
prlateva|ueofc
Remark 1: Theorem I glvesaso|utlononthee/elnterva|l fora//e.
dlerentla|equatlon, ln contrastwlth Theorem I ofSectlon I 3, whlch guarantees
on|yaso|utlononaposslb|ysma||erlnterva|
Remark 2: Theorem I te||susthateveryso|utlonof Eq(3)lslnc|udedln
thegenera| so|utlonglvenlnEq (6) Thusa//e.nrst-orderdlfferentla|equatlon
hasoslngu|arso|utlons
Remark 3: Theapproprlateva|ueoftheconstantclnEq(6)asneeded
to so|ve the lnltla| va|ueprob|emlnEq ( I I )canbe se|ected 'automatlca||yby
wrltlng
,,x; expi,; a , .
y,x;
I
y

+,

,, ; _, ; a

,,x;
,
( I 2)
cXump| e
I . Li near Fi rst-Order Equati ons 51
Thelndlcated |lmltsx, andx effecta cholceoflndennlte lntegra|sl nEq (6) that
guaranteeslnadvancethatj(x,) = I andthat,(x,) = ), (as you canverlfydlrect|y
bysubstltutlngx =x,lnEqs ( I 2)

So|vethelnltla| va|ueprob|em
:
a,
.
( I ) x - +x,= smx, , = ),
ax
( I 3)
Sol ution Dlvlslonbyx
:
glvesthe|lnearnrst-orderequatlon
5
2
- |
- 2

5
| I -5)
o | | z0
x
l 1.5.3. Typical solution
UYCb defned by Eq. ( 1 5) .
ay I sln x
- + -, =
ax x x
:
wlthP(x) = IJx and Q(x) = (sln x)Jx
:
Wlthx, = I thelntegratlngfactor ln( I 2)
l s
j(x)
-
expa,= exp(|n x) =x,
sothedeslredpartlcu|arso|utlonl s glvenby
I

sln t
,(x) =

),+

a ( I 4)
InaccordwlthTheorem I , thlsso|utlonlsdennedonthewho|eposltlvex-axls
Comment: Ingenera|, anlntegra|suchastheonelnEq ( I 4) wou|d (for
glvenx) needtobeapproxlmatednumerlca||yuslngSlmpson' sru|e,forlnstance
to nndthe va|ue ,(x) ofthe so|utlonatx Inthlscase, however, wehavethe slne
lntegra|functlon
sln t
Si (x) = a.
,
whlch appears wlth sumclent frequency ln app|lcatlons that lts va|ues have been
tabu|ated A good set of tab|es of specla| functlons ls Abramowltz and Stegun,
u.a/oo/o)M./e-./../i./os(NewYork. Dover, I 965) Thenthepartlcu-
|arso|utlonlnEq ( I 4)reducesto
I
sln

sln I ,
.
,(x) = - ), + a - a = - )
a
+ Si (x) - Si ( | ) ]
x

, x
( I 5)
Theslnelntegra|functlonlsaval|ab|eln mostsclentlnccomputlngsystemsandcan
beused to p|ottyplca| so|utlon curves denned by Eq ( I 5) Flgure I 5 3 shows a
se|ectlonofso|utloncurveswlthlnltla|va|ues,( I ) = ),ranglngfrom),
-
-3 to
), = 3 Itappears thatoneach so|utloncurve, ,(x) casx +, andthlsls
lnfacttruebecausetheslnelntegra|functlonlsbounded.
Intheseque|wewl||seethatltlstheexceptlonratherthantheru|ewhena
so|utlonofadlfferentla|equatloncanbeexpressedlntermsofe|ementaryfunctlons.
We wl|| study varlous devlces for obtalnlnggoodapproxlmatlonstothe va|ues of
the none|ementaryfunctlonsweencounter In Chapter6wewl||dlscussnumerlca|
lntegratlonofdlerentla|equatlonslnsomedetal|
52 Chapter I First-Order Di fferenti al Equati ons
!aat.rj .,j,
Ameaatx(t)
Ve|ameV(t)
Cea;eattat|ea _ (/)
Oatat.
r_ .,
_

FIGURE 1.5.4. The single-tank


mixture problem.
Mixture Problems
Asanrstapp|lcatlonof|lnearnrst-orderequatlons,weconslderatankcontalnlnga
so|utlonamlxtureofso|uteandso|ventsuchassa|tdlsso|vedlnwater. Therels
both lnowandoutow, and we wantto computethe.-ox,; ofso|uteln the
tank at tlme , glven the amountx,o; x,at tlme o Supposethat so|utlon
wlth aconcentratlon of., grams ofso|uteper |lter ofso|utlon ows lnto the tank
at the constantrate of, |lters persecond, and thatthe so|utlonlnthe tankkept
thorough|ymlxedbystlrrlngowsoutattheconstantrateofr |lterspersecond.
To set up a dlfferentla| equatlon for x, ; , we estlmate the change cxlnx
durlngthe brlef tlme lnterva| .+ c } The amountofso|utethatowslnto the
tankdurlng c seconds ls , .,cgrams. Tocheckthls, notehowthecance||atlon
ofdlmenslonschecksourcomputatlons.
|lters grams
,

.,

, c seconds)
second |rter
yle|dsaquantltymeasuredlngrams.
Theamountofso|utethatowsoutofthetankdurlngthesametlmelnterva|
dependsontheconcentratlon.,;ofso|utelntheso|utlonattlme Butasnotedln
Flg. I 5 4, . ,; x,;}r,; ,where r,; denotes the vo|ume (notconstantun|ess
, ;ofso|utlonlnthetankattlmeThen
cx gramslnput- gramsoutput , .,c- .c
Wenowdlvldebyc
cx

, .,

c
Flna||y, wetake the |lmltasc c, l fa|| thefunctlonslnvo|vedarecontlnuous
andx,; l sdlerentlab|e,thentheerrorlnthlsapproxlmatlona|soapproacheszero,
andweobtalnthedlfferentla|equatlon
ax
lnwhlch, .., .andareconstants,but.denotesthevarlab|econcentratlon
x ,;
. ,;
r,;
( I 6)
( I 7)
ofso|ute ln the tank attlme Thusthe amountx,; ofso|ute ln the tanksatlsnes
thedlerentla|equatlon
ax
, ., -x
a r
( I S)
If r, r,o; , then r,; r, + (, - ; , so Eq. ( I S) ls a |lnear nrst-order
dlfferentla|equatlonfortheamountx,;ofso|utelnthetankattlme
Imporant: Equatlon( I S) need notbecommlttedto memory. Itls them-
.essweusedtoobtaln thatequatlonexamlnatlon ofthebehavlorofthesystem
overashorttlmelnterva| ,+At]thatyoushou|dstrlvetounderstand,because
ltlsaveryusefu|too|forobtalnlnga||sortsofdlfferentla|equatlons.
Remark: Itwasconvenlentforustousegmass/vo|umeunltslnderlvlng
Eq. ( I S) Butanyotherconslstentsystemofunltscanbeusedtomeasureamounts
ofso|ute and vo|umes ofso|utlon. In the fo||owlng examp|ewemeasure both ln
cublckl|ometers
cXump| e4
I . Li near Fi rst-Order Equati ons 53
Assu
g

e E
g
s a

|u
_
e of4S0k' andthatltsrate oflnHow(fron
LakeHuron)andoutHow(to LakeOntarlo)areboth350km

peryear. Supposethat
atthe tlme = 0(years), the po||utantconcentratlonofLake Erlecausedbypast
lndustrla|po||utlonthathas now been orderedto ceaselsnvetlnesthatofLake
Huron Iftheoutowhenceforthlsperfect|ymlxed|ake water, how|ongwl||lttake
toreducethe po||utlonconcentratlonlnLakeErletotwlcethatofLakeHuron!
Sol uti on Herewehave
cXump| e
t
=4S0(km

) ,
= = = 350(km

Jyr) ,
c = c(thepo||utantconcentratlonofLakeHuron), and
x,= x,c;=cr,
andthequestlonl sthls. Whenlsx,; = 2cr! Wlth thlsnotatlon, Eq ( I S)l sthe
separab|eequatlon
ax
- = c - x
a r

whlchwe rewrlte lnthe|lnearnrst-orderform


ax
a
+px= g
( I 9)
(20)
wlthconstantcoemclents p = J\, g = c. and lntegratlngfactor p e You
caneltherso|vethlsequatlondlrect|yor app|ytheformu|aln( I 2) The|atterglves
x,; =e
-

x,+

a = e
-

x,+ e

- I j
=e
-


c
t
+
_
e

'- I j ,
x,;=cr+:c re
-

'
To nndwhenx,; =2c r,wethereforeneedon|yso|vetheequatlon
(2I )
r 4S0
cr+:c re
-

'=zc r for = - |n 4 = |n 4 I 90I (years) .


350
. - . ... .... ... ... .. .
A I 20-ga||on (ga|) tanklnltla||ycontalns90|bofsa|tdlsso|vedln90ga|ofwater.
Brlnecontalnlng2 |bJga|ofsa|towslntothe tankattherateof:ga|Jmln,andthe
we||-stlrredmlxtureows out ofthe tank at the rateof3 ga|Jmln How much sa|t
does thetankcontalnwhenltlsfu||!
Sol uti on The lnterestlngfeatureofthls examp|e l sthat, duetothedlerlngratesoflnHow
andoutow,thevo|umeofbrlnelnthetanklncreasessteadl|ywlth r,; = 90+
ga||ons The change cx ln the amountx of sa|t ln the tank from tlme to tlne
+ c (mlnutes)lsglvenby
cx (4) (2)c- 3
x
c ,
90 +
54 Chapter I First-Order Di fferenti al Equati ons
soourdlerentla|equatlonls
dx
-+ x = S.
dt c+t
Anlntegratlngfactorls
whlchglves
p(x) = exp

dt

= c

,
.-
@
,c+)

,
c +t

,c+t )

x] = s,c+t)

,
,c+t )

x = z,c+ )
.
+
.
Substltutlonofx,c) = cglves c = -,c)
.
, sotheamountofsa|tlnthetankat
tlmet ls
c
.
x(t ) = z,c+t ) -
,c+t )

The tank ls fu||aftercmln,andwhent = c, wehave


ofsa|tlnthetank
:
h|
..........., ..,.....
.1 .... .............
.,....,............,...
......,x.
1. y' + y = 2, y(O) = 0
2. y' - 2y = 3e
2
x , y(O) = 0
3. y' + 3y = 2xe-
3
x
4. y' - 2xy =

5. xy' + 2y = 3x, y( 1 ) = 5
6. xy' + 5y = 7x
2
, y(2) = 5
7. 2xy' + y = 1 O.
8. 3xy' + y = 1 2x
9. xy' - y = x, y( 1 ) = 7
10. 2xy' - 3y = 9x
3
11. xy' + y = 3xy, y( 1 ) = 0
12. xy' + 3y = 2x
5
, y(2) = 1
13. y' + y = y(O) = 1
14. xy' - 3y = x
3
, y( 1 ) = 1 0
15. y' + 2xy = x, y(O) = -2
16. y' = ( 1 - y) cos x, y(n) = 2
17. (1 + x)y' + y = cos x, y(O) = 1
18. xy' = 2y + x
3
cos x
19. y' + y cot x = cos x
20. y' = 1 + x + y + xy, y(O) = 0
c
.
x,c)= z,c+c)-
l zc

zcz(|b)
21. xy' = 3y + x
4 cos x, y(2n) = 0
22. y' = 2xy + 3x
2
exp(x
2
) , y(O) = 5
23. xy' + (2x - 3) y = 4X
4
24. (x
2
+ 4) y' + 3xy = x, y(O) = 1
dy
25. (x
2
+ 1 ) -+ 3x
3
y = 6x exp (-x
2
) , y(O) = 1
dx

.., ..,..........2.
.....y .....,..........x.
26. (1 4xy
2
)
dy
= y
3
dx
28. ( 1 + 2xy)
dy
= 1 + y
2
dx
.y
27. (x + yeY)
dx
= 1
29. Express the general solution of dyjdx = 1 + 2xy in terms
of the error function
2 2
erf(x) = _
.
30. Express the solution of the initial value problem
.y
2x - = y + 2x cos x, y( 1 ) = 0
dx
as an integral as in Example 3 of this section.
.........,....
...,......,.....,..
..............,.......,
31. (a) Show that
c (x) =

P(x
) dx
is a general solution of ....+ (x) = O. (b) Show
that
._ (x) ~

P(x)
dx

Q(x)
P(X) dX
dx
}
is a particular solution of ...+ . = Q(x) .
(c) Suppose that Yc (x) i s any general solution of ....+
(x) .= 0 and that ._(x) is any particular solution of
...+ . = Q(x) . Show that ..~ Yc (x) + _ (x)
i s a general solution of ....+ . = Q(x) .
32. (a) Find constants and B such that ._ (x) ~ sin x +
B cos x is a solution of ....+ .= 2 sin x. (b) Use the
result of part (a) and the method of Problem to fnd the
general solution of ...+ .= 2 sin x. (c) Solve the
initial value problem dy/dx + .= 2 sin x, y(O) = 1 .
33. A tank contains 1 000 liters (L) of a solution consisting of
1 00 kg of salt dissolved in water. Pure water is pumped
into the tank at the rate of 5 L/s, and the mixture-kept
uniform by stirring-is pumped out at the same rate. How
long will it be until only 1 0 kg of salt remains in the tank?
34. Consider a reservoir with a volume of 8 billion cubic feet
(f
3
) and an initial pollutant concentration of 0. 25%. There
is a daily infow of 500 million ft
3
of water with a pollu
tant concentration of 0. 05% and an equal daily outfow of
the well-mixed water in the reservoir. How long will it
take to reduce the pollutant concentration in the reservoir
to 0. 1 O%?
35. Rework Example 4 for the case of Lake Ontario, which
empties into the St. Lawrence River and receives infow
from Lake Erie (via the Niagara River). The only difer
ences are that this lake has a volume of 1 640 k
3
and an
infow-outfow rate of 41 0 k
3
/year.
36. A tank initially contains 60 gal of pure water. Brine
containing 1 Ib of salt per gallon enters the tank at
2 gal/min, and the (perfectly mixed) solution leaves the
tank at gal/min; thus the tank is empty afer exactly 1 h.
(a) Find the amount of salt in the tank afer minutes.
(b) What is the maximum amount of salt ever in the tank?
37. A 400-gal tank initially contains 1 00 gal of brine contain
ing 50 Ib of salt. Brine containing Ib of salt per gallon
enters the tank at the rate of 5 gal /s, and the well-mixed
brine in the tank fows out at the rate of gal/s. How
much salt will the tank contain when it i s full of brine?
38. Consider the ....of two tanks shown in Fig. 1 . 5. 5,
with V
I
= 1 00 (gal ) and V
2
= 200 (gal) the volumes of
brine in the two tanks. Each tank also initially contains
50 Ib of salt. The three fow rates indicated in the fg
ure are each 5 gal/min, with pure water fowing into tank
1 . (a) Find the amount .of salt in tank 1 at time
(b) Suppose that .i s the amount of salt i n tank 2 at
I . Li near First-Order Equati ons 55
time Show frst that
..
.
. .
1 00
-
200
'
and then solve for . using the function .found in
part (a). (c) Finally, fnd the maximum amount of salt
ever in tank 2.
FIGURE 1.5. 5. A cascade of two tanks.
39. Suppose that in the cascade shown in Fig. 1 . 5. 5, tank 1
initially contains 1 00 gal of pure ethanol and tank 2 ini
tially contains 1 00 gal of pure water. Pure water fows
into tank 1 at 10 gal/min, and the other two fow rates
are also 1 0 gal/min. (a) Find the amounts .and .
of ethanol in the two tanks at time _ O. (b) Find the
maximum amount of ethanol ever in tank 2.
40. A multiple cascade i s shown in Fig. 1 . 5. 6.
FIGURE 1.5.6. A multiple cascade.
At time ~ 0, tank 0 contains 1 gal of ethanol and 1 gal
of water; all the remaining tanks contain 2 gal of pure wa
ter each. Pure water is pumped into tank 0 at 1 gal/min,
56 Chapter I First-Order Di fferenti al Equati ons
and the varying mixture in each tank i s pumped into the
one below it at the same rate. Assume, as usual, that the
mixtures are kept perfectly uniform by stirring. Let .

denote the amount of ethanol in tank .at time


(a) Show that . =

(b) Show by induction on .


that

= -- for .~
.

(c) Show that the maximum value of .

for .~ 0 is

= .

.= .

. (d) Conclude from Stirling's


approximation ..


.that

41. A 30-year-old woman accepts an engineering position


with a starting salary of $30, 000 per year. Her salary
increases exponentially, with =

thou
sand dollars afer years. Meanwhile, 1 2% of her salary
i s deposited continuously in a retirement account, which
accumulates interest at a continuous annual rate of 6%.
(a) Estimate .in terms of t o derive the differential
equation satisfed by the amount . in her retirement
account afer years. (b) Compute A(40) , the amount
available for her retirement at age 70.
42. Suppose that a falling hailstone with density = 1 starts
from rest with negligible radius = Thereafter its ra
dius is = ..is a constant) as it grows by accretion
during its fall . Use Newton' s second law-according to
which the net force acting on a possibly variable mass
equals the time rate of change .,.of its momentum
,= to set up and solve the initial value problem
.
= . = 0,
.
where is the variable mass of the hai lstone, = ...
is its velocity, and the positive y-axis points downward.
Then show that ..= ..Thus the hailstone falls as
though it were under ...the infuence of gravity.
43. Figure I . S. 7 shows a slope feld and typical solution
curves for the equation . = . .
|
s
-
+
z
0
- 2
-+
--
-s
- |

x
rcukr1.5.7. Slope feld and solution
curves for . = . .
(a) Show that every solution curve approaches the
straight line .= . I as . +0. (b) For each
of the fve values ;, = 3 . 998, 3. 999, 4. 000, 4. 001 , and
4. 002, determine the initial value
;,(accurate to four dec
imal places) such that .= ;,for the solution satisfying
the initial condition .-S) =
44. Figure I . S. 8 shows a slope feld and typical solution
curves for the equation . = .+ .(a) Show that every
solution curve approaches the straight line .=
as . -0. (b) For each of the fve values ;, = -1 0,
-S, 0, S, and 1 0, determine the initial value ;,(accurate to
fve decimal places) such that .= ;, for the solution
satisfying the initial condition .-S) = ;,

|
s
-
+
2

- z
.
- -
- s
- 0

x
rcukr1.5. 8. Slope feld and solution
curves for . = .+ .
.45 ...46 ...............
...,..................
.,.2 ..............
= 0 ..........,..,......
.........200 .......
.,.............
...... ............
,..............
45. The incoming water has a pollutant concentration of
= 10 liters per cubic meter (Um
3
). Verify that
the graph of . resembles the steadily rising curve in
Fig. I . S. 9, which approaches asymptotically the graph of
the equilibrium solution . = 20 that corresponds to the
reservoir' s long-term pollutant content. How long does it
take the pollutant concentration in the reservoir to reach
S Um
3
?
:

:
.

..
..
. : .

s.
rcukr1.5.9. Graphs of solutions in
Problems 4S and 46.
46. The incoming water has pollutant concentration
1 0( 1 + cos Um
3
that varies between 0 and 20, with an
average concentration of 1 0 Um
3
and a period of oscilla
tion of slightly over 6i months. Does it seem predictable
that the lake' s polutant content should ultimately oscillate
periodically about an average level of 20 million liters?
I . Li near Fi rst-Order Equati ons 57
Verify that the graph of .does, indeed, resemble the
oscillatory curve shown in Fig. 1 . 5. 9. How long does it
take the pollutant concentration in the reservoir to reach
5 Llm
3
?
1 . b | COl O
Foranlnterestlng app|led prob|em that lnvo|ves the so|utlon ofa |lneardleren-
tla| equatlon, conslder lndoor temperature oscl||atlons that are drlven by outdoor
temperatureoscl||atlonsoftheform
A(t ) = a, + a, cos ot + b, sln ot ( l )
Ifo JI 2, thentheseoscl||atlonshaveaperlodof24hours(sothatthecyc|eof
outdoortemperaturesrepeatsltse|fdal|y)andEq ( I ) provldesarea|lstlcmode|for
thetemperatureoutsldeahouseon adaywhenno changelntheovera||day-to-day
weather pattem ls occurrlng For lnstance, for a typlca| Ju|y day ln Athens, GA
wlth a mlnlmum temperature of70 Fwhen t :(4 v , and a maxlmum of
90 Fwhent I 6(4e v , , we wou|dtake
A(t ) S0- I 0 cos o(t - 4) S0 5cos ot - .sln ot (2)
WederlvedEq. (2)byuslngtheldentltycos (o - f) cosocosf +slnosln to
getao = S0,al -, andb, -.lnEq ( I )
If we wrlte Newton' s |aw of coo|lng (Eq (3) ofSectlon I I ) forthe corre-
spondlnglndoortemperature u (t ) at tlme t , butwlth the outsldetemperature A(t)
glven by Eq ( I ) lnstead ofa constant amblent temperature A, we get the |lnear
nrst-orderdlfferentla|equatlon
thatls,
du
dt
-k(u - A(t ) ) ,
du
b
.
-+ku k(ao +al cos ot+ ,smot)
dt
(3)
wlth coemclent functlons P(t ) = k and Q(t ) kA(t ) Jplca| va|ues of the
proportlona|ltyconstantk rangefrom0 2 to0 5 (a|thoughk mlghtbe greater than
0 5 forapoor|ylnsu|atedbul|dlngwlthopenwlndows, or|ess than 0 2for awe||-
lnsu|atedbul|dlngwlthtlght|ysea|edwlndows)
SCENARIO: Supposethatouralrcondltlonerfal|sat tlme to = 0onemldnlght,
and wecannotafford to haveltrepalred untl| paydayattheendofthemonth. We
thereforewantto lnvestlgatetheresu|tlnglndoortemperaturesthatwemustendure
forthenextsevera|days
Beglnyour lnvestlgatlonbyso|vlngEq (3)wlththe lnltla|condltlonu(0) *
Uo (the lndoor temperature at the tlme ofthe fal|ure ofthe alr condltloner) You
maywantto usethelntegra|formu|asln49 and50of theendpapers,orposslb|ya
computera|gebrasystem Youshou|dgettheso|utlon
u (t ) ao +coe-
kt
+.,cos ot+d, sln ot , (4)
58 Chapter I Fi rst-Order Di fferenti al Equati ons
IOO
a
0
&5

_ &O

7a
7O
6a
' / " 36
6O
O

I O

2O

3O

/,a)
rcukr1. 5. 10. Solution curves
given by Eq. (5) with ,= 65, 68,
71 , . . , 92, 95.
I O0 ~
a
&5

_ &O

7a
7O
utdoor
tcmpcraturc
65
6O
O

' / = I 2
tcmpcraturc
'
' / 36
/,a)
rcukr1. 5. 11. Comparison of
indoor and outdoor temperature
oscillations.
where
wlth.-}l z
/..
.
+ /
:
/
.
a
.

/
:
+ .
:
Wlth.,sc, ., -, /, -.(as ln Eq. ,z) ,6 -}l z,and/c z
(forlnstance),thlsso|utlonreduces(approxlmate|y)to
- -
,; sc+c

, , - sz l ) +,z l )cos - , :c:)s m ,)


Observenrstthatthe'dampedexponentla|temlnEq. ,)approaches zero
as- +~.|eavlngthe|ong-term'steadyperlodlcso|utlon
-

-
, ,;sc+,z l )cos - , :c:)s m. ,:)
Consequent|y, the |ong-tem lndoor temperatures oscl||ate every z:hours around
thesameaveragetemperatureS0 Fastheaverageoutdoortemperature.
Flgure l l cshows a numberof so|utlon curves correspondlng to posslb|e
lnltla| temperatures , ranglng from : tto t Observe thatwhatever the
lnltla| temperaturethe lndoor temperature 'sett|es down wlthln about l s hours
to a perlodlc dal|y oscl||atlon. But the amp|ltude of temperature varlatlon ls |ess
lndoorsthanoutdoors. Indeed,uslngthetrlgonometrlc ldentltymentlonedear|ler,
Eq. ,:)canberewrltten(verlfythls | ) as
,; sc- ,: cc)cos - l ::,
-
sc- ,: cc)cos
l z
(- csz) ,)
Doyouseethatthlslmp|lesthatthelndoortemperaturevarlesbetweenamlnlmum
ofabout:Fandamaxlmumofabouts:F!
Flna||y, comparlsonofEqs. ,z)and ,)lndlcatesthatthelndoortemperature
|agsbehlndtheoutdoortemperaturebyabout csz- : hours,asl||ustrated
ln Flg. l l l Thus thetemperaturelnsldethe housecontlnuesto rlseuntl|about
cl. N. eachevenlng,sothehottestpartofthedaylnsldelsear|yevenlngrather
than |ate aftemoon (as outslde).
For apersona| prob|emtolnvestlgate, carry outa slml|arana|yslsuslngav-
erageJu|y dal|y maxlmumJmlnlmumnguresforyourown |oca|e andava|ueof/
approprlate to your own home. You mlght a|so conslder a wlnterday lnstead of
a summerday. (What ls the wlnter-summerdlfferenceforthe lndoortemperature
prob|em!) Youmaywlshto exp|orethe use ofaval|ab|etechno|ogyboth to so|ve
thedlerentla|equatlonandtographltsso|utlonforthelndoortemperaturelncom-
parlsonwlththeoutdoortemperature.
I . Substi tuti on Methods and Exact Equati ons 59
Substitution Methods and Exact Eq1ations
cXump| e 1
The nrst-orderdlerentla| equatlons we have so|vedlntheprevlous sectlonshave
a|| been eltherseparab|eor|lnear Butmany app|lcatlonslnvo|vedlerentla|equa-
tlonsthatareneltherseparab|enor|lnear Inthlssectlonwel||ustrate(maln|ywlth
examp|es) substltutlon methods that sometlmes can be used to transform aglven
dlfferentla|equatlonlnto onethatwea|readyknowhowtoso|ve
Forlnstance,thedlerentla|equatlon
ay
ax
=),x, y; , , l )
wlthdependentvarlab|eyandlndependentvarlab|ex, maycontalnaconsplcucus
comblnatlon
c = :,x, y; ,z)
ofxandythatsuggests ltse|fasanewlndependentvarlab|ecThusthedlerentla|
equatlon
ay
:
ax
=,x
'
y
'

practlca||ydemandsthesubstltutlonc=x
'
y
'

oftheformlnEq ,z)
Ifthesubstltutlonre|atlonlnEq ,z)canbeso|vedfor
y=;,x, c ,

then app|lcatlon ofthe chaln ru|eregardlng c as an (unknown) functlon cfx-


yle|ds
ay _ -;ax -;ac

ac
ax

-xax
'
-cax

'
ax

,:)
where the partla| derlvatlves -;,-x = ;

,x,c and -;,- c = ;, ,x, c areknown


functlonsofxandc Ifwesubstltutetherlght-handsldeln,:)foray,axlnEq , l )
andthenso|veforac,ax.theresu|tl s anew dlerentla|equatlonoftheform
ac
- = ,,x, c
ax
(5)
wlth new dependentvarlab|e c Ifthls new equatlon ls eltherseparab|eor |lnear,
then wecanapp|ythemethodsofprecedlngsectlonstoso|velt
Ifc= c,x ls aso|utlonofEq ,), theny=;,x, c,x, wl||beasc|utlonct
theorlglna|Eq , l ) The trlck ls to se|ecta substltutlon suchthatthetransfcrmed
Eq ,)ls one we can so|ve Even when posslb|e, thls ls not a|ways easy, ltmay
requlreafalramountoflngenultyortrla| anderror
So|vethedlfferentla|equatlon
ay
:
- =,x
'
y
'


ax
60 Chapter I Fi rst-Order Di fferenti al Equati ons
Sol ution As|aa|.a:eaea:i|e:,ie: s::y:aesa|s:|:a:|ea

Z
> 0
- Z

- 6
- &
- I 0

x
FIGURE 1.6.1. Slope feld and
solution curves for
y
'
.+ y + 3)
2
.
1aea
c=x+y+ , :aa:|s, y=c- x-
ay ac
- = - - l ,
ax ax
se:ae::aasie:meaeaa:|ea|s
ac

- = l + c
ax
1a|s|sasea:a|ieeaa:|ea,aaaweaaveaea|ia.ai:y|ae|:a|a|a,|:sseia:|ea
ac
x =
l+c

=:aa
-
,
c+C.
se c = :aa,x- C) . se.aasec = x+y+, :ae,eae:aiseia:|eaei:aee:|,|aai
eaa:|eaay}ax = ,x+y+)

|sx+y+=:aa,x- C) , :aa:|s,
y,x) =:aa,x- C)- x-
Remark: t|,a:e l : l saewsasieeaeiaaaa:y|.aiseia:|ea.a:vesie:
:aea|ne:ea:|aieaa:|eaeisamiel wesee:aa:,ai:aea,a:aeiaa.:|ea),x,y)=
,x +y +)

|s.ea:|aaeasiya|n::ea:|a|ieie:aiixaaay, ea.aseia:iea|s.ea:|aaeas
eaiyeaa|eaaaea|a:e:vai iaa::|.aia:,|e.aase:ae:aa,ea:iaa.:|ea|s.ea:|aaeas
ea :ae eea |a:e:vai ( - }z,}z) , :aea::|.aia:seia:iea w|:a a:||::a:y .eas:aa:
vaiaeC|s.ea:|aaeasea:ae|a:e:vaiwae:e-}z< x-C < }z,:aa:| s, C- n<
x < C +}z 1a|ss|:aa:|ea|sia|:iy:y|.aieiaeai|aea:a|ne:ea:|aieaa:|eas, |a
.ea::as:w|:ai|aea:a|iie:ea:|aieaa:|eas, waeseseia:|easa:e.ea:|aaeaswae:eve:
:ae.eeia.|ea:iaa.:|eas|a:aeeaa:|eaa:e.ea:|aaeas
samieI |iias::a:es:aeia.::aa:aaya|ne:ea:|aieaa:|eaei:aeie:m
ay
ax
= I,ax+/y+c) ,:)
.aa|e::aasie:mea|a:easea:a|ieeaa:|ea|yaseei:aesa|s:i:a:|eac = ax+
/y+c ,seer:e|iem) 1aea:a,:aas :aa:ieiiewaeai w|:a e:ae:.iassesei
a:s:e:ae:eaa:|easie:wa|.a:ae:ea:e s:aaaa:a sa|s:|:a:|eas:aa:a:e |aewa :e
sa..eea
Homogeneous Equations
Ahomogeneous a:s:e:ae:a|iie:ea:|aieaa:|ea|seae:aa:.aa|ew:|::ea|a:ae
ie:m
ay
= I,
ax x
iiwema|e:aesa|s:|:a:|eas
y
c= -, y=cx,
x
ay ac
- = c + x -,
ax ax
,)
,s)
cXump| eZ
I . Substi tuti on Methods and Exact Equati ons 61
:aea ,)| s ::aasie:mea|a:e:aesep..//eeaa:|ea
ac
x- =i,c;- c
ax
1aaseve:yaeme,eaeeasa:s:e:ae:a|ne:ea:|ai:aa:|ea.aa|e:eaa.:a:eaa|a::
,:a:|ea:e|iem|ymeaasei:aesa|s:|:a:|eas|a,s)
Remark: Aa|.:|eaa:yaeaa|:|eaeiaeme,eaeeas|seias|m|ia:||aa
e:aa:a:eCeas|ae:aa|ne:ea:|aieaa:|eaei:aeien
waeseeiyaem|ai.eeu.|ea:iaa.:|easa:eaeme,eaeeas|a:aesease:aa::a.a
ei:ae|::e:msaas:aesame:e:aiae,:ee,m +h = p+g = +s = . ii:
a|v|aeea.as|aeei( *) |yx
t
, :aea:ae:esai:-|e.aasex
-
y
.
,x
-
.
= ,y,x;
.
.aaa
seie::a-|s:aeeaa:|ea

.
ay

y

y
.
A - -

B - + c -
x ax x x
wa|.aev|aea:iy.aa|ew:|::ea,|yaae:ae:a|v|s|ea)|a:aeienei ,) Me::
,eae:aiiy,aa|iie:ea:|aieaa:|eaei:aeienP,x.y;y= _,x.y;w|:aeiyaem|ai
.eeia.|ea:sP aaa_|saeme,eaeeas|i:ae:ens|a:aeseeiyaem|aisaiiaa::a:
same:e:aiae,:eeK. 1aea|ne:ea:|aieaa:|ea|a:aeieiiew|a,esamie|sei:a|s
ie:mw|:aK =z
seive:aea|ne:ea:|aieaa:|ea
ay
zxy- =:x
:
+y
ax
Sol ution 1a|seaa:|ea|sae|:ae:sea:a|ieae:i|aea:,|a:we:e.e,a|ze|:asaaeme,:a:eas
eaa:|ea|yw:|:|a,|:|a:aeien
ay
=
:x
:
+y
:
=z

.
ax :y y z x
1aesa|s:|:a:|eas|a,s):aea:a|e:aeie:m
y=cx.
1aesey|eia
aaaaea.e
ay ac y
- = c + x -.
ax ax
c= -.
x
ac z
c + x - = - +-c.
ax c z
x
aaa - =
c y
ac z : c
:
+:
x - = - + - = ,
ax c z zc
ac= - ax,
zc
c
:
+ : x
ia,c
:
+:;=ia x +iac
62 Chapter I Fi rst-Order Di fferenti al Equati ons
x
FIGURE 1.6.2. Slope feld and
solution curves for
2xyy' .x

+ .

weaiy:aeeseaea:|aiiaa.:|ea:e|e:as|aesei:aeias:eaa:|ea:ee|:a|a
Ne:e:aa::aeiei:aaaas|aeei:a|seaa:|ea|sae.essa:|iyaeaae,a:|ve i:ieiiews
:aa:/ > c |a:ae.aseeiseia:|eas:aa:a:eaeaaeaie:x > c, wa|ie/ < cie:
seia:|easwae:ex < 0. iaaeea,:aeiam|iyeiseia:|ea.a:ves|iias::a:ea|at|,i : z
esa|||:ssymme::ya|ea:|e:a.ee:a|aa:eases A.:aaiiy,:ae:ea:ees|:|vevaiaea
aaaae,a:|vevaiaeaseia:|easei:aeie:msy,x) = /x

- :x

:aa:a:eaeaaea
ie:x > :}/|i:ae.eas:aa:/|ses|:|ve,aaaie:x < :}/ |i/|sae,a:|ve
seive:a:|a|:|aivaiae:e|iem
wae:ex,> 0.
ay
x
ax
= ) +x

- y

, y,x,) = c,
Sol ution wea|v|ae|e:as|aes|yxaaaaaa:aa:
a0
40
J0
Z0
I 0
> 0
-I 0
-Z0
-J0
-40
x
FIGURE 1.6.3. Solution curves
for xy' y + ,.

sewema|e:aesa|s:|:a:|eas|a,s), we,e:
ac

c+x
ax
= c+ i- c

,
,
.
ac= , ax,

x
s|a

c= ia x+c
weaeeaae:w:|:eia x |e.aasex > caea:x = x,> 0. Newae:e:aa:c,x,) =
y,x,)}x,= c,sec= s|a

c- ia x,= - ia x, uea.e
aaa:ae:eie:e
c= = s|a,ia x- ia x,)= s|aia

,
x x,
y,x) = xs|aia

|s:aeaes|:eaa::|.aia:seia:|ea t|,a:ei : saewsseme:y|.aiseia:|ea.a:es


se.aaseei:ae:aa|.ai|a:aea|ne:ea:|aieaa:|ea,:aeseseia:|ea.a:vesa:e.eaaaea
:e:ae|aa|.a:ea::|aa,aia::e,|eax y Yea.aa.ae.|:aa::ae|eaaaa:yi|aes
y= xaaay= -x,ie:x > c)a:es|a,aia:seia:|ea.a:ves:aa:.eas|s:eie|a:sei
:aa,ea.yw|:a:aeseia:|ea.a:vesieaaaea:i|e:
cXump| e4
I . Substi tuti on Methods and Exact Equati ons 63
Bernoulli Equations
Aa:s:e:ae:a|iie:ea:|aieaa:|eaei:aeie:m
ay
-+i,x y= _,x y
.
ax
,)
|s .aiieaaBernoulli equation. iie|:ae: = ce: = i .:aea ,) |si|aea:
O:ae:w|se,asweas|yea:esaew|ar:e|iem:, :aesa|s:|:a:|ea
::aasie:ms,|a:e:aei|aea:eaa:|ea
ac
-+,i- i,x c= ,i- _,x
ax
, i
ka:ae::aaameme:|z|a,:aeie:mei:a|s::aasie:meaeaa:|ea,|:|sme::eu.|:a:
:ema|e:aesa|s:|:a:|ea|a, i esi|.|:iy,as|a:aeieiiew|a,:sami:s
If we:e:|:etheaeme,eaeeaseaa:|eazxyy = :
,
:
+yeisami:z|a:a:
ie:m
ay zx
- - -y = -.
ax zx y
we see :aa: |:|s aiseaseneaii|eaa:|ea w|:a i,x = -},zx . _,x = zx.
= -i , aaai- = z uea.ewesa|s:|:a:e
1a|s,|es
@ :
:~ y . aaa
ay ayac i
-

:
ac
- = -- = -u
.
ax acax z 1x
i ac

c
-
,

:
. a ,
:
,

:
= zxc
-
,

z ax zx
1aeamai:|i|.a:|ea|yzc

:
:eaa.es:aei|aea:eaa:|ea
ac
- - -c = :x
ax x
w|:a|a:e,:a:|a,ia.:e:p = c

= x
-

sewee|:a|a

:
n,x c = _,
x

64 Chapter I Fi rst-Order Di fferenti al Equati ons


cXump| e
cXump| e
1aeeaa:|ea
x
ay
+:y=xy
-

ax
|sae|:ae:sea:a|ieae:i|aea:ae:aeme,eaeeas, |a:|:|saseneaii|eaa:|ea|:a
= , i- h =- 1a:sa|s:|:a:|eas
c=y

.
::aasie:m|:|a:e

) =c . aaa
ay ayac

-
ac
- = -- = -c -
ax ac ax ax
ac
xc
-
-
-+:c
-

=xc
-
-

ax
D|v|s|ea|y-xc
-
y|eias:aei|aea:eaa:|ea
ac z
- - -c = -i
ax x
w|:a|a:e,:a:|a,ia.:e:p =e'
-
:

-
=x
-
:
1a|s,|:s
aaaaaaiiy,
l
y,x =
,x+cx
:

1aeeaa:|ea

, l l )
| s ae|:ae:sea:a|ie,ae:i|aea:,ae:aeme,eaeeas, ae:| s | : ase:aeaii|eaa:|easa:
wee|se:ve:aa:yaea:seaiy|a:ae.em||aa:|ease
:

aaan,e
:

=ze
:

y 1a| s
:em:s:aesa|s:|:a:|ea
ac
=ze
:

ay
ax ax
:aa:::aasie:ms , l l ) |a:e:aei| aea:eaa:|eaxc ,x =x
-
+c,x , :aa:|s,
ac l

- - -c =x
ax x
Ai:e:mai:|iy|a,|y:ae|a:e,:a:|a,ia.:e:p = i}x. weaaa:aa:
c = x
:
ax = x

+ c. se e
:
= c = x
-
+ cx.
aaaaea.e
y,x= ia x
-
+cx
Y
FIGURE 1. 6.4. The airplane
headed for the origin.
X
FIGURE 1.6.5. The components
of the velocity vector of the
airplane.
I . Substi tuti on Methods and Exact Equati ons 65
Flight Trajectories
saes::aa:aaa|:iaaeaea::si:em:aee|a:,. .;ie.a:eaaa::as:ei|:s|a::aa:a
aes:|aa:|ea-aaa|:e::ie.a::aa::aee:|,|a (0, 0) 1aeiaae::a:is|:a.eas:aa:
seea :,:eia:|ve:e:aew|aa,wa|.a|s|iew|a,aa:ae::a|:a .eas:aa:s::aU.
As |aa|.a::a| at|, l . 6 4, w:assam::aa::a:iaae s|ie:ma|a:a|as|:sa:aa|a,
a|:e.:iy:ewa:a:aee:|,|a
t|,a:e I . 6. 5aeisasae:|::a:iaa: sveie.|:y.emea:a:s::ia:|::e:a:
,:eaaa 1aeya::
ax :,x
- = -c, .es ^ =- ,
a
,x
:
+y
:
ay :,y
- =:,sm^+=- +
1
,x
:
+y
:
uea.::ae::a]e.:e:yy=),xei:aeiaaesa:|saes:aea|ii:::a:|ai:aa:|ea
liwes::
1y ay,a I

- = = :,y- , x
:
+y
:

1x 1x,1 :,x

/ = -,
:,
( l 2)
( l 3)
:a::a:|eei:a:w|aas::a:e:a:iaa: sa|:seea,:aea( I 2) :ases:a:aeme,:
aeeasie:m
1aesa|s:|:a:|eay=x:. y=:+xc:a:ai:aas:ea:|aeiy:e

ax

x
( l 4)
( l 5)
sy::|,eaem:::|.sa|s:|:a:|ea,e:|y.easai:|a,a:a|i:ie::a:|a::,:aiea:a:|:i:,
weaaa:aa:
|a

c +

=-/ ia x+ c.
aaa:ae|a|:|ai.eaa|:|eac,.=y,.,. =0y|eias
c=/ |a .
( I 6)
( I 7)
Asw:as|yea:esaew|ar:e|i:m6S, :ae:esai:eisa|s:|:a:|a,( I 7)|a( l 6)aaa
:aeaseiv|a,ie::|s
s:.aasey=x:. eaaaiiye|:a|a

.
,

y,x
-

2 .
ie::a:eaa:|eaei:a:iaa: s::a]e.:e:y
( l )
( l 9)
66 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| e
Ne:e:aa:eaiy|a:ae.ase/ < l ,:aa:|s, U < c,;aees:ae.a:e|a ( l 9)
ass:a:ea,a:aee:|,|a, se:aa::aeiaae:ea.aes |:s aes:|aa:|ea iiU :, ,se
:aa:/ :aea l 9:a|es:aeie:m y,x; , .

x
:
}.
:
; . se:aeiaae s
::a]e.:e:y a:ea.aes :aee|a: .:a:ae::aaa 1ae s|:aa:|ea|seea
we:se|i > :,,se/ > i )-|a:a|s.ase|:ieiiewsi:em, l ):aa:y as
x 0 1ae:a:ee.asesa:e|iias::a:ea|at|, l : :
ii.|:,= : |,a aaa m|}a, :aea/,c, se:aeiaae
w|iisa..eea|a:ea.a|a,:aea|:e::a: w|:a:aesevaiaes, y|eias


y,x;

New saese:aa:wewaa::e aaa:ae mas|mamameaa:|y wa|.a:aeiaae |s


|iewaeii.ea:seaa:|a,|:s::| 1aa:|s, waa:|s:aemas|mamvaiaeeiy,x;ie:
0 x
Sol ution D|iie:ea:|a:|eaei:aeiaa.:|ea|a y|eias

(u,0)
X
FIGUR 1.6.6. The three cases
U 0q (plane velocity exceeds
wind velocity), U = 0q (equal
velocities), and U ~ 0q (wind is
greater) .
ay



ax

aaawe:eaa|iyseive:aeeaa:|eay,x;:ee|:a|a,x

uea.e
_

)max

+

: s l
1aas:aeiaae|s|iewaaimes:l m|ae::|a:eaee|a:aa:|a,|:swes:wa:a]ean:y
,1ae,:aaei:aeiaa.:|ea|a |s:aeeaeasea:e.eas::a.:t|, l : : 1ae
ve::|.ais.aie:ae:e|sesa,,e:a:ea|yaia.:e:ei: )
Exact Diferential Equations
we aaveseea:aa:a,eae:aiseia:|eay,x; eiaa:s:e:ae:a|ne:ea:|aieaa:|ea|s
ei:eaaeaaea|mi|.|:iy|yaaeaa:|eaei:aeie:m
i,x. y,x; ; c.
wae:ec| s a.eas:aa:Oa:aee:ae:aaaa,,|vea:ae|aea:|:y|a we.aa:e.ee:
:aee:|,|aaia|iie:ea:|aieaa:|ea|ya|iie:ea:|a:|a,ea.as|aew|:a:ese.::exr:e
v|aea:aa: |mi|.|:iyaeaaesyasaa|ne:ea:|a|ieiaa.:|eaeix. :a|s,|v:s
:aee:|,|aaia|ne:ea:|aieaa:|ea|a:aeie:m
:aa:|s,
-i -iay

-x -yax
ay
M,x. y;x,x. y; -
ax
wae:eM,x. y;i,x. y;aaax,x. y;i

,x.y;

cXump| ed
I . Substi tuti on Methods and Exact Equati ons 67
.....,.,..
. ax+. ay= .
..diferential . ..,....,..*
).......= )...= -l ..,
.,......,.... ..
/e.,..

..
.

- = x,

. = c
.,,..,..., ...,...
exact diferential equation-the ..
a= i

ax+i

ay
. .,ax+ay
...,... ............
,.............
..= ..= ....,......
....,............,
..,...,.,.. = .,.....x
.....,..........
....,..

= = =

.
=
.

.e.ess..oa//o.....,..ax+ay= .
...= .....,..,...
...,........= ..= x.
...
....,..
,ax+.yay=
.........,..... = ....
,,,..= ..= . ....,...,
,.,. = .


68 Chapter I Fi rst-Order Di fferenti al Equati ons
...,,.........de...,....,
s,y
:
.
y ax+x ay0.
..,.......My..xx,....
-M
- x
= o
-y -x
..,.., ,z:)..
,z:)
...............,..
,z)..,z:)..,,......,....,....
,.........,...,..
....,.Max+xay....
.......,..,....,..
,...,.,z:)..s./e..
... M
,
x

,....,..Max+xay
.
THEOREM 1 Criterion for Exactness
..,,...M,x. y; ..x,x. y;........
...,.......,.,r . < x < /,
c < y < a.....,..
M,x.yax+x,x. yay
.r..,
-M - x

-y -x
,z
,z:)
...,r ......i,x.y..r..
- i }-xM..- i }-yx..,.,,z+).r
Proof: ......,...,.,,z:)..
., ,z ..,...........,,z:).
......i,x. y....- i}-x M..- i}-y x
...y.,,y , ..
i,x. y,M,x. yax+,,y ,z)
. ..-i}-xM.,,z), ..]M,x.yax.
......M,x. y..,x ,..,,y..
x
- i
,M,x. y ax+,

,y
-y -y
.. .. ..

.
, _ ,yx- ~ M,x.yax
-y
,zs
cXump| e
I . Substi tuti on Methods and Exact Equati ons 69
..../s....y, .......,.....
.,(2S)..y....,,y;,,.,..,
y ....,......,(2S)....,...
......x, ............,x
..,..
x- ,M,x. y; ax,=
-x
- ,M,x. y; ax
-x -y -x -x-y
- x - -
, = -- -- M,x. y; ax
-x -y -x
- x -M
= -- -=
-x -y
,.,,. ..........,,y;,,.,
.,(2S). ... ..,(27).
i,x. y; =,M,x. y; ax +,x,x. y; -
-
-
y
,M,
. y; ax, ay (29)
.......i

=M..i
,
=x
....,.,(29), ...,...,..
Max+xay=,.,,..,...,.,(27)..(2S) .
.,.M,x.y;..,x...
i,x. y; =,M,x.y; ax+ ,,y; .
..,..,,y;....,.,.....
...x .......,,y;,.,,....
- i}-y=x,x. y; .. ,..,....,.i,x. y;=c
......,..
,:xy
-
ax+,:y+x

- x;;ay=0
,;
5o| Ul| oD .M,x. y; =:xy- y

..x,x. y;=:y+x

- xy

..,.,..
...
-M
=:x

=
- x

-y -x
,.,- i}-x=M,x.y; ..,x, .,
i,x. y;=,,:xy ax=x

y- x,+,,y;
........, y..- i}-y=x,x. y; ..,.
- i
=x

x;+,
,y;=:y+x

- x;.
-y
70 Chapter I Fi rst-Order Di fferenti al Equati ons
x
FIGURE 1.6.7. Slope feld and
solution curves for the exact
equation in Example 9.
andltfo||owsthat, ,y; =4y. Hence,,y; =2y

+c, .andthus
Therefore,agenera|so|utlonofthedlerentla|equatlonlsdennedlmp|lclt|ybythe
equatlon
(3 l )
(wehaveabsorbedtheconstantc, lntotheconstantc;
Remark: Flgure l . 6. 7 shows a rather comp|lcated structure of so|utlon
curvesforthedlerentla|equatlon ofExamp|e 9. The so|utlon satlsfylngaglven
lnltla| condltlony(x,) =y,ls dennedlmp|lclt|ybyEq. (3 l ),wlthcdetermlnedby
substltutlngx=x,andy=y,lntheequatlon. Forlnstance, thepartlcu|arso|utlon
satlsfylngy,c; = l ls denned lmp|lclt|y by theequatlon 3x

y xy

+2y

= 2.
Theothertwospecla|polntslnthengureat(0, 0) andnear(0. 75 ,2.l 2)areones
wherebothcoefnclentfunctlons lnEq. (30)vanlsh, sothetheorem ofSectlon l . 3
doesnotguaranteeaunlqueso|utlon.
Reducible Second-Order Equations
se.oaoaea_ ee/./e,./olnvo|vesthesecondderlvatlveoftheunknown
functlony(x) , andthushasthegenera|form
i,x. y. y .y

;=c (32)
Ife//ethedependentvarlab|eyothelndependentvarlab|exlsmlsslngfrom a
second-orderequatlon, then ltls easl|yreducedby a slmp|esubstltutlontoa nrst-
orderequatlonthatmaybeso|vab|ebythemethodsofthlschapter.
Dependent variable ymissing. Ifylsmlsslng, thenEq. (32)takestheform
Thenthe substltutlon
i,x. y .y

;= c
_ ay
//
ap
p=y =
ax

y
ax
resu|tslnthe]soaedlerentla|equatlon
i,x. p.p;= c
(33)
(34)
Ifwecanso|vethlsequatlonforagenera|so|utlonp,x. c, lnvo|vlnganarbltrary
constantc| , then weneedon|ywrlte
y,x; =,y (x) ax=,p,x. c, ax + c

togetaso|utlonof Eq.(33)thatlnvo|vestwoarbltraryconstants c, andc

(as lsto
beexpectedln thecaseofasecond-orderd|erentla|equatlon) .
cXump| e 1
I . Substi tuti on Methods and Exact Equati ons 71
So|vetheequatlonxy+zy=6xlnwhlchthedependentvarlab|eylsnlsslng
Sol uti on Thesubstltutlondennedln(34)glvesthenrst-orderequatlon
,
3 4 5

FIGURE 1.6.8. Solution curves


C]
of the form y =
2
+ for

L 0, 3, 1 O, 20, 35, 60,


1 00.
ap
x
ax
+zp =:x,
ap z
thatls, - +-p=6
ax x
Observlngthattheequatlonontherlghtherels|lnear,wenu|tlp|ybyltslntegratlng
factorp =exp
,z}x;ax

=e


=x

andget
n

,x

p;=6x

,
x

p=zx
3
+c, .
ay c

p = - = zx + -
ax x

nna|lntegratlonwlthrespecttoxyle|dsthegenera|so|utlon

c
y,x; =x +- +c

X
ofthe second-orderequatlonxy+zy = 6x So|utloncurves wlth c

= 0but
c

/0areslmp|yvertlca|trans|atesoftheparabo|ay=x

(forwhlchc =c

0) Flgure l 6 S showsthlsparabo|aandsonetyplcal so|utloncurves wi thc

0
butc

/c So|utloncurveswlth c, andc

bothnonzeroarevertlca|trans|atesol
those(other thantheparabo|a) shown lnFlg l 6 S

Independent variable X missing. Ifxls nlsslng,thenEq ,z;takestheforn


/, y. y .y

;=c
Then thesubstltutlon
, ay
p=y =
ax

ap apay ap
y = - = -- = p-
ax ayax ay
resu|tslnthe]soaedlerentla|equatlon
(35)
(36)
forpasafunctlonofyIfwecanso|vethl sequatlonforagenera|so|utlonp,y. c, ;
lnvo|vlnganarbltraryconstantc ,then(assunlngthaty/0)weneedon|y wrlte

ax

ay
x,y;=
ay
ay=
ay}ax
ay=

ay=
p,y. c ;
+c

Ifthenna| lntegra| P =], l}p;aycanbeeva|uated,theresu|tlsanlnp|lcltso|utlon


x,y;= i,y. c+c

ofoursecond-orderdlerentla|equatlon
72 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| e 1 1
So|vetheequatlon,,= (,')
,
lnwhlchthelndependentvarlab|ex lsmlsslng.
Sol uti on
Weassumetemporarl|ythat,and,'arebothnonnegatlve,andthenpolntoutatthe
end thatthls restrlctlon ls unnecessary. The substltutlondenned ln(36) glves the
nrst-orderequatlon
5
4
3
:
j ~
- 0
[

:
-
3
, ,
: 3 4 5
x
FIGUR 1.6.9. The solution
curves y = .with B = 0 and
.= 0, 1 are the horizontal lines
y 0, 1 . The exponential curves
with B ~ 0 and .= 1 are in
color, those with B 0 and
. 1 are black.
Problems
.,
:
,- =
.,
Then separatlonofvarlab|esglves
.

=
.

ln,= ln,+ c (because, > 0and,= ,


'
> 0),
,= c, y
where c, = eC. Hence
dx I
., c, y

c, x=
.

= |n ,+cj .
Theresu|tlnggenera|so|utlonofthesecond-orderequatlon,, = (,' )
:
ls
where A = e-c
z
ands = c[ . Desplteourtemporary assumptlons, whlchlmp|y
thattheconstants A and sareboth posltlve, wereadl|y verlfy that,(x) = AeB
x
satlsnes ,, = (,' )
:
for .//rea| va|ues ofA and s Wlth s = 0 and dlfferent
va|uesofA, we get a|| horlzonta||lnes ln the p|aneas so|utloncurves. The upper
ha|fofFlg. I . 6. 9showstheso|utloncurves obtalnedwlth A = I (forlnstance)and
dlerent posltlve va|ues ofs Wlth A = - l these so|utlon curves are reHected
ln the x- axls, and wlth negatlve va|ues of s they are reHected ln the ,-axls. In
partlcu|ar, wesee thatwegetso|utlonsof,, = (,' )
:
, a||owlngbothposltlveand
negatlveposslbl|ltlesforboth, and,' .
.........., ..,.....
.1 ....30. .........,x
.....
15. x(x + y)y' + y(3x + y) = 0
16. y' = Jx + y + 1 17. y' = (4x + y)z
18. (x + y)y' = 1 19. xZy' + 2xy = 5y3
1. (x + y) y' = x -
;
3. xy' = y + 2.
5. x(x + y)y' = y(x - y)
7. xyzy' = x3 + y3
9. xZy' = xy + yZ
11. (xz - yZ)y' = 2xy
12. xyy' = yZ + x.r4x -
Z
=+-y"
z
13. xy' = y + , x2 + y2
14. yy' + x = Jx2 + y2
2. 2xyy' = x2 + 2y2
4. (x - y)y' = x + y
6. (x + 2y)y' =
;
8. x2y' = xy + x2Y/
10. xyy' = x2 + 3y2
20. y2y' + 2xy3 = 6x 21. y' =
;
+ y3
22. xZy' + 2xy = 5y4 23. xy' + 6y = 3xy4/3
24. 2xy' + y3

2= 2xy
25. y2(xy' + y)( l + X4) 1 /2 = x
26. 3y2y' + y3 =
27. 3xy2y' = 3x4 + y3
28. xeY y' = 2(eY + x3 e2x
)
29. (2x sin y cos y) y' = 4x2 + sin
2
y
30. (x + eY)y' = x

Y - 1
..31 ....42, ,......, ..
,.........
31. .. .... ..= 0
32. .. . ..6. . ..= 0
33. .

.....6.

..= 0
34. ..

...

...

..= 0
35. .

, .. .

ln . .. =
36. ( 1 . .... ..= 0
37. (cos .

.. ..= 0
38. .tan-
1
...
.Y
..= 0
1 .

39. .

...

...

..= 0
..sin y tan y) ..cos y + x sec
2
...= 0
41.

..

2;
.
.

_
1
_
..= 0
; . .

.
.
.

.
.
.

42.
.

..
.

..= 0
..................,
..,.....43-54. ..........
,....,... ....,11).
...= . 44. ...

= 0
....= 0 46. ...= ..
47. .= .

48. .

... =
....

= .. 50. .= ..

. .= ..

52. .

.= 1
. .= .. ...= .

. Show that the substitution = ...transforms


the differential equation ....= ...into a
separable equation.
56. Suppose that ./ 0 and ./ 1. Show that the sub
stitution = y
l
-
n
transforms the Beroulli equation
..... .= ,. .

into the linear equation


.
..
( 1 . . . = ( 1 . ,.
1. Show that the substitution = I n .transforms the difer
ential equation ..... .= ,. .In .into the
linear equation .... = ,. .
. Use the idea i n Problem to solve the equation
..

.
..
....In .= .
. Solve the diferential equation
.. . . 1
.. ..

by fnding h and .so that the substitutions .= ..


. .transform it into the homogeneous equation
. .
.. .
I . Substi tuti on Methods and Exact Equati ons 73
60. Use the method in Problem 59 to solve the diferential
equation
.. . .
.. .. . 1
61. Make an appropriate substitution to fnd a solution of the
equation ....= sin . . Does this general solution
contain the linear solution .. = . rf that is readily
verifed by substitution in the diferential equation?
62. Show that the solution curves of the diferential equation
.. ..
.
.

.. ..

are of the form .

= ..
63. The equation ....= .. .

.. .. is called
a Riccati equation. Suppose that one particular solution
Yl .of this equation is known. Show that the substitution
1
Y = Yl

V
transforms the Rccati equation into the ..equation
.
..
.2AYl ) V = .
....63 ..,.....
.64 ...65, ....Yl .= ........
..
64. .

= 1 .

..
..
65. ..= 1 .

..
66. An equation of the form
.= ....
is called a Clairaut equation. Show that the one
parameter family of straight lines described by
.. = ..
is a general solution of Eq.
67. Consider the Clairaut equation
.= ..
.
.

for which .. = .

in Eq. Show that the line


.= .

is tangent to the parabola .= .

at the point

,
Explain why this implies that .= .

is a singular solu
tion of the given Clairaut equation. This singular solution
and the one-parameter family of straight line solutions are
illustrated in Fig. 1 . 6. 1 0.
74 Chapter I Fi rst-Order Di fferenti al Equati ons
FIGURE 1. 6. 10. Solutions of the Clairaut
equation of Problem 67. The "typical" straight
line with equation .= .

is tangent to
the parabola at the point ,C,


68. Derive Eq. ( 1 8) i n this section from Eqs. ( 1 6) and ( 1 7) .
69. I n the situation of Example 7, suppose that .= 1 00 mi,
0q = 400 milh, and U = 40 mi/h. Now how far north
ward does the wind blow the airplane?
70. As in the text discussion, suppose that an airplane main
tains a heading toward an airport at the origin. If 0q = 500
milh and U = 50 milh (with the wind blowing due north),
and the plane begins at the point (200, 1 50) , show that its
trajectory is described by
.+ ,X

+ .

2(200X
9
)
1 / 1
O
.
Ppulation Models
71. A river 1 00 f wide is fowing north at U feet per second.
A dog starts at ( 1 00, 0) and swims at 0q = 4 ftls, always
heading toward a tree at (0, 0) on the west bank directly
across from the dog' s starting point. (a) If U = 2 ftls,
show that the dog reaches the tree. (b) If U = 4 f/s,
show that the dog reaches instead the point on the west
bank 50 f north of the tree. (c) If U = 6 ftls, show that
the dog never reaches the west bank.
72. In the calculus of plane curves, one lears that the ..
.k of the curve .= .(.) at the point ..is given
by

(.)

k =

[ 1 + . (.)

P
/

and that the curvature of a circle of radius is k = 1 I


[See Example 3 i n Section 1 1 . 6 of Edwards and Penney,
................7th edition (Upper Sad
dle River, NJ: Prentice Hall, 2008). ] Conversely, substi
tute = . to derive a general solution of the second-order
diferential equation
(with constant) in the form
Thus a circle of radius (or a part thereof) is the ..plane
curve with constant curvature l l .

n Sectlon l . 4 we lntroducedthe exponentla| dlfferentla| equatlonai,a = /i.


wlth so|utlon i, ; = i,e

, as a mathematlca| mode| for natura| popu|atlon


growth that occurs as a resu|t ofconstantblrth and death rates Here we present
amoregenera|popu|atlonmode|thataccommodatesblrthanddeathratesthatare
not necessarl|yconstant As before, however, ourpopu|atlonfunctlon i,;wl||be
a.o/osapproxlmatlontotheactua| popu|atlon,whlchofcoursechangeson|y
bylntegra|lncrementsthatls, byoneblrthordeathatatlme
Suppose that the popu|atlon changes on|y by the occurrence of blrths and
deathsthere ls no lmmlgratlon or emlgratlon from outslde the country or envl-
ronment underconslderatlon It ls customary to trackthegrowthor dec|lne ofa
popu|atlonln terms oflts///meandae./mefunctlonsdennedasfo||ows.
;,;ls thenumberofblrthsperunltofpopu|atlonperunltoftlmeattlme ,
,;ls thenumberofdeathsperunltofpopu|atlonperunltoftlmeattlme
Then the numbers ofblrths and deaths that occur durlng the tlme lnterva|
.+ } ls glven(approxlmate|y)by
blrths. ;,; i,; . deaths. ()

i,;


Hencethechangeilnthepopu|atlondurlngthetlmelnterva| . +} of
|ength ls
i= blrths- deaths ;,;

i,;

- ,;

i,;

I . / Popul ati on Model s 75
so
AP
At
f(t)- o (t ) ] P(t ) .
Theerrorl nthlsapproxlmatlonshou|dapproachzeroas At- 0, sotaklng
the|lmltwegetthedlerentla| equatlon
dP
dt
=(f - o) P, ( l )
l nwhlchwewrlte f = f(t ) , o = o (t ) , and P = P(t ) forbrevlty. Equatlon( l ) ls
thegeneral population equation. Iff ando areconstant, Eq. ( l ) reducestothe
natura|growthequatlonwlth/=f - o . Butlt a|solnc|udestheposslbl|ltythat
and o are varlab|efunctlonsoft . Theblrth anddeathrates need notbeknowa la
advance,theymay we|| dependontheunknownfunctlon P(t)
Supposethatana||lgatorpopu|atlonnumbers l 00lnltla||y,andthatltsdeathratels
o =0(sononeofthea||lgators ls dylng) . Iftheblrthratels f = (0. 0005) Paad
thuslncreasesasthepopu|atlondoesthenEq. ( l ) glvesthelnltla|va|ueprob|em
_=(0. 0005) P
:
, P(0) = l 00
(wltht lnyears). Then upon separatlngthevarlab|esweget

:
dP =(0. 0005)dt ,
I
=(0. 0005) t +c
P
Substltutlonoft =0, P = I 00glves c=- lJI 00,andthenwereadl|y so|vefor
2000
P(t ) = -.
20- t
Forlnstance, P( I 0) = 2000J l 0= 200, soafter l 0years thea||lgatorpopu-
|atlon has doub|ed. Butwe see that P - +c as t - 20, so area| popu|atlon
exp|oslonoccursln20years. Indeed,thedlrectlonne|dandso|utloncurvesshown
lnFlg. l . 7. I lndlcatethatapopu|atlonexp|oslona|ways occurs, whateverthe slze
ofthe (posltlve) lnltla| popu|atlon P(0) = P,. In partlcu|ar, lt appears thatthe
popu|atlona|ways becomesunboundedlna]/teperlodoftlme.
o t:
o I 0 Z0 J0 40 a0
FIGUR 1.7. 1. Slope feld and solution curves for the equation
dP/dt = (O. 0005) P
2
in Example 1 .
76 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| eZ
Bounded Populations and the Logistic Equation
Insltuatlonsasdlverseasthehumanpopu|atlonofanatlonandafrultHypopu|atloa
lnac|osed contalner, ltlsoenobservedthattheblrthratedecreasesasthepopu-
|atlon ltse|flncreases Thereasonsmayrangefrom lncreasedsclentlncorcu|tura|
sophlstlcatlon to a |lmltedfood supp|y Suppose, forexamp|e, that theblrthrate
; ls a//e.decreaslngfunctlonofthepopu|atlonslze i. sothat; = ;

- ;, i.
where;

and;
.
areposltlveconstants I fthedeathrate =

remalnsconstant,
thenEq , i ; takestheform
ai
a
=,;

- ;,i-

; i,
thatl s,

where.=;

and/=;,
ai

- = . i - /i
a

(2)
Ifthecoefnclents.and/arebothposltlve,thenEq (2)lsca||edthelogistic
equation. For the purposeofre|atlng the behavlor ofthe popu|atlon i,;to the
va|ues oftheparameterslntheequatlon,ltlsusefu|torewrltethe|oglstlcequatlon
lntheform
ai
a
=/ i,M- i; .
where/=/andM=.,/areconstants
(3)
InExamp|e:ofSectlon l 3 weexp|oredgraphlca||y apopu|atlonthatlsmode|ed
bythe|ogl stlcequatlon
ai
- = :i, i - i; = :i- :r

a
,:;
Toso|vethl sdlerentla|equatlonsymbo|lca||y, weseparatethevarlab|esandlnte
grate Weget
i, i

- i;
=

: a.
i


+
i
i
.
_ ,
ai=

: a partla|fractlons] ,
ln i - ln i - i = :+c.
i

=+e

=se

wheres=+e

}
i - i
If we substltute = and i= i,= i lnto thl s |ast equatlon, we nnd that
s= i,} , i - i,; Hence
i
i - i
i,e

i - i,
Flna||y,thlsequatlonls easy toso|veforthepopu|atlon
i,; =
i i,
i,+, i - i,;e-

,;

300

..
25 50 75 1 00
FIGURE 1.7.2. Typical solution
crs for the logistic euation
P
-
O. 06P - O. 0004P .
F

/:
FIGURE 1.7.3. Typical solution
curves for the logi stic equation
= k P(M - P) . Each solution
curve that starts below the line
P = Mj has an infection point
H this line. (See Problem 34. )
cXump| e
I . / Popul ati on Model s 77
attlmet lntermsofthelnltla|popu|atlon P,
-
P(0) . Flgure l . 7. 2showsanumber
ofso|utloncurvescorrespondlngtodlerentva|uesofthelnltla|popu|atlonranglng
from P, = 20to P, = 300 Notethata||theseso|utlon curves appeartoapproach
the hori zonta| |lne P = l 50 as an asymptote Indeed, you shou|dbe ab|e to see
dlrect|ynomEq (5)that|lm,,_ P(t )
-
l 50,whateverthelnltla|va|ue P, > 0.
Limiting Populations and Carryng Capacity
The nnlte |lmltlngpopu|atlon noted lnExamp|e2 lscharacterl stlcof|oglstlcpop
u|atlons InProb|em 32 weask youto use the methodofso|utlonofExamp|e2to
showthattheso|utlonofthe|oglstlclnltla|va|ueprob|em
ls
dP

-
kP( M - P) , P(0)
-
P,
MP,
P) =
P,+ (M P
a
) e''
,6)
(7)
Actua| anlma| popu|atlons are posltlve va|ued If P,
-
M, then(7)reduces
to the unchanglng (constant-va|ued) equl|lbrlum popu|atlon P(t ) = M Other-
wlse, the behavlorofa |oglstlc popu|atlon depends on whether0 < P, < M or
P, > M If0 < P, < M,thenweseefrom (6)and(7)that P' > 0and
P(t )
-
MP,
P,+ (M P,)e''
MP, MP,
< M
P,+ pos number P,
However,lfP, > M, thenweseefrom(6)and(7)that P' < 0and
P)
-
MP,
P, +(M- P
a
) e''
MP, MP,
= >
-
M.
P, + neg number P,
Inelthercase,theposltlvenumberornegatlvenumberlnthedenomlnatorhas
abso|ute va|ue |essthan P, andbecauseoftheexponentla|factorapproaches0
ast +c. Itfo||owsthat

MP,
hm P(t )
- -
M.
l
P,+0
,S)
Thus a popu|atlon that satlsnesthe|oglstlcequatlon does notgrow wlthoct
bound |lke a natura||y growlng popu|atlon mode|ed by the exponentla| equatlon
P' -
kP. Instead, ltapproaches the nnlte limiting population M as t +o.
As l||ustratedbythe typlca| |ogl stlc so|utlon curves ln Flg l 7 3, the popu|atlon
P(t ) steadl|ylncreasesandapproaches Mfrombe|owlf0 < P, < M, butsteadl|y
decreases and approaches M fromabove lf P, > M Sometlmes M lsca||edthe
carrying capacity oftheenvlronment,conslderlnglttobethemaxlmumpopu|atlon
thattheenvlronmentcansupportona|ong-termbasls
Supposethat ln l SS5 the popu|atlon ofa certaln country was 50 ml||lonand was
growlng at the rate of750, 000 peop|e per year at that tlme Suppose a|so that ln
l 940ltspopu|atlonwas l 00ml||lon and wasthengrowlngat the rate of l ml||lon
peryear Assumethatthl spopu|atlonsatl snesthe|oglstlcequatlonDetermlneboth
the|lmltlngpopu|atlonM andthepredlctedpopu|atlonfortheyear2000
78 Chapter I Fi rst-Order Di fferenti al Equati ons
Sol ution We substltutethetwoglvenpalrsofdatalnEq. (3)andnndthat
cXump| e4
cXump| e
0. 75=50k( M- 50) , I . 00= l 00k(M- l 00) .
Weso|veslmu|taneous|yfor M =200andk =0. 000 l . Thus the|lmltlngpopu|a-
tlon ofthecountry lnquestlon ls 200ml||lon. Wlth theseva|uesofM andk, and
wlth = 0 correspondlng to the year l 940 (ln whlch i, = l 00), we nnd that
accordlngtoEq. (7)thepopu|atlonlnthe year 2000 wl||be
l 00 200
i(60) =
l 00+
(200 - i e-
,

.
about l 53. 7ml||lonpeop|e.
Historical Note

The |oglstlcequatlonwas lntroduced(around l S40)bytheBe|glanmathematlclan


and demographerP. Verhu|stas aposslb|emode|forhumanpopu|atlongrowth.
Inthenexttwoexamp|eswecomparenatura| growthand|oglstlcmode| nts tothe
l 9th-century U. S. popu|atlon census data, then compare proectlons forthe 20th
century.
TheU. S. popu|atlonln l S00was5. 30Sml||lonandln I 900was76. 2l 2 ml||lon. If
wetakei,=5. 30S(wlth=0ln I S00)lnthenatura|growthmode| i() = i,e
and substltute= l 00,i=76. 2l 2,wenndthat
l s
76. 2I 2= se
,

.
l 76. 2l 2
so =
l 00
ln
5. 30S
0. 026643.
Thusournatura|growthmode|fortheU. S. popu|atlondurlngthel 9thcentury
i() =, s


(9)
(wlthlnyearsand ilnml||lons). Becausee

l . 02700,theaveragepopu-
|atlongrowthbetween I S00and l 900wasabout2. 7peryear.
The U. S. popu|atlonln I S50was 23. l 92ml||lon. Ifwetake i, = 5. 30Sandsub-
stltutethedatapalrs=50, i =23. I 92(forI S50)and= l 00, i =76. 2l 2(for
I 900)lnthe|oglstlcmode|formu|alnEq. (7), wegetthetwoequatlons
(5. 30S) M
=23. l 92
5. 30S
+
(M - 5. 30S)e
a
'

(5. 30S) M
=76. 2l 2
5. 30S+ ( M- s)e-
,

,
( I 0)
ln the two unknowns k and M. Non|lnear systems |lke thls ordlnarl|y are so|ved
numerlca||y uslng an approprlate computer system. But wlth the rlght a|gebralc
trlck(Prob|em36 lnthlssectlon)theequatlonsln( l 0) canbeso|vedmanua||yfor
k =0. 000I 677 I 6, M = l SS. l 2 l . Substltutlonoftheseva|ues ln Eq. (7)yle|dsthe
|oglstlcmode|
99S. 546
i,;
-

5. 30S+
( l S2 S l 3) e-
1 1

( l l )
c-
40%
Lxoaeat|a|
Lo,|st|;
...
-40%
FIGURE 1.7.5. Percentage
errors in the exponential and
logistic population models for
1800-1 950.
I . / Popul ati on Model s 79
The tab|e lnFlg. l . 7. 4compares the actua| l S00l 990U S census popu|a-
tlon ngures wlth those predlcted by the exponentla| growth mode| ln (9) and the
|oglstlcmode| ln ( l l ) . Both agree we||wlththe l 9th-centuryngures But theex-
ponentla| mode| dlverges appreclab|y from the census data ln the ear|y decades
ofthe 20th century, whereas the |oglstlc mode| remalns accurate untl| l 940. By
the endofthe 20th century the exponentla| mode| vast|y overestlmates the actua|
U. S. popu|atlonpredlctlng over abl||lon ln the year 2000whereas the |oglstlc
mode|somewhatunderestlmateslt.
; : , ; :ponential Logtic Logtic
' Error Model Error
1 800 5. 308 5. 308 0. 000 5. 308 0.000
1 8 1 0 7. 240 6. 929 0. 3 1 1 7. 202 0.038
1 820 9. 638 9. 044 0. 594 9. 735 -0. 097
1 830 1 2. 861 1 1 . 805 1 . 056 1 3.095 -0. 234
1 840 1 7. 064 1 5.409 1 . 655 1 7. 501 -0. 437
1 850 23. 1 92 20. 1 1 3 3. 079 23. 1 92 0.000
1 860 3 1 .443 26. 253 5 . 1 90 30. 405 1 . 038
1 870 38. 558 34. 268 4. 290 39. 326 -0. 768
1 880 50. 1 89 44. 730 5. 459 50. 034 0. 1 55
1 890 62. 980 58. 387 4. 593 62. 435 0. 545
1 900 76. 21 2 76. 21 2 0. 000 76. 21 3 -0. 001
1 91 0 92. 228 99.479 -7. 25 1 90. 834 1 . 394
1 920 1 06. 022 1 29. 849 -23. 827 1 05. 61 2 0.410
1 930 1 23. 203 1 69.492 -46. 289 1 1 9. 834 3. 369
1 940 1 32. 1 65 221 . 237 -89. 072 1 32. 886 -0. 721
1 950 1 5 1 . 326 288. 780 -1 37. 454 1 44. 354 6. 972
1 960 1 79. 323 376. 943 -1 97. 620 1 54. 052 25. 271
1 970 203. 302 492. 023 -288. 721 1 61 . 990 41 . 3 1 2
1 980 226. 542 642. 236 -41 5. 694 1 68. 3 1 6 58.226
1 990 248. 7 1 0 838. 308 -589. 598 1 73. 252 76. 458
2000 28 1 .422 1 094. 240 -81 2. 8 1 8 1 77. 038 1 04. 384
FIGURE 1.7.4. Comparison of exponential growth and logistic models with U. S. census
populations (in millions).
Thetwomode|s are compared ln Flg. l . 7 5, where p|ots ofthelrrespectlve
errorsas ape.e.,e ofthe actua| popu|atlonare shown for the l S00l 950
perlod. Weseethatthe|oglstlcmode|trackstheactua|popu|atlonreasonab|ywe||
throughout thls l 50-year perlod. However, the exponentla| error ls conslderab|y
|argerdurlngthe l 9thcenturyand|ltera||ygoesothechartdurlngthenrstha|fof
the20th century.
Inordertomeasuretheextenttowhlchaglvenmode|ntsactua|data,ltlscus-
tomarytodennetheaverage error (lnthe mode|)as/es,.eoo)/e..em,e
o)/es,.eso)/e/a/./a./eos(the |atterappearlng lnthefourthand slxth
co|umnsofthetab|elnFlg. l . 7. 4) Uslngon|y the l S00l 900data, thlsdennltlon
glves 3 . l 62fortheaverageerrorlntheexponentla| mode| , whl|etheaverageerror
lnthe|oglstlcmode|ls on|y0. 452. Consequent|y,even ln l 900wemlghtwe||have
antlclpatedthatthe|oglstlcmode|wou|dpredlcttheU. S. popu|atlongrowthdurlng
the20thcenturymoreaccurate|ythantheexponentla|mode|.
80 Chapter I Fi rst-Order Di fferenti al Equati ons
cXump| e
Themora|ofExamp|es4and5ls slmp|ythatone shou|dnotexpecttoomuch
ofmode|sthatarebasedonsevere|y|lmltedlnformatlon(suchasustapalrofdata
polnts). Muchofthe sclenceofs./s/.s ls devoted to the ana|yslsof|arge data
setstoformu|ateusefu|(andperhaps re|lab|e)mathematlca|mode|s.
More Applications of the Logistic Equation
Wenextdescrlbe some sltuatlonsthat l||ustratethevarledclrcumstances lnwhlch
the |oglstlcequatlonlsasatlsfactorymathematlca|mode|.
1. i/-/eae./-es/./o A certalnenvlronmentcansupportapopu|a-
tlonofatmostM lndlvldua|s. Itlsthenreasonab|e toexpectthegrowthrate
f - (the comblnedblrth and death rates)tobeproportlona|to M - i, be-
causewe may thlnkofM - i as the potentla| for furtherexpanslon. Then
f -
-
k(M - i; , sothat
ai
a
-
(f- ) i
-
kP(M - i;
Thec|asslcexamp|eofa|lmltedenvlronmentsltuatlonl s afrultypopu|atlon
lnac|osedcontalner.
2. co-pe//os/./o Ifthe blrth rate f ls constant but the deathrate ls
proportlona|to i,sothat=oi,then
ai

-
(f - oi) i
-
ki(M- i;
a
Thls mlghtbeareasonab|e worklng hypothesls lna study ofacannlba|lstlc
popu|atlon, lnwhlch a|| deaths resu|tfrom chanceencounters betweenlndl-
vldua| s. Ofcourse, competltlonbetweenlndlvldua|slsnotusua||ysodead|y,
norltseectssolmmedlateanddeclslve.
J. 1o/ppo/os/./o Let i,; denote the number of lndlvldua|s ln a
constant-slzesusceptlb|epopu|atlon M who are lnfectedwltha certaln con-
taglous and lncurab|e dlsease. The dlsease ls spread by chance encounters.
Then P',;shou|dbeproportlona|totheproductofthenumberioflndlvld-
ua|s havlng the dlsease and the number M - i ofthose not havlng lt, and
therefore ai}a
-
ki(M - i; Agaln we dlscover that the mathematlca|
mode| ls the|ogl stlcequatlon. Themathematlca|descrlptlonofthe spreadof
arumorlnapopu|atlonofMlndlvldua|slsldentlca| .
. . . . . .
Supposethatattlme = c, i cthousandpeop|elnacltywlthpopu|atlonM
-
i cc
thousand peop|e have heard a certaln rumor After i week the number i,; of
thosewhohaveheardlthaslncreasedto i, i ;
-
zcthousand. Assumlngthati,;
satlsnesa|oglstlcequatlon,whenwl||scoftheclty' spopu|atlonhaveheard the
rumor!
Sol ution Substltutlng i, = icandM
-
i cc(thousand)lnEq.(7), weget
i ccc
i,;
-
, i z;
i c

ce-
,

Then substltutlonof
-
i , i
-
zcglvestheequatlon
zc=
i ccc
i c

ce-
,

Exampl e
I . / Popul ati on Model s 81
thatlsreadl|yso|vedfor
e
-
.


so k
,
_ln c ccs i c
Wlth P( ) sc, Eq. ( i 2)takesthefom
sc
i ccc
i c

ce-
.

whlchwe so|vefore
-
,

Itfo||ows that scofthepopu|atlonhasheard


therumorwhen
ln 36 ln 36
4. 42,
l 00k ln_
thus afterabout4weeksand3 days.
Doomsday versus Extinction

Conslderapopu|atlonP,;ofunsophlstlcatedanlma|slnwhlchfema|esrelyso|e|y
onchanceencounters to meetma|esforreproductlvepurposes. Itlsreasonab|eto
expectsuchencounters to occurat aratethatlsproportlona|to theproductofthe
number P J2ofma|esandthenumber PJoffema|es, henceat arateproportlona|
to P
z
. We therefore assumethat blrths occur at the rate kP
z
(per unlt tlme, wlth
k constant). Theblrth rate (blrthsJtlmeJpopu|atlon) ls then glven by f kP If
the death rate o ls constant, then the genera| popu|atlonequatlon ln , i ; yle|dsthe
dlerentla|equatlon
dP
z
kP - o PkP( P- M;
a
(whereMoJk > c;asamathematlca|mode|ofthepopu|atlon.
( I 3)
Note thattherlght-hand slde lnEq. ( I 3) ls thee,./.eoftherlght-handslde
lnthe|ogl stlcequatlonln (3). Wewl|| see thattheconstantMls now athreshold
population, wlththewaythepopu|atlonbehaveslnthefuturedependlngcrltlcal|y
onwhetherthelnltla|popu|atlonP,ls|essthanorgreaterthanM
Conslderananlma|popu|atlon P(t ) thatls mode|edbytheequatlon
dP
z
0. 0004P( P- i c;0. 0004P - 0. 06P.
dt
We wanttonndP(t ) lf(a)P(0) zcc, (b)P(0) i cc
,i:;
Sol uti on To so|vetheequatlonln( I 4), weseparatethevarlab|esandlntegrate. We get
P( P i c;

0. 0004 dt ,
-

-
I
dP

0. 0004 dt partla|fractlons] ,
i c P P - i c
ln P - ln P- i c -0. 06t

c.

- +e

e
-

se
-

P - i c
where s+e

} , i ;
82 Chapter I Fi rst-Order Di fferenti al Equati ons
P


FIGURE 1.7.6. Typical solution
curves for the explosion/extinction
equation = . M) .
Problems
(a) Substltutlonof=candi=zcclnto( I 5)glvess=:Wlththlsvalueofs
weso|veEq. ( I 5)for
:cce

i, ; =
:e-

- I
( I :)
Note that, as lncreases and approaches = |a,:; }c c: z i c, the posltlve
denomlnatorontherlghtln( I :)decreasesandapproaches0. Consequent|yi,;
+~as

Thlsls aaoo-sa.ysltuatlonarea|popu|atlonexp/os/o
(b) Substltutlonof =cand i = i cclnto ( I 5) glves s= -zWlththlsva|ue of
swe so|veEq. ( I 5)for
cce
-

cc
i, ;

z

+I

z+e


( I 7)
Note that, as lncreases wlthout bound, the posltlve denomlnator on the rlght ln
, i :;approaches+~ Consequent|y, i,; cas +~ Thlslsan(eventual)
ex/./osltuatlon.
Thusthepopu|atlonlnExamp|e7eltherexplodesorlsanendangeredspecies
threatenedwlthextlnctlon,dependlngonwhetherornotltslnltla| slzeexceedsthe
thresho|d popu|atlon M = I 50. An approxlmation to thls phenomenon ls some-
tlmesobservedwlthanlma|popu|atlons,suchastheal|lgatorpopu|atlonlncertaln
areasofthesouthernUnltedStates.
Flgure I . 7. 6showstyplca| so|utloncurvesthatl||ustratethetwoposslbllltles
for apopu|atlon i,; satlsfylngEq. , i ; If i, = M (exactly' ) , then thepopu|a-
tlon remalns constant. However, thls equl|lbrlum sltuatlon lsveryunstab|e If i,
exceeds M (even s|lght|y), then i,; rapldly lncreases wlthoutbound, whereas lf
thelnltla| (posltlve)popu|atlonls|essthanM (howevers|lghtly), thenltdecreases
(moregradua||y)towardzeroas +~SeeProb|em
,..........,.........
..,...1-8. ........
.,....., .......,...
........, ..,...........
.....,......
the population numbers 1 00 rabbits and is increasing at
the rate of 20 rabbits per month. How many rabbits will
there be one year later?
10. Suppose that the fsh population in a lake is attacked
by a disease at time = 0, with the result that the fsh
cease to reproduce (so that the birth rate is f = 0) and the
death rate (deaths per week per fsh) is thereafer propor
tional to 1/
..
. If there were initially 900 fsh in the lake
and ..were lef afer 6 weeks, how long did it take all
the fsh in the lake to die?
..
1. - = . .

.= 2 2.
.

..
. - = 1 - .

.= .
.
..
5. - = . . .= 8
.
..
6. - = .. .= 2
.
..
7. - = .. . .= 1 1
.
..
8. = ..- 1 3) , .= 1
.
..
- = . .

X (0) = 1
.
..
- = 9 - 4x
2
, x (0) = 0
.
9. The time rate of change of a rabbit popUlation is pro
portional to the square root of At time = 0 (months)
11. Suppose that when a certain lake is stocked with fsh, the
birth and death rates f and are both inversely propor
tional to
..
. (a) Show that
where .is a constant. (b) If ,= 1 00 and afer 6 months
there are 1 69 fsh in the lake, how many will there be afer
1 year?
12. The time rate of change of an alligator population in
a swamp is proportional to the square of The swamp
contained a dozen alligators in 1 988, two dozen in 1 998.
When wi l l there be four dozen alligators in the swamp?
What happens thereafer?
13. Consider a prol ifc breed of rabbits whose birth and death
rates, f and , are each proportional to the rabbit popula
tion = ( , with f ~ . (a) Show that

-
.constant.
-
1 - . ,
'
Note that +0 as ., This is dooms
day. (b) Suppose that = 6 and that there are nine
rabbits afer ten months. When does doomsday occur?
14. Repeat part (a) of Problem 1 3 in the case f . What
now happens to the rabbit population in the long run?
15. Consider a population satisfying the logistic equa
tion ..= .

where .= . is the time rate


at which births occur and = 2 is the rate at which
deaths occur. If the initial population is = , and
.births per month and ,deaths per month are occur
ring at time = 0, show that the li miting population is
M = .
16. Consider a rabbit population satisfying the logistic
equation as in Problem 1 5. If the initial population is 1 20
rabbits and there are 8 births per month and 6 deaths per
month occurring at time = 0, how many months does it
take for to reach 95% of the limiting population M?
17. Consider a rabbit population satisfying the logistic
equation as in Problem 1 5. If the initial population is 240
rabbits and there are 9 births per month and 12 deaths per
month occurring at time = 0, how many months does it
take for to reach 1 05% of the limiting population M?
18. Consider a population sati sfying the extinction
explosion equation ./.= .

- bP, where .= .

is the time rate at which births occur and = is


the rate at which deaths occur. If the initial population
is = ,and .,births per month and ,deaths per
month are occurring at time = 0, show that the threshold
population is M = , ,.,
19. Consider an alligator population pet) satisfying the
extinction/explosion equation as in Problem 1 8. If the ini
tial population is 1 00 alligators and there are 10 births per
month and 9 deaths per months occurring at time = 0,
how many months does it take for to reach 10 times
the threshold population M?
20. Consider an alligator population satisfying the
extinction/explosion equation as in Problem 1 8. If the ini
tial population is 1 1 0 alligators and there are 1 1 births per
month and 12 deaths per month occurring at time = 0,
how many months does it take for to reach 1 0% of
the threshold population M?
21. Suppose that the population of a country sati sfes the
differential equation ./.= . with .con
stant. Its population in 1 940 was 1 00 million and was then
I . / Popul ati on Model s 83
growing at the rate of 1 million per year. Predict this coun
try' s population for the year 2000.
22. Suppose that at time = 0, half of a "logistic" popula
tion of 1 00, 000 persons have heard a certain rumor, and
that the number of those who have heard it is then increas
ing at the rate of 1 000 persons per day. How long will it
take for this rumor to spread to 80% of the population?
....Find the value of .by substituting and
in the logistic equation, Eq. (3). )
23. As the sal t KN03 dissolves in methanol , the number x(t)
of grams of the salt in a solution afer seconds satisfes
the differential equation ...= 0. 8x - 0. 004x2.
(a) What is the maximum amount of the salt that will ever
dissolve in the methanol ?
(b) If x = 50 when = 0, how long will it take for an
additional 50 g of salt to dissolve?
24. Suppose that a community contains 1 5,000 people who
are susceptible to Michaud' s syndrome, a contagious dis
ease. At time = 0 the number of people who have
developed Michaud' s syndrome is 5000 and is increasing
at the rate of 500 per day. Assume that is propor
tional to the product of the numbers of those who have
caught the disease and of those who have not. How long
will it take for another 5000 people to develop Michaud
'
s
syndrome?
25. The data in the table in Fig. 1 . 7. 7 are given for a certain
population that sati sfes the logistic equation in (3).
(a) What is the limiting population M? ....Use
the approximation
+ h) - h)

2h
with h = 1 to estimate the values of when
25. 00 and when = 47. 54. Then substitute these values
in the logistic equation and solve for .and M. ) (b) Use
the values of .and M found in part (a) to determine when
= 75. .... Take = 0 to correspond to the
year 1 925. )
Year (milli0ns)
1 924 24. 63
1 925 25.00
1 926 25. 38
1 974 47. 04
1 975 47. 54
1 976 48.04
FIGURE 1.7.7. Population data for Problem 25.
26. A population of small rodents has birth rate f
(0. 001 )(births per month per rodent) and ..death
rate . If = 1 00 and 8, how long (in
84 Chapter I Fi rst-Order Di fferenti al Equati ons
months) wi l l it take this population to double to ro
dents? ....First fnd the value of . )
27. Consider an animal population with constant death
rate (deaths per animal per month) and with
birth rate proportional to Suppose that
and = (a) When i s (b) When does
doomsday occur?
28. Suppose that the number .(with in months) of alliga
tors in a swamp satisfes the diferential equation ...=
.

.
(a) I f initially there are alligators in the swamp, solve
this diferential equation to determine what happens
to the alligator population in the long run.
(b) Repeat part (a), except with alligators initially.
29. During the period from to the u. S. population
in years) grew from million to million.
Throughout this period, remained close to the solu
tion of the initial value problem
.
_ .


(a) What population does this logistic equation pre
dict?
(b) What limiting population does it predict?
(c) Has this logistic equation continued since to ac-
curately model the U. S. population?
[This problem is based on a computation by Verhulst, who
in .used the .U. S. population data to pre
dict accurately the U. S. population through the year
(long afer hi s own death, of course) . ]
30. A tumor may be regarded as a population of multiplying
cells. It is found empirically that the "birth rate" of the
cells in a tumor decreases exponentially with time, so that
,(where .and

are positive constants) ,


and hence
.
_ ,
,

Solve this initial value problem for


, exp

,

Observe that ) approaches the fnite limiting popula


tion ,exp ,.as +0.
31. For the tumor of Problem suppose that at time
there are ,

cells and that is then increasing


at the rate of X

cells per month. Afer months the


tumor has doubled (in size and in number of cell s). Solve
numerically for .and then fnd the limiting population of
the tumor.
32. Derive the solution
)
,
,+ ,


of the logistic initial value problem .
() Make i t clear how your derivation depends on
whether or , ~
33. (a) Derive the solution
,

,+ ( ,


of the extinction-explosion initial value problem
. ) (
(b) How does the behavior of ( as increases depend
on whether , or ,~
34. If satisfes the logistic equation in )use the chain
rule to show that
() .

) (
Conclude that ~ if

if

if

and ~

if ~ In particular, it follows that any solution


curve that crosses the line has an infection point
where it crosses that line, and therefore resembles one of
the lower S-shaped curves in Fig.
35. Consider two population functions and

( ) , both
of which satisfy the logistic equation with the same limit
ing population but with different values kl and .

of the
constant .in Eq. () Assume that . .

Which pop
ulation approaches the most rapidly? You can reason
..by examining slope felds (especially if ap
propriate sofware is available), ..by analyzing
the solution given in Eq. ()or ..by substitut
ing successive values of
36. To solve the two equations i n ( ) for the values of .and
begin by solving the frst equation for the quantity
.

and the second equation for .

Upon equating the two resulting expressions for .

in
terms of you get an equation that is readily solved for
With now known, either of the original equations
is readily solved for .This technique can be used to "ft"
the logistic equation to any three population values ,
and

corresponding to ,...,..times ,
and


37. Use the method of Problem to ft the logistic equation
to the actual U. S. population data (Fig. .)for the years
and Solve the resulting logistic equa
tion and compare the predicted and actual populations for
the years and
38. Fit the logistic equation to the actual U. S. population data
(Fig. .)for the years and Solve the
resulting logistic equation, then compare the predicted and
actual populations for the years and
39. Birth and death rates of animal populations typically are
not constant; instead, they vary periodically with the pas
sage of seasons. Find if the population satisfes
the differential equation
.
- = .+ cos
.
I . d Accel erati on-Vel ocity Model s 85
where t is in years and .and are positive constants. Thus
the growth-rate function r (t ) = .+ cos 2:t varies pe
riodically about its mean value .Construct a graph that
contrasts the growth of this population with one that has
the same initial value Po but satisfes the natural growth
equation P' = k P (same constant . How would the two
populations compare afer the passage of many years?
cXump| e 1
InSectlonI . 2wedlscussedvertlca|motlonofamass-nearthesurfaceoftheearth
under thelnuenceofconstantgravltatlona|acce|eratlon. Ifweneg|ectanyeects
ofalrreslstance,thenNewton' ssecond|aw , i=-.;lmp|lesthattheve|ocltytof
the mass-satlsnestheequatlon
ac
-- =
l
,
a

( I )
where i, = --_ l s the (downward-dlrected)force ofgravlty, wherethegravlta-
tlona|acce|eratlonls _ 9. SmJs
:
(ln mksunlts ,_ 32ftJs
:
lnfpsunlts).
. . . _ __ . . _ _ _ __ _ _._. _ _ " _m. _ _ __ _ _" N N _ _ _ _ __ _ _ _ . .. . ... . . . _ ____ h _ _ .
Supposethatacrossbowbo|tlsshotstralghtupwardfromtheground,y,=0) wlth
lnltla|ve|oclty:,=49(mJs). ThenEq. ( I ) wlth_ =9. Sglves
ac
- =-9. S, s o t (t ) =-(9. S) t +:,=-(9. S) t +49.
a
Hencethebo|t' shelghtfunctlony(t)lsglvenby
(t)=, -(9. S) t +49]a=-(4. 9)t
:
+49t+y,=-(4. 9)t
:
+49t .
The bo|treaches lts maxlmumhelght when t = -(9. S) t +49 = 0, hence when
t =5 (s). Thus lts maxlmumhelghtls
y,,=y,;=-(4. 9) (5
:
) +(49) (5) = I 22. 5(m) .
The bo|t retums tothe ground when = -(4. 9) t (t - I 0) = 0, andthus after I 0
secondsa|oft.
Nowwe want to take accountofalrreslstancelna prob|em |lkeExamp|e I .
The force i , exerted by alr reslstance on the movlng mass - must be added ln
Eq.( I ), sonow
ac
-- = i, + i,

a
(2)
Newtonshowedlnhlsi/./p/.M./e-./..thatcertalnslmp|ephyslca|assump-
tlons lmp|y that i,lsproportlona| to the s,.eofthe ve|oclty. i, = kt
:
. But
emplrlca|lnvestlgatlonslndlcatethattheactua| dependenceofalrreslstanceonve-
|ocltycanbequltecomp|lcated. Formanypurposesltsufncestoassumethat
where I p 2andtheva|ue ofk dependsontheslzeandshapeofthebody,as
we||asthedensltyandvlscosltyofthealr. Genera||yspeaklng,p=Iforre|atlve|y
Chapter I First-Order Di fferenti al Equati ons

(Note: F R acts upward when


the body is falling.)

Net force F FR -FG


Ground level
FIGUR 1.S. 1. Vertical motion
with air resistance.
|owspeedsand =2forhlghspeeds,whereas I < < 2forlntermedlatespeeds.
Buthows|ow|owspeedandhowfasthlghspeedaredependonthesamefactors
thatdeterminetheva|ueofthecoefnclent/
Thusalrreslstancel sacomp|lcatedphyslca|phenomenon. Buttheslmp|lfy-
ingassumptlonthati,lsexact|yoftheformglvenhere,wlthelther = l or =2,
yle|ds atractab|emathematlca| mode|thatexhlblts the mostimportantqua|ltatlve
featuresofmotlonwlthreslstance.
Resistance Proportional to Velocit
Letus nrstconslderthevertlca| motlon ofabodywlth mass- nearthesurface
ofthe earth, subecttotwoforces. adownwardgravltatlona|force i ,andaforce
i, ofalrreslstancethat ls proportlona| to ve|oclty (sothat = I ) andofcourse
dlrected opposite the dlrectlon ofmotlon ofthe body. Ifwe set up a coordlnate
systemwlththepositlve ,-dlrectlonupwardand wlth , = 0 atground|eve|, then
i,= --,and
i,=-/c, (3)
where/lsaposltlveconstantandc =d,Jdt lstheve|ocltyofthebody. Notethat
themlnusslgnlnEq. (3)makesi,posltlve(anupward force) lf thebody lsfa||lng
,cl snegatlve)andmakes i ,negatlve(adownwardforce)lfthebodylsrlslng,cls
posltlve). AslndlcatedlnFlg. I . S. I , thenetforceactlngonthebodylsthen
i=i,+i,=-/c- -,,
andNewton' s|awofmotlon i=-,ac}a;yle|dstheequatlon
Thus
ac
- = -/c - -,
a
ac
- =-,c - ,,
a
(4)
where,= /}- ~ 0. You shou|dverlfyforyourse|fthatlftheposltlve,-axlswere
dlrecteddownward,thenEq. (4)wou|dtaketheforma:}a=-,c+,
Equatlon(4)lsaseparab|enrst-orderdlerentla|equatlon,andltsso|utlonls
c, ; = c,+

e
-
.

-
Here, :,=c,c;lsthelnltla|ve|ocltyofthebody. Notethat
c= |lm c, ; =.

-
,
(5)
(6)
Thusthespeedofabodyfa||lngwithalrreslstancedoesolncreaselndennlte|y,
lnstead,itapproaches+]/e|lmltlngspeed,orterminal speed,
, -,

: = - =

, k
(7)
cXump| eZ
I .d Accel erati on-Vel ocity Model s 87
Thls fact l s what makes a parachute a practlca| lnventlon, lt even he|ps exp|aln
theoccaslona|survlva|ofpeop|ewhofa|| wlthoutparachutesfromhlgh-ylngalr-
p|anes.
WenowrewrlteEq. ,;lntheform
Integratlonglves
ay

( t, - c, ; e +t, .
a
i
y,;= --(c,- c, ; e

+c, +c
j
(S)
We substltutecforand|et)g = y(0) denotethelnltla|helghtofthebody. Thus
wenndthatc=)g +(t,- t, )Jj,andso
i
.

y, ; =)g +c, +-(t,- t, ) ( I - e


-
) .
j
,;
Equatlons,s;and ,;glvetheve|ocltycandhelghtyofabody movlngver-
tlca||yunderthe lnuence ofgravlty and alrreslstance. The formu|as depend on
the lnltla|helght)g ofthe body, lts lnltla| ve|oclty t,, andthedrg .oe./ej,
theconstantsuchthattheacce|eratlonduetoalrreslstancelsO
R
= -jt. Thetwo
equatlonsa|solnvo|vethetermlna|ve|oclty t, dennedln Eq. (6).
For aperson descendlng wlththealdofaparachute, a typlca| va|ueofj ls
i ,whlchcorrespondstoatermlna|speedof t, z i ftJs, orabouti : ml/h.
Wlthanunbuttonedovercoatapplnglnthewlndlnp|aceofaparachute,anun|ucky
skydlvermlghtlncreasej toperhaps as much asc , whlch glves atermlna|speed
ofc, 65ftJs,about::mlJh. SeeProb|ems i cand i i forsomeparachute-ump
computatlons.
....... ..... .........
We agaln conslder a bo|t shot stralght upward wlth lnltla| ve|oclty u, = :mJs
fromacrossbowatground|eve|. Butnowwetakealrreslstancelntoaccount,wlth
j = c c:ln Eq. ,:; We ask how the resu|tlng maxlmum helght and tlme a|oft
comparewlththeva|uesfoundlnExamp|e i
Sol uti on We substltute)g = c,v, = :, and c, = -g/j = -z:lnEqs. ,;and ,;,and
obtaln
c,;= z:e
-

- z:,
y,; =:c- z:- :ce

Tonndthe tlme requlredforthe bo|t to reach lts maxlmumhelght (when t =c;,


weso|vetheequatlon
c( ) =z:e
-

- z:=c
for = zln,z:}z:; : s(s). Its maxlmumhelghtls then)gg = c,;
i cs zscmeters (as opposedto i zz meterswlthoutalrreslstance). Tonnd when
thebo|tstrlkestheground,wemustso|vetheequatlon
y,; =:c- z:- :ce
-

=0.
UslngNewton' smethod,wecanbeglnwlththelnltla|guess, = icandcaout
thelteratlon
-., =
-
- y,
-
;}y ,
-
;togeneratesuccesslveapproxlmatlonstothe
root. Orwecanslmp|yusetheSolve commandonaca|cu|atororcomputer. We
88 Chapter I Fi rst-Order Di fferenti al Equati ons
nnd that the bo|t ls ln the alr for , 9. 4I I seconds (asopposedto icseconds
wlthoutalrreslstance). Ithltsthegroundwlthareducedspeedof

: ,,; 43. 227


mJs (asopposedtoltslnltla|ve|ocltyof49mJs).
Thustheeectofalrreslstancelstodecreasethebo|t' smaxlmumhelght,the
tota| tlme spenta|oft, andlts nna|lmpactspeed. Notea|sothatthebo|tnowspends
moretlmelndescent,,- 4. S53s)thanlnascent, 4. 55Ss).
Resistance Proportional to Square of Velocit
Nowweassumethattheforceofalrreslstancelsproportlona|tothes,.eofthe
ve|oclty.
, i c;
wlth / > 0. The cholceofslgns here dependson the dlrectlon ofmotlon, whlch
theforceofreslstancea|waysopposes. Taklngtheposltlve,-dlrectlonas upward,
i, < cforupward motlon (when c > c; whl|e i, > cfordownward motlon
(when c < c; Thusthe slgn ofi,lsa|waysopposltethatofc.sowecanrewrlte
Eq. , i c;as
i,=-/c c
ThenNewton' ssecond|awglves
ac
-- =i, +i, m = /c c .
a
thatls,
ac
= - ,c c
a

, i c;
, i i ;
where, =/}- > 0. We mustdlscussthecasesofupwardanddownwardmotlon
separate|y.
UPWARD MOTION: Supposethataproectl|els|aunched stralghtupwardfrom
the lnltla|posltlon)g wlth lnltla|ve|oclty 0g > 0. ThenEq. ( I I ) wlth c > cglves
thedlerentla|equatlon
= - ,c
:
= i+c
:
( I 2)
In Prob|em I 3 we ask you to make the substltutlon = c, jJg and app|y the
faml|larlntegra|
,
i

a =tan
|
+C
i +
toderlvetheproectl|e' sve|ocltyfunctlon
0 ( ) = _tan C = _ ) wlth C, =tan
,
cg_. , i ;
Because ]tana== lncos+C, a second lntegratlon (see Prob|em I4)
yle|dstheposltlonfunctlon
i cosC, - )
y, ; =)g + ln
, cosC
|
( I 4)
cXump| e
I . d Accel erati on-Vel ocity Model s 89
DOWNWARD MOTION: Suppose that a proectl|e ls |aunched ,or dropped)
stralght downward from the lnltla| posltlon y, wlth lnltla| ve|oclty :, 0. Then
Eq. ( I I ) wlthc < 0glvesthedlerentla|equatlon
ac
:
,
:
a
=-,+,c =-, I -
_
: , I 5)
In Prob|em I 5 we ask you to make the substltutlon = c,,}, and app|y the
lntegra|

:
a =tanh
|
+C
i -
toderlvetheproectl|e' sve|ocltyfunctlon
c,;=tanhC
:
) wlth , I 6)
Because,tanha=lncosh+C, anotherlntegratlon,Prob|em l 6)yle|dsthe
posltlonfunctlon
I coshC
:
)
y,;=y, - ln

, cosh C
:
, I 7)
(Note the ana|ogybetween Eqs. ( I 6) and , I 7) and Eqs. ( I 3) and ( I4)forupward
motlon. )
If:,=0,thenC
:
=0,soc,;=-, ,} ,tanh ) Because

slnhx

|
,,
-
,

,
hmtanh x = hm = hm
|
= I ,
.- .-
cosh x .-
,,
+
,

,
ltfo||ows that ln the case ofdownward motlon the body approaches the temlna|
speed
, I S)
(ascomparedwlth c

=,},lnthecaseofdownwardmotlonwlth|lnearreslstance
descrlbedbyEq. (4)).
Weconslderoncemoreabo|tshotstralghtupwardwlthlnltla|ve|oclty:,=49m/s
fromacrossbowatground|eve|,aslnExamp|e2. Butnowweassumealrreslstance
proportlona|tothesquareoftheve|oclty, wlth,=0. 00I I lnEqs. ( I 2)and( I 5). In
Prob|ems I 7and I S weask you toverlfytheentrleslnthe|ast|lneofthefo||owlng
tab|e.
Desen
0d
.
. . . . .

. . . . . .
90 Chapter I Fi rst-Order Di fferenti al Equati ons
)
. za
. aa
sa
-a
a
za
w..-...
FIGUR I.S.2. The height fnctions in Example I (without air
resistance), Example (with linea air resistance), and Example 3
(with air resistance proportional to the squae of the velocity) ae all
plotted. The graphs of the latter two visually indistinguishable.
Comparlson ofthe |ast two |lnes ofdatahere lndlcates |ltt|e dlerencefor the
motlonofourcrossbowbo|tbetween|lnearalrreslstance and alrreslstancepro-
portlona|tothesqueoftheve|oclty. AndlnFlg. I . &. 2, wheretheconespondlng
helght functlonsaregraphed, thedlerencels hard|yvlslb|e. However,thedler-
ence between |lnear and non|lnereslstance can be slgnlncant ln more comp|ex
sltuatlonssuch as, forlnstance, the atmospherlc reentry and descent ofa space
vehlc|e.
Variable Gravitational Acceleration
Un|essaproectl|elnvertlca|motlonremalnslnthelmmedlatevlclnltyoftheearth' s
surface, the gravltatlona| acce|eratlon actlng on lt l snotconstant. Accordlng to
Newton' s|aw ofgravltatlon,the gravltatlona|forceofattractlonbetweentwopolnt
massesM andm |ocatedatadlstancerapartlsglvenby
GMm
F =

r
( I 9)
where G ls acertalnemplrlca| constant (G :. ::z: I 0
t t
N (mJkg)
:
lnmks
unlts). Theformu|alsa|sova|ldlfeltherorbothof thetwomassesarehomogeneous
spheres, lnthlscase,thedlstancer lsmeasuredbetweenthecentersofthespheres.
The fo||owlng examp|e ls slml|to Examp|e zlnSectlon I . 2, but now we
takeaccountof|unargravlty.

un

fa||lg

to

the

and atana|tltudeof53 kl|ometers


abovethe|unarsurface lts downwardve|oclty ls measuredat I 477k. Itsretro-
rockets, when nred ln free space, provldea dece|eratlonof = 4 ms
:
Atwhat
helghtabovethe|unarsurfaceshou|dtheretrorocketsbeactlvatedtoensureaso
touchdown,c= 0atlmpact)!
Sol uti on Let r (t ) denote the |ander' s dlstance nom the center of the moon at tlme t
(Flg. I . &. 3). When wecomblnethe (constant)thrustacce|eratlon andthe (neg-
atlve) |unar acce|eratlon FJm = GMJr
:
of Eq. ( I 9), we get the (acce|eratlon)
dlerentla|equatlon
(z0)
where M = : I 0

(kg) ls the mass of the moon, whlch has a radlus of


r = I . 74 X I 0
-
meters (or I 740k, a|ltt|eovera quarter oftheearth' sradlus).
FIGUR 1.8.3. The lunar lander
descending to the surface of the
moon.
I . d Accel erati on-Vel ocity Model s 91
Notlngthatthlssecond-orderdlerentla|equatlondoesnotlnvo|vethelndependent
varlab|e .wesubstltute
a a
:
ac ac a ac
: = -.
a
-= - = - - = c -
a
:
a a a a
(as ln Eq. (36)ofSectlon I 6)andobtalnthenrst-orderequatlon
ac cM
c - = - -
a
:
wlth the newlndependentvarlab|e Integratlon wlthrespectto now yle|ds the
equatlon
I
:
cM
c = + -+ c
2
thatwecan app|y bothbeforelgnltlon, =0 ) andafterlgnltlon, =4) .
Before ignition: Substltutlonof=0ln(2I ) glvestheequatlon
I
:
cM
-c = -+ c,
2
wheretheconstantls glvenbyc, =c}z- cM},wlth
k m I h I 4770 m
:,=-I 477 Z I 000 Z =--
h k 3600s 36 s
(2l )
(2l a)
and,= ( I . 74Z I 0
-
)+53, 000= I . 793Z I 0
-
m(fromthelnltla|ve|ocltyposltlon
measurement) .
After ignition: Substltutlonof =4and c=0, = r(attouchdown)lnto(2l )
glves
I
:
cM
c = : +-+ c
:
2
(2lb)
wheretheconstantc
:
=-:r- cM}rlsobtalnedbysubstltutlngtheva|uesc=0,
=rattouchdown.
Atthelnstantoflgnltlonthe|unar|ander' sposltlonandve|ocltysatlsfyboth
(2I a) and(2I b). Therefore wecan nnd ltsdeslredhelght/above the|unarsurface
at lgnltlon by equatlng the rlght-hand sldes ln (2I a) and (2I b). Thls glves =
,c,- c
:
;= I . 7& I &7Z I 0
-
andnna||y/=- r=4I , S70meters(thatls, 4I . S7
kl|ometersustover26ml|es). Moreover, substltutlonofthlsva|ueofln(2I a)
glvestheve|ocltyc=-450m/satthelnstantoflgnltlon.
92 Chapter I Fi rst-Order Di fferenti al Equati ons

Ve|o;|tye (t )
FIGUR 1.8.4. A mass m at a
great distance from the earth.
Escape Velocit
Inhls nove| io-/eIo/o/eMoo( I &65)Ju|esVeme ralsed the questlon
ofthe lnltla| ve|oclty necesary fora proectl|e nred nom the surface ofthe earth
toreach the moon. Slml|ar|y, we can ask whatlnltla| ve|oclty :,l snecessary for
the proectl|e to escape from the earth a|together. Thl s wl|| be so lf lts ve|oclty
c= a }aremalnspos//.efora|| >0,soltcontlnuesforevertomoveawayfrom
theearth.Wlth , ; denotlngtheproectl|e' sdlstancefromtheearth' scenterattlme
(Flg. I . &. 4), wehavetheequatlon
ac a

cM
a
=
a
,
=
`
(22)
slml|ar to Eq. (2O), butwlth = O (nothrust)and wlth M = 5. 975 I 0

(kg)
denotlngthemassoftheearth, whlchhasanequatorla|radlusofr=6. 37&I 0

(m). Substltutlonofthechalnru|e expresslonac}a = t (dtJdr) as ln Examp|e:


glves
ac cM
c = --
a

Thenlntegratlonofboth sldeswlthrespecttoyle|ds
I

cM
-c = -+ c
2
Now c= :,and = rwhen = 0, soc= c- cMJr, andhenceso|utlonfor
c
:
glves
, ,
I
c = :
,
+zcM _ -
r
(23)
Thlslmp|lcltso|utlonofEq.(22)determlnestheproectl|e' sve|ocltycasafunctlon
ofltsdlstancefromtheearth' scenter. Inpartlcu|ar,

zcM
c > c
,
- -
r

socwl||remalnposltlveprovldedthatc 2cMJr Therefore,theescape velocity


fromtheearthlsglvenby
:,=
2M
. (2:)
InProb|em27 we ask you to show that, lftheproectl|e' slnltla|ve|oclty exceeds
zcM}r, then , - c as - c, so lt does, lndeed, escape from the
e. Wlth theglvenva|uesofcand the earth' smass Mand radlusr, thlsglves
:, I I , I &O(mJs)(about36, 6&0ftJs, about6. 95mlJs, about25,000mlJh) .
Hemarkt Equatlon(24;glvestheescapeve|ocltyforanyother(spherlca|)
p|anetary body when we use /smass and radlus. Forlnstance, when weusethe
massMandradlusrforthemoonglvenlnExamp|e:,we nndthatescapeve|oclty
fromthe |unarsurface ls :, 2375mJs. Thls lsustoverone-nfth oftheescape
ve|ocltyfromtheearth' ssurface,afactthatgreat|yfacl|ltatestheretumtrlp(From
theMoontotheEarth).
Problems
1. The acceleration of a Maserati is proportional to the dif
ference between 250 km/h and the velocity of this sports
car. If this machine can accelerate from rest to 1 00 km/h
in 1 0 s, how long will it take for the car to accelerate from
rest to 200 km/h?
2. Suppose that a body moves through a resisting medium
with resistance proportional to its velocity v, so that
..-.v. (a) Show that its velocity and position
at time are given by
v

and
.. , ( -


(b) Conclude that the body travels only a fnite distance,
and fnd that distance.
3. Suppose that a motorboat is moving at 40 f/s when its
motor suddenly quits, and that 1 0 s later the boat has
slowed to 20 f/s. Assume, as in Problem that the re
sistance it encounters while coasting is proportional to its
velocity. How far will the boat coast in all?
4. Consider a body that moves horizontally through a
medium whose resistance is proportional to the .,..of
the velocity v, so that dvd-.v
2
Show that
and that

1 .
1
.._lnO .
Note that, i n contrast with the result of Problem .
as Which ofers less resistance when the
body is moving fairly slowly-the medium in this prob
lem or the one in Problem 2? Does your answer seem
consistent with the observed behaviors of .as o?
. Assuming resistance proportional to the squae of the ve
locity (as in Problem 4), how far does the motorboat of
Problem 3 coast in the frst minute afer its motor quits?
6. Assume that a body moving with velocity v encounters
resistance of the form dvd.3
2
Show that
and that
.

2

.
,
. . ,
. '

Conclude that under a -power resistance a body coasts


only a fnite distance before coming to a stop.
1. Suppose that a car starts from rest, its engine providing an
acceleration of 1 0 f/s
2
, while air resistance provides 0. 1
f/S
2
of deceleration for each foot per second of the ca' s
velocity. (a) Find the car' s maximum possible (limiting)
velocity. (b) Find how long it takes the ca to attain 90%
of its limiting velocity, and how fa it travels while doing
so.
I . d Accel erati on-Vel ocity Model s 7J
8. Rework both parts of Problem 7, with the sole diference
that the deceleration due to air resistance now is (0. 001 ) v
2
f/s
2
when the ca's velocity is v feet per second.
9. A motorboat weighs 32,000 lb and its motor provides a
thrust of 5000 lb. Assume that the water resistance is 1 00
pounds for each foot per second of the speed v of the boat.
Then
dv
1 000- 5000 - 1 OOv.
dt
If the boat starts from rest, what is the maximum velocity
that it can attain?
10. A woman bails out of an airplane at an altitude of 1 0,000
f, falls freely for 20 s, then opens her parachute. How
long will it take her to reach the ground? Assume lin
ear air resistance pv f/S
2
, taking p 0. 1 5 without the
paachute and p 1 . 5 with the paachute. .,,..
First determine her height above the ground and velocity
when the paachute opens. )
11. According to a newspaper account, a paatrooper survived
a training jump from 1 200 f when his paachute failed to
open but provided some resistance by fapping unopened
in the wind. Allegedly he hit the ground at 1 00 mi/h after
falling for s. Test the accuracy of this account. .,,.
. Find p in Eq. (4) by assuming a terminal velocity
of 1 00 mi/h. Then calculate the time required to fall 1 200
ft. )
12. It is proposed to dispose of nuclea wastes-in drums with
weight W 640 lb and volume f3-by dropping them
into the ocean 0). The force equation for a drum
falling through water is
where the buoyant force .is equal to the weight (at 62. 5
lb/ft3) of the volume of water displaced by the drum
(Archimedes' principle) and ,is the force of water re
sistance, found empirically to be 1 lb for each foot per
second of the velocity of a drum. If the drums are likely
to burst upon an impact of more than 75 f/s, what is the
maximum depth to which they can be dropped in the ocean
without likelihood of bursting?
13. Separate variables in Eq. and substitute .v"
to obtain the upward-motion velocity function given in
Eq. ( 1 3) with initial condition v(O)
14. Integrate the velocity fnction in Eq. ( 1 3) to obtain the
upwad-motion position function given in Eq. (4) with
initial condition y(O)
15. Separate vaiables i n Eq. (5) and substitute .v"
to obtain the downwad-motion velocity function given in
Eq. ( 1 6) with initial condition v(O)
16. Integrate the velocity function in Eq. ( 1 6) to obtain the
downwad-motion position fnction given in Eq. ( 1 7) with
initial condition y(O)
94 Chapter I Fi rst-Order Di fferenti al Equati ons
17. Consider the crossbow bolt of Example 3, shot straight
upwad from the ground 0) at time 0 with initial
velocity co 49 m/s. Take ,9.8 m/s
2
and p 0. 001 1
in Eq. ( 1 2). Then use Eqs. ( 1 3) and ( 1 4) to show that
the bolt reaches its maximum height of about 1 08. 47 m in
about 4. 61 s.
18. Continuing Problem 1 7, suppose that the bolt is now
dropped (vo 0) from a height of Yo 1 08. 47 m. Then
use Eqs. ( 1 6) and ( 1 7) to show that it hits the ground about
4. 80 s later with an impact speed of about 43. 49 m/s.
19. A motorboat starts from rest (initial velocity v()
0). Its motor provides a constant acceleration of 4 f/S
2
,
but water resistance causes a deceleration of v
2
/ 400 f/ S
2
.
Find v when 1 0 s, and also fnd the .,
as (that is, the maximum possible speed of the
boat) .
20. An arrow is shot straight upwad from the ground with an
initial velocity of 1 60 f/s. It experiences both the decel
eration of gravity and deceleration v
2
/800 due to air resis
tance. How high in the air does it go?
21. If a ball is projected upwad from the ground with initial
velocity Vo and resistance proportional to v
2
, deduce from
Eq. ( 1 4) that the maximum height it attains is
1

,c
_
Y
max
-In 1 .
2p ,
22. Suppose that p 0. 075 (in fps units, with ,32 ft/s
2
)
in Eq. ( 1 5) for a paratrooper falling with paachute open.
If he j umps from an altitude of 1 0, 000 f and opens his
parachute immediately, what will be his terminal speed?
How long will it take him to reach the ground?
23. Suppose that the paratrooper of Problem 22 falls freely for
30 s with p 0. 00075 before opening his parachute. How
long will it now take him to reach the ground?
24. The mass of the sun is 329, 320 times that of the earth and
its radius is 1 09 times the radius of the earth. (a) To what
radius (in meters) would the earth have to be compressed
in order for it to become a ...-the escape velocity
from its surface equal to the velocity 3 7 1 08 m/s of
light? (b) Repeat part (a) with the sun in place of the
earth.
25. (a) Show that if a projectile is launched straight upward
from the surface of the earth with initial velocity Vo less
than escape velocity "2GM/R, then the maximum dis
tance from the center of the earth attained by the projectile
is
2GMR
r
-
-----
max
-
2GM
~
Rv
2
'
q
where M and R are the mass and radius of the earth, re
spectively. (b) With what initial velocity Vo must such a
prjectile be launched to yield a maximum altitude of 1 00
kilometers above the surface of the earth? (c) Find the
maximum distance from the center of the earth, expressed
in terms of earth radii, attained by a projectile launched
from the surface of the earth with 90% of escape velocity.
26. Suppose that you are stranded-your rocket engine has
failed-on an asteroid of diameter 3 miles, with density
equal to that of the earth with radius 3960 miles. If you
have enough spring in your legs to jump 4 feet straight up
on earth while wearing your space suit, can you blast of
from this asteroid using leg power alone?
27. (a) Suppose a projectile is launched vertically from the
surface r R of the earth with initial velocity co =
"2GM/R so c

e
/R where .
2
2GM. Then solve
the diferential equation dr/d./
v
(from Eq. (23)
in this section) explicitly to deduce that r ) as

(b) If the projectile is launched vertically with initial ve
locity co "2GM/R, deduce that
dr

.
.

_
dt r
v
Why does it again follow that r ) as o?
28. (a) Suppose that a body is dropped (vo 0) from a dis
tance ro R from the earth' s center, so its acceleration
is dv/dt -GM/r
2

Ignoring air resistance, show that it
reaches the height r ro at time

ro

,rro - r2 + ro cos-1
2GM

.,,.. Substitute r ro cos
2
d to evaluate
"r/(ro - r) dr. ) (b) If a body is dropped from a height
of 1 000 km above the earth's surface and air resistance
is neglected, how long does it take to fall and with what
speed will it strike the earth' s surface?
29. Suppose that a projectile is fred straight upwad from the
surface of the earth with initial velocity c, "2GM/R.
Then its height above the surface satisfes the initial
value problem
GM
R)
2 '
0,
Substitute dv/dt v(dv/d) and then integrate to obtain
2GMy
v
2
c-
R( R
for the velocity v of the projectile at height What maxi
mum altitude does it reach if its initial velocity is 1 km/s?
30. In Jules Vere's original problem, the projectile launched
from the surface of the earth is attracted by both the earth
and the moon, so its distance r (t ) from the center of the
eath satisfes the initial value problem
d
2
r GMe GMm
dt2

R, c,
where Me and Mm denote the masses of the earth and
the moon, respectively; R is the radius of the earth and
384, 400 km is the distance between the centers of
the earth and the moon. To reach the moon, the projectile
must only j ust pass the point between the moon and earth
where its net acceleration vanishes. Thereafer it is "under
the control" of the moon, and falls from there to the luna
surface. Find the ..launch velocity c,that suffces
for the projectile to make it "From the Earth to the Moon."
1 . | COl O
)
F
'
FIGUR 1.8.5. An ascending
rocket.
I . d Accel erati on-Vel ocity Model s 95
Rocket Propulsion
. ...
SupposethattherocketofFlg. I . &. 5b|astsostralghtupwardfromthe surfaceof
theearthattlme=0. We wanttoca|cu|atelts helghtyandve|ocltyc=ay;aat
tlme Therocket ls prope||edbyexhaustgasesthat exlt (rearward) wlth constant
speed . (re|atlve to the rocket) . Because ofthe combustlon oflts me|, the mass
-=-, ; oftherocketlsvarlab|e.
To derlvetheequatlonofmotlonoftherocket, weuseNewton' ssecond|aw
ln theform
ai
- = i
a
( I )
whereil smomentum(theproductofmassandve|oclty)andidenotesnetextema|
force(gravlty,alrreslstance,etc. ). Ifthemass-oftherocketlsconstantsom
'
,; =
0when ltsrocketsaretumedoorbumed out, forlnstancethenEq.( I ) glves
a,-c; ac a- ac
i = = -- + -c = --
a a a a

whlch(wlthac}a=o; lsthemorefaml|larformi=-.ofNewton' ssecond|aw.


But here-ls notconstant. Suppose-changesto-+cm andcto c+ cc
durlngthe short tlme lnterva|fromto+c Thenthechangelnthemomentum
o)/eo./e/se(l s
ci ,-+c-; , c+cc;- -c=-cc+cc-+c-cc
Butthe system a|so lnc|udes the exhaustgasesexpe||ed durlng thls tlme lnterva|,
wlth mass -c-and approxlmateve|oclty c - . Hence thetota| changelnmo-
mentumdurlngthetlme lnterva| cl s
ci ,-cc+cc-+c-cc;+, -c-; ,c- .;
=-cc+.c-+c-cc
Now we dlvlde by c and take the |lmlt as c 0, so c- 0, assumlng
contlnulty of-, ; The substltutlon ofthe resu|tlng expresslon for ai,a ln ( I )
yle|dstherocket propulsion equation
ac a-
-- + .-=i ,z;
a a
Ifi~ i +i,,wherei =--,l saconstantforceofgravltyandi, =-/cls
aforceofalrreslstanceproportlona|tove|oclty,thenEq. (2)nna||yglves
ac a-
-- + .- =--,- /c ,3)
a a
Constant Thrust
Nowsupposethattherocketfue|lsconsumedattheconstantbumrate;durlng
thetlmelnterva| 0, , } , durlngwhlchtlmethemassoftherocketdecreasesfromm
o
to -, . Thus
-,o;=-,.
-,;=-,- ;,
wlthburnoutoccunlngattlme=,
-,,=-, ,
a-
- =-; for_, ,
a
,:;
96 Chapter I Fi rst-Order Di fferenti al Equati ons
PROBLEM 1 Substltutetheexpresslonsln(4)lntoEq. ,;toobtalnthedlerentla|
equatlon
ac
, -- ; ; +/c=/
c- ,-,- / ;
a
So|vethls|lnearequatlonfor
where:,=c,o;and
M=
-
.
,

;
=
-
-
,
-
-

;
.

-, -,
denotestherocket' sfractional mass attlme
No Resistance
(5)
PROBLEM Z Forthecaseofnoalrreslstance,set/= 0ln Eq. (5)andlntegrate
toobtaln
-,
c,; =:,- ,+c |n
-,- ;
(7)
Because-,- ;, =-, , ltfo||owsthattheve|ocltyof therocketatbumout,= , ;
l s
PROBLEM 3 StartwlthEq. (7)andlntegratetoobtaln
I _ c -,
y,;= ,c,+c;- , - - ,-,- ; ; ln

2 ; -,-
/

Itfo||owsthattherocket' sa|tltudeatbumoutl s
(&)
(9)
( I O)
PROBLEM 4 TheV-2rocketthatwasusedtoattackLondonlnWor|dWarIIhad
anlnltla|massofI 2, &50kg, ofwhlch6&. 5was fue|. Thlsfue|bumedunlform|y
for70 secondswlth anexhaustve|oclty of2 kms. Assumeltencountersalrresls-
tanceofI . 45Npermsofve|oclty.Thennndtheve|ocltyanda|tltudeofthevzat
bumoutundertheassumptlonthat lt ls |aunchedvertlca||yupwardomrest onthe
ground.
I . d Accel erati on-Vel ocity Model s 97
PROBLEM 5 Actua||y,ourbaslcdlerentla|equatlonln( 3) app|leswlthoutqua|-
lncatlon on|y when the rocket ls a|ready ln motlon. However, when a rocket ls
slttlng onlts |aunchpadstandandltsenglnesaretumedonlnltla||y,ltlsobserved
thatacertaln tlme lnterva|passesbeforetherocketactua||yb|astso''andbeglns
toascend.Thereasonlsthatlfc=0ln(3), thentheresu|tlnglnltla|acce|eratlon
ac c a-
- = --- ,
a - a
ofthe rocket may be e,o/.e Butthe rocket doesnotdescend lnto the ground,
ltustslts there whl|e (because- ls decreaslng) thls ca|cu|ated acce|eratlon ln-
creasesuntl|ltreaches 0and(thereafter)posltlveva|uessotherocketcanbeglnto
ascend. Wlth the notatlonlntroduced to descrlbed the constant-thrust case, show
thattherocketlnltla||yustsltstherelftheexhaustve|ocltyc ls|essthan-,,};.
andthatthetlme,whlchthene|apsesbeforeactua|b|astols glvenby
Free Space
-,,- ;c
,=
/,
Supposenna||ythattherocket ls acce|eratlng lnfreespace, wherethere ls nelther
gravltynorreslstance, so,=k =oWlth,=0lnEq. (&)weseethat,asthemass
oftherocketdecreasesfrom-,to-| , ltslncreaselnve|ocltyl s
-,
cc=:, - :,=c |n .
-,
( I I )
Notethat cc dependson|yontheexhaustgasspeedcandthelnltla|-to-nna|mass
ratlo -,}-, . but does notdepend on the bum rate ; For examp|e, lfthe rocket
b|astsoomrest,c,=0)andc =5 kmsand-,}-, = 20, then lts ve|ocltyat
bumoutls :, =5 ln20 I 5 kms. Thuslfarocketlnltla||yconslstspredomlnant|y
offue|, thenltcanattalnve|ocltlesslgnlncant|ygreaterthanthe(re|atlve)ve|oclty
oflts exhaustgases.
______ __ _ __
Inthlschapterwehavedlscussedapp|lcatlonsofandso|utlon methodsforsevera|
lmportanttypesofnrst-orderdlerentla|equatlons,lnc|udlngthosethataresepara-
b|e (Sectlon I . 4), |lnear(Sectlon l . 5), orexact(Sectlon I . 6). InSectlon I . 6wea|so
dlscussedsubstltutlontechnlquesthatcansometlmesbeusedtotransform aglven
nrst-orderdlerentla|equatlonlntoonethatls eltherseparab|e,|lnear,orexact.
Lestltappearthatthesemethodsconstltuteagrabbag ofspecla|andunre-
|ated technlques, ltls lmportantto note thatthey are a|| verslonsofa slng|eldea.
Glvenadlerentla|equatlon
),x.y,y

=0,
weattempttowrlteltlntheform
a
ax
c,x. y } =o
( I )
(2)
98 Chapter I First-Order Di fferenti al Equati ons
Itl spreclse|ytoobtalntheformlnEq. (2)thatwemu|tlp|ythetermslnEq. ( I ) byan
approprlatelntegratlngfactor (evenlf a|| wearedolngls separatlngthevarlab|es).
Butoncewehavefoundafunctlonc,x.y;suchthatEqs. ( I ) and(2)areequlva|ent,
agenera|so|utlonlsdennedlmp|lclt|ybymeansoftheequatlon
c,x. y;= c ,;
thatoneobtalnsbylntegratlngEq.(2).
Glvenaspeclncnrst-orderdlerentla|equatlontobeso|ved,wecanattacklt
bymeansofthefo||owlngsteps.

Isltsep.m//e:Ifso, separatethevarlab|esandlntegrate(Sectlon I . 4).

Islt//e.:Thatl s, canltbewrltten lntheform


ay
- + i,x; y= _,x; :
ax
Ifso,mu|tlp|ybythelntegratlngfactorp = exp]iax) ofSectlon I . 5.

Isltex..:Thatl s, whentheequatlonlswrlttenlntheformMax+xay= 0,
ls - M}-y= -x }-x(Sectlon I . 6)!

Iftheequatlonasltstandslsnotseparab|e,|lnear,orexact,lsthereap|auslb|e
substltutlonthatwl|| makelt so! Forlnstance, l s lt homogeneous (Sectlon
I . 6)!
Manynrst-orderdlerentla|equatlonssuccumbtothe|lneofattackout|lned
here. Neverthe|ess, many more do not. Becauseofthe wlde aval|abl|lty ofcom-
puters, numerlca| technlques arecommon|y used to .ppax/-.ethe so|utlons of
dlerentla| equatlonsthatcannotbeso|vedreadl|y orexp|lclt|ybythe methods of
thls chapter. Indeed, mostofthe so|utloncurves shown ln ngures ln thls chapter
were p|otted uslngnumerlca| approxlmatlonsratherthan exact so|utlons. Severa|
numerlca|methodsfortheapproprlateso|utlonofdlerentla|equatlonswl||bedls-
cussedlnChapter6.
LOl| 1 Revew Problems
- - --- --- --- - - - - - - - - - - - - - - -- - -- - - - - - - - - - -- -- - - - - -- - - -- -- -- - -- - -- - -- - -- - - ----- - -- -- - - - -- - - - --- -- ---- ---- ---
i/a,ee./so//oso)/ea_ ee/./e,./os/ia//e-s1 /o,/J0. i/-esaeoeae/../.es//espe.
o x
1. x3 + 3y - xy' = 0
2. xy2 + 3y2 - x2y' 0
3. xy + y2 - x2y' = 0
4. 2xy3 +
eX + (3x2y2 + sin y) y' = 0
S. 3y + x4y' = 2xy
6. 2xy2 + x2y' y2
7. 2x2y + x3 y' = 1
8. 2xy + x2y' = y2
9. xy' + 2y = 6x2 .
10. y' = 1 + x2 + y2 + X2y2
11. x2y' xy + 3y2
12. 6xy3 + 2y4 + (9x2y2 + 8xy3 )y' = 0
13. 4xy2 + y' 5x4y2
14. x3 y' = x2y y3
15. y' + 3y = 3x2e
-3x
16. y' = x2 - 2xy + y2
17. eX
+ yexy
+ (eY + xeYX
)y' = 0
18. 2x2y x3 y' = y3
19. 3X5y2 + x3 y' = 2y2
20. xy' + 3y = 3X-3/2
21. (x2 - l )y' + (x - l ) y = 1
22. xy' 6y + 1 2x4y2/3
23. eY + y cos x + (xeY + sin x) y' = 0
24. 9x2y2 + X3/2y' = y2
25. 2y + (x + l )y' " 3x + 3
26.
.
.

0
27. ...

...

0
28. ...

29. . ..

30. .

..
...., ..,......31 ....36
..., .,..........,.,..
...............
I . d Accel erati on-Vel ocity Model s 99
.........,......., .....
........
31.
.. ..
~
.. .

32.
~
..

..
.. ..
33.
.. ..

34.
..

. ..

.. ... .. . ..
35.
.. .. .
36.
..
,
]
.. .

.. tan .
linear Iquations
olHigherrder
Introduction: Second-Order Iinear Equations
\

n Chapter l we investigated frst-order diferential equations. We now mm to


equations of higher order 2, beginning in this chapter with equations that are
linear. The general theory of linear diferential equations parallels the second-order
case ,= 2), which we outline in this initial section.
Recall that a second-order diferential equation in the (unknown) function
y(x) is one of the form
c(x. y. y .y;= c , l ;
This diferential equation i s said to be linear provided that ci s linear i n the depen
dent variable yand its derivatives yand y Thus a linear second-order equation
takes (or can be written in) the form
n(x)y+s(x) y+c(x) y= i,x; (2)
Unless otherwise noted, we will always assume that the (known) coefcient func
tions n,x; , s,x; , c,x; , and i,x; are continuous on some open interval i(perhaps
unbounded) on which we wish to solve this diferential equation, but we do o
require that they be linear functions of x Thus the diferential equation
e

y+(cos x) y+, l +.; y= tan


-
1
x
is linear because the dependent variable yand its derivatives yand yappear lin
early. By contrast, the equations
y= yy and y+ ,y ;
:
+:,= c
are not linear because products and powers of yor its derivatives appear.
Vass
) - O ) > O
La|||ot| am
os|t|oa
uasaot

FIGUR 2. 1. 1. A mass
spring-dashpot system.
), c > 0
FIGUR 2. 1.2. Directions of the
forces acting on m,
Z. I I ntroducti on: Second-Order Li near Equati ons 1 01
Ifthefunctloni,x; ontherlght-handsldeofEq. (2) vanlshesldentlca||yon
i, then weca|| Eq. (2) ahomogeneous |lnearequatlon,otherwlse,ltls nonhomo
geneous. Forexamp|e,thesecond-orderequatlon
x
:
y+2xy'+ y~ cosx
ls nonhomogeneous , ltscsso./ceahomogeneousequatlonls
x
:
y+2xy
'
+y~ 0.
Ingenera|,thehomogeneous|lnearequatlonassociated wlthEq. (2)ls
n,x; y+ s,x; y+c,x; y~ 0. (3)
Incasethedlerentla| equatlonl n(2)mode|saphyslca| system,thenonhomoge-
neousterm i,x; frequent|ycorrespondstosomeeec/lnuenceonthesystem.
Remark: Notethatthemeanlngofthetermhomogeneousforasecond-
order|lneardlerentla|equatlonlsqultedlerentfromlts meanlngforanrst-order
dlerentla| equatlon (as ln Sectlon I . 6). Ofcourse, lt ls not unusua|elther ln
mathematlcsorlntheEng|lsh|anguagemoregenera||yforthesamewordtohave
dlerentmeanlngslndlerentcontexts.
A Typical Application
Llneardlerentla|equatlonsnequent|yappearasmathematlca|mode|sofmechan-
lca|systemsande|ectrlca|clrcults. Forexamp|e,supposethatamasslsattached
bothtoasprlng that exerts on lt aforce iand toadashpot(shockabsorber)that
exerts aforce i, on the mass (Flg. 2. I . I ). Assume that the restorlngforce iof
the sprlng l sproportlona| to the a/sp/c.eex ofthe mass from lts equl|lbnum
posltlonandactsopposltetothedlrectlonofdlsp|acement. Then
i~ -/x (wlth/> c)
soi < clfx > c(sprlngstretched)whl|ei > clfx < c,sprlngcompressed).
Weassumethatthe dashpotforce i , ls proportlona| to the .e/o./-c ~ ax}aof
themass andactsopposltetothedlrectlonofmotlon.Then
ax
i,~ -.c~ -.- (wlth.> c)
a
soi , < clfc> c(motlontotherlght)whl|ei , > clfc < c(motlontothe|ef).
Ifi,andiaretheon|y forcesactlng onthemassandltsresu|tlngacce|-
eratlon lsc~ ac}a, thenNewton' s|aw i~ cglves
,:;
thatl s,
(5)
Thus wehaveadlerentla|equatlonsatlsnedbytheposltlonfunctlonx,;ofthe
massThlshomogeneoussecond-order|lnearequatlongovemsthe)ee.//a/os
ofthemass, wewl||retumtothlsprob|emlndetal|lnSectlon2. 4.
1 02 Chapter Z Li near Equati ons of Hi gher Order
If, ln addltlon to F and F, the mass l s acted onbyanextema| force
F(t )whlch mustthen be added to the rlght-hand slde ln Eq. (4)the resu|tlng
equatlonl s
a
:
x ax
-
:
+.- +/x=F( ; .
a
a
,:;
Thls nonhomogeneous |lneardlerentla|equatlongovemsthe)o.ea.//c/osol
themassunderthelnuenceoftheextema|force F(t) .
Homogeneous Second-Order Linear Equations
Conslderthegenera|second-order|lnearequatlon
~,x; y+s,x; y+c,x; y~ i,x; . (7)
wherethecoefnclentfunctlons n,s, c,and F arecontlnuousontheopenlnterval
i Hereweassumeln addltlonthat~,x; y Oateachpolntofi. sowecandlvlde
eachtermlnEq. (7)byA(x) andwrlteltlntheform
y+p,x; y+, ,x; y~ ),x; ,s)
Wewl||dlscussnrsttheassoclatedhomogeneousequatlon
y+p,x; y+, (x; y~ 0 (9)
Apartlcu|ar|yusefu|propertyofthl s/oo,eeos|lnearequatlonlsthefactthatthe
sumof any twoso|utlons ofEq. (9) ls agalnaso|utlon, as l sany constantmu|tlp|e
ofaso|utlon. Thls lsthecentra|ldeaofthefo||owlngtheorem.
THEOREM 1 Principle of Superposition for Homogeneous
Equations
Lety,andy
:
betwoso|utlonsofmehomogeneous|lnearequatlonln(9)onthe
lnterva|. IfCj andc
:
areconstants,thenthe|lnearcomblnatlon
( I O)
ls a|soaso|utlonofEq. (9)oni
Proof: Theconc|uslonfo||owsa|mostlmmedlate|yfromthe|lnearltyofthe
operatlonofdlerentlatlon,whlchglves
Then
' ' '
d " y ~ ., y
,
+.
:
y
:
an y ~ ., y
,
+.
:
y
:
y+py+,y~ ,., y,+.
:
y
:
;+
;
,., y,+.
:
y
:
;+, ,., y,+.
:
y
:
;
~ ,., y
+.
:
y
;+p,., y,+.
:
y,
;+, ,., y,+.
:
y
:
;
~ .,(y

+
;y,+,y, ;+.
:
,y

+;y,
+,y
:
;
~ ., O+.
:
O~ O
becausey, andy
:
areso|utlons. Thus y~ ., y,+.
:
y
:
ls a|soaso|utlon.
s
s

Z
P a
-Z

s
s
R
. .. . .
cXump| e 1

[

a R
X
Z. I I ntroduction: Second-Order Li near Equati ons 1 03
We can see by inspection that
Yl (x) ~ cos x and Y
2
(X) ~ sin x
are two solutions of the equation
Y" +Y ~
O.
Theorem 1 tells us that any linear combination of these solutions, such as
y(x) ~ 3Yl (X) - 2Y
2
(X) ~ 3 cos x - 2 sin x,
is also a solution. We will see later that, conversely, e.esolution of y" +Y =c
is a linear combination of these two particular solutions Yl and Y
2
. Thus a general
solution of y" +Y ~ cis given by
y(x) ~ Cl cos x +C
2
sin x.
It i s important to understand that this single formula for the general solution encom
passes a "twofold infnity" of particular solutions, because the two coefcients Cl
and C
2
can be selected independently. Figures 2. 1 . 3 through 2. 1 . 5 illustrate some of
the possibilities, with either Cl or C
2
set equal to zero, or with both nonzero.
s . a
s
s

Z
Z
P a P a
-Z
- Z


s
s

s
z
s
R a R ZK
. a
R a R ZK 1a
X
FIGURE 2. 1.3. Solutions
y(x) = C[ cos x of y" y =
FIGURE 2. 1.4. Solutions
y(x) = C

sinx of y" y
FIGURE 2.1.5. Solutions of
y" y = 0 with C[ and

both
nonzero.
Earlier in this section we gave the linear equation mx" +CX' +kx ~ i,;as
a mathematical model of the motion of the mass shown in Fig. 2. 1 . 1 . Physical con
siderations suggest that the motion of the mass should be determined by its initial
position and initial velocity. Hence, given any preassigned values of x ,c)and x' (
O),
Eq. ( 6) ought to have a /,e solution satisfying these initial conditions. More
generally, in order to be a "good" mathematical model of a deterministic physical
situation, a diferential equation must have unique solutions satisfing any appro
priate initial conditions. The following existence and uniqueness theorem (proved
in the Appendix) gives us this assurance for the general second-order equation.
1 04 Chapter Z Li near Equati ons of Hi gher Order
Z

0
- I `
)'()= -
- Z
. a Z J .

X
FIGUR 2. 1.6. Solutions of
y" + 3y' + 2y 0 with the same
initial value y eO) but diferent
initial slopes.

.
J
Z
)()= J

- I
- Z
- J

.
- I 0
`
)() = -J
Z J 4
X
FIGUR 2. 1.7. Solutions of
y" + 3y' + 2y 0 with the same
initial slope y' (0) but diferent
initial values.
Exampl e 1
COlI|lUCO
THEOREM Z Existence and Uniqueness for Linear Equations
Suppose that the functions , g, and I are continuous on the open interval i
containing the point c. Then, given any two numbers bo and bl , the equation
y

+p,x; y+, ,x; y= I(x) ,s)


has a unique (that is, one and only one) solution on the entire interval i that
satisfes the initial conditions
y,c; = bo, y ,o;= bl . ( I I )
Remark 1 : Equation ,s)and the conditions i n ( I I ) constitute a second
order linear initial value problem. Theorem 2 tells us that any such initial value
problem has a unique solution on the /o/einterval iwhere the coefcient func
tions in (S) are continuous. Recall from Section I . 3 that a o//ecdiferential
equation generally has a unique solution on only a smaller interval.
Remark 2: Whereas +]soaediferential equation ay,ax= i,x , y;
generally admits only a single solution curve y = y,x; passing through a given
initial point ,c,b), Theorem 2 implies that the se.oaoaeequation in ,s)has
infnitely many solution curves passing through the point ,c, bo)-namely, one for
each (real number) value of the initial slope y,c; = bl . That is, instead of there
being only one line though ,c,bo) tangent to a solution curve, e.enonvertical
straight line through ,c,bo) is tangent to some solution curve of Eq. ,s) Figure
2. I . 6 shows a number of solution curves of the equation y+y+ zy = 0 all
having the same initial value y,c;= I , while Fig. 2. I . 7shows a number of solution
curves all having the same initial slope y(0) = I . The application at the end of
this section suggests how to construct such families of solution curves for a given
homogeneous second-order linear diferential equation.

We saw in the frst part of Example I that y,x;= 3cos x 2sin xis a solution (on
the entire real line) of y+y = c It has the initial values y,c; = 3, y ,c;= 2
Theorem 2 tells us that this is the o/y solution with these initial values. More
generally, the solution
y,x; = bo cos x +bl sin x
satisfes the c//minitial conditions y(0) = bo, y(0) = bl ; this illustrates the
ex/se.eof such a solution, also as guaranteed by Theorem 2.

Example I suggests how, given a /oo,eeossecond-order linear equation,


we might actually fnd the solution y,x; whose existence is assured by Theorem 2.
First, we fnd two "essentially different" solutions y,and y
:
,second, we attempt to
impose on the general solution
the initial conditions y,c;
simultaneous equations
for the coefcients .,and .
:

( I 2)
bo, y,c; = bl . That is, we attempt to solve the
.l y, (c) +.
:
y
:
(c) = bo,
.l y,,c;+.
:
y ,c;= bl
, l )
Exampl e Z
Z. I I ntroducti on: Second-Order Li near Equati ons 1 05
Verlfythatthefunctlons
areso|utlonsofthedlerentla|equatlon
y- zy
+y~ c,
andthennndaso|utlonsatlsfylngthelnltla| condltlonsy,c;~ ,y ,c;=I
Sol ution Theverlncatlonl sroutlne, weomltlt.Welmposetheglvenlnltla|condltlonsonthe
genera|so|utlon
l 0

?
6
4
z
P 0
- z
- 4
- 6
- ?
X
FIGUR 2. 1.8. Diferent
solutions y (x)

.of
y" - 2y' y 0 with the same
initial value y (O)
Exampl e
z
forwhlch
y ,x;~ ,.,+.
:
;e

+.
:
xe

.
toobtalntheslmu|taneousequatlons
y,c; ~ ., ~ .
y,c)~ ., +.
:
~ I
Theresu|tlngso|utlonls ., ~ ..
:
~ -z Hencetheso|utlonoftheorlglna|lnltla|
va|ueprob|eml s
y,x; =e

- zxe

Flgure 2. I . S showssevera| addltlona|so|utlonsofy- zy+y~ c,a||havlngthe


samelnltla|va|uey,c;~
InorderfortheprocedureofExamp|e2to succeed,thetwo so|utlons) and
y
:
musthavethee|uslvepropertythattheequatlonsln( I 3)cana|waysbesolvedfor
., and.
:
,no matterwhatthelnltla| condltlons/, and/, mlghtbe. The fo||owlng
dennltlonte||spreclse|yhowdlerentthetwofunctlonsy, andy
:
mustbe.
DEFI NITI ON Li near I ndependence of Two Functions
Tofunctlonsdennedonanopenlnterva|iaresaldtobelinearly independent
oniprovldedthatneltherls aconstantmu|tlp|eoftheother.
Twofunctlonsaresaldtobelinearly dependent onanopenlnterva|provlded
thattheyarenot|lnear|ylndependentthere,thatl s, one ofthem ls aconstantmultl-
p|eoftheother. Wecana|waysdetermlnewhethertwoglvenfunctlons)and,are
|lnear|ydependenton anlnterva| iby notlngatag|ancewhethereltherofthe two
quotlents) },or,})l saconstant-va|uedmnctlonon i
Thus ltls c|earthatthefo||owlngpalrsoffunctlonsare|lnear|ylndependentonthe
entlrerea||lne.
sln x and cos x,
e

and e
-
:

,
e

and xe

,
x + i and x
:
,
x and x

1 Chapter Z Li near Equati ons of Hi gher Order


That ls, nelther sln xJ cos x ~ tan x norcos xJ sln x ~ cot x l sa constant-va|ued
functlon, nelther e

,e
-
:

~ e

nor e
-
:
,e

ls aconstant-va|uedfunctlon, and so
forth. Butthe ldentlca||y zero mnctlon )(x) = O and any other functlon , are
|lnear|y dependent on every lnterva|, because O
.
,(x) ~ O ~ )(x) . A|so, the
functlons
)(x) ~ sln 2x and g(x) ~ sln x cos x
are|lnear|ydependentonanylnterva|because)(x) ~ 2,(x)lsthefaml|lar trlgono-
metrlc ldentlty sln2x ~ 2slnxcosx.
General Solutions
Butdoesthehomogeneousequatlon, + py+q, ~ Oa|wayshavetwo|lnear|y
l ndependentso|utlons!Theorem2saysyes | Weneedon|ychoose), andy
:
sothat
It ls then lmposslb|ethatelthery ~ /y
:
ory
:
~ /y, because/ O = I forany
constant/ Theorem2 te||s us that two such|lnear|y lndependentso|utlonse/s,
actua||y nndlng them ls a crucla| matterthatwewl||dlscussbrley atthe endof
thl ssectlon, andlngreaterdetal|beglnnlnglnSectlon2. 3.
Wewanttoshow,nna||y,thatglvencytwo|lnear|yl ndependentso|utlons)
andy
:
ofthehomogeneousequatlon
,(x) +p,x; y ,x;+q (x) ,(x) ~ O, (9)
e.eso|utlon
yofEq. (9)canbeexpressedasa|lnearcomblnatlon
( I 2)
ofy,andy
:
Thlsmeansthatthefunctlonln( I 2)lsa,eec/so//oofEq. (9)lt
provldesa||posslb|eso|utlonsofthedlerentla|equatlon.
Assuggestedbytheequatlonsln, i ; , thedetermlnatlonoftheconstants.,
and .
:
ln ( I 2) depends on a certaln 2 2 determlnant ofva|ues ofy, , y
:
, and
thelrderlvatlves. Glventwo functlons )and ,, theWronskian of)and, ls the
determlnant
u =
)
)
,
~ ),- )
,

,
Wewrlteeltherw(], ,;orw(x) , dependlngonwhetherwewlshtoemphaslzethe
twofunctlonsorthepolntxatwhlchtheWronsklanlstobeeva|uated. Forexamp|e,
and

cosx slnx

: :
w(cosx, smx) ~ + ~ cos x +sm x ~ i
- smx cos x

u,e xe ) ~

+

~ e
e e xe
Theseareexamp|esof|lnear|y /aepeaepalrs ofso|utlonsofdlerentla| equa-
tlons(seeExamp|es iand2). NotethatlnbothcasestheWronsklanlseverywhere
o,eo
Z. I I ntroducti on: Second-Order Li near Equati ons 1 07
Ontheotherhand,lfthefunctlons] and,are|lnear|ydependent,wlth] =
/,(forexamp|e),then
u, ),,; ~ , ~ /,,- /, ,= c
Thus the Wronsklan oftwo |lnear|y aepeaefunctlons ls ldentlca||y zero. ln
Sectlon 2. 2 we wl|| prove that, lfthe two functlons y, and y
:
are so|utlons ofa
homogeneous second-order|lnearequatlon,thenthe strong converse statedlnpm
(b)ofTheoremho|ds.
Wronskians O Solutions
SugOsO
md} mC lWO so|utlonsofthehomogeneoussecond-order|lnear
equation(9)
j
n
+ p(x;y+, ,x; y O
on OpcintervalI on whichp and,arecontlnuous.
(a) Ify m |inearIydependent,then u,y, ,y
:
;= Oon.
|lnearlyindependent,then u,y, , y
:
;yOateachpolntofi
Thus, glven two so|utlons ofEq. (9), there are ust two posslbl|ltles. The
Wronsklan uls ldentlca||yzerolftheso|utlonsare|lnear|ydependent,theWron-
sklan ls neverzerolftheso|utlonsare|lnear|ylndependent. The|atterfactls what
weneedtoshowthaty~ ., y, +.
:
y
:
ls thegenera|so|utlonofEq. (9)lfy, andy
:
are|lnear|yl ndependentso|utlons.
dfO8fM4 General Sol utions of Homogeneous Equations
Let
)I
mtwoIinear|yIndependentsoIutlonsofthehomogeneousequatlon
, |'J
}
n
p(x; y+, ,x; y~ O
wi antinonopen lnterva| i IfY ls anyso|utlonwhatsoever
ofEq.(9)OH , thenthereexlstnumbers., and.
:
suchthat
I
In essence, Theorem :te||s us that when we have found o|lnear|y lnde-
pendentso|utlonsofthe second-orderhomogeneousequatlonln,9),thenwehave
foundc//ofltsso|utlons. Wethereforeca||the|lnearcomblnatlonY =.,y,+.
:
y

a,eec/so//oofthedlerentla| equatlon.
Proof of Theorem 4: Chooseapolntcofi.andconslderthe slmu|taneous
equatlons
., y,,c;+.
:
y
:
,c;~ r,c; ,
., y,c;+.
:
y ,c;~ I' ,c;
, i:;
The determlnantofthe coemclents ln thls system of|lnear equatlons ln the un-
knowns ., and .
:
ls slmp|y the Wronsklan u,y, ,y
:
; eva|uated at x = c By
1 08 Chapter Z Li near Equati ons of Hi gher Order
cXump| e4
Theorem , this determinant is nonzero, so by elementary algebra it follows that the
equations in ( 1 4) can be solved for CI and C2 . With these values of CI and C2 , we
defne the solution
of Eq. (9), then
G(a) =CI YI (a) +c
2
Y2 (a) =Y(a)
and
G' (a) =CI y,(a) +
c2y (a) ~ Y' (a) .
Thus the two solutions Y and G have the same initial values at a; likewise, s o do I'
and G' . By the uniqueness of a solution determined by such initial values (Theorem
2), it follows that Y and G agree on i Thus we see that
Y(x) = G(x) ~ CI YI
(x) +C
2
Y2 (X) ,
as desired.
y,~ (2) (2)e
2
x ~ 4e
2
x ~ 4YI and y ~ (-2) ( _2)e
-
2
x ~ 4e
-
2
x =4Y2 .
Therefore, y, and y
:
are linearly independent solutions of
y

- 4y ~ c ( 1 5)
But y

,x; ~ cosh 2x and y


-
,x; =sinh 2x are also solutions of Eq. ( 1 5) , because
d
2
d
-
2
(cosh 2x) ~ -(2 sinh 2x) ~ 4 cosh 2x
dx dx
and, similarly, (sinh 2x) " ~ 4 sinh 2x. It therefore follows from Theorem 4 that the
fnctions cosh 2x and sinh 2x can be expressed as linear combinations of y,(x) *
e
2
x and y
:
,x;~ e-
2
x . Of course, this is no surprise, because
by the defnitions of the hyperbolic cosine and hyperbolic sine.
Remark: Because e
2
x , e-
2
x and cosh x, sinh x are two diferent pairs of
linearly independent solutions of the equation Y" - 4y ~ 0 in ( 1 5), Theorem 4
implies that every particular solution Y (x) of this equation can be written both in
the form
and in the form
Y(x) ~ a cosh x +/sinh x.
Thus these two diferent linear combinations (with arbitrary constant coefcients)
provide two diferent descriptions of the set of all solutions of the same diferential
equation y" - 4 y ~ c Hence each of these two linear combinations is a general
solution of the equation. Indeed, this is why it is accurate to refer to a specifc such
linear combination as "a general solution" rather than as "the general solution
.
"
Z. I I ntroducti on: Second-Order Li near Equati ons 1 09
Linear Second-Order Equations with Constant Coeficients
As d l||ustratlon ofthe genera| theory lntroduced ln thls sectlon, we dlscuss the
homogeneoussecond-order|lneardlerentla|equatlon
.y

+/y+.y=0 , i :;
wlth.osccoefnclentsc,/,and.Wenrst|ookforas/,/cso|utlonof Eq., i :;
andbeglnwlththeobservatlonthat
, i:;
so any derlvatlve of c ls a constant mu|tlp|e ofc
Hence, lfwe substltuted
y = c lnEq. , i :; . then each termwou|dbeaconstantmu|tlp|eofc .wlththe
constantcoemclentsdependentonandthecoemclentsc,/,and. Thlssuggests
thatwetry to nndava|ueofsothatthesemu|tlp|esofc wl||havesumzero. lt
wesucceed,theny=c wl||beaso|utlonof Eq., i :;
Forexamp|e,lfwesubstltutey=clntheequatlon
y
- y+:y=0,
weobtaln
Thus
,
2
- +:;c
=0, ,- z; ,- ; c
=0
Hencey = c wl||bea so|utlonlfelther = zor = So, ln searchlngfora
slng|eso|utlon,weactua||yhavefoundtwoso|utlons. Yl ,x;=c
2
andY2 ,x;=c

To carry out thls procedure ln the genera| case, we substltute y = c ln


Eq. , i :; Wlththealdoftheequatlonsln, i :;, wenndtheresu|ttobe
Becausec
ls neverzero, weconc|udethaty,x; =c wl|| satlsfythedlerentla|
equatlonln , i :;preclse|y when ls arootof thec/,c/m/.equatlon
c
:
+/+.=0 , i s;
Thlsquadratlcequatlon ls ca||edthecharacteristic equation ofthehomogeneous
|lneardlerentla|equatlon
.y
+/y+.y=0 , i :;
lfEq. , i s;has two a/s/.(unequa|) roots r, andr2 , thenthecorrespondlngso|u-
tlonsYl ,x; =c

andY2 (X) =c

of, i :;are|lnear|ylndependent. (Why!) Thls


glves thefo||owlngresu|t.
THEOREM 5 Disti nct Real Roots
Iftheroots , and
:
ofthechacterlstlcequatlon ln , i s;arerea|anddlstlnct,
then
, i ;
ls agenera|so|utlonof Eq. , i :;
1 1 0 Chapter Z Li near Equati ons of Hi gher Order
cXump| e Flndthegenera|so|utlonof
zy

- :y+y= 0.
Sol uti on Wecanso|vethecharacterlstlcequatlon
cXump| e
x
FIGURE 2.1.9. Solutions
y(x) = C2 e-
2
x
of y" 2y' = 0
with diferent values of C
2 .
zr
z
:+= c
byfactorlng.
,zr l ) (r - ;= 0
Theroots r, = |andr
z
~ arerea|anddlstlnct,soTheorem 5 yle|dsthegenera|
so|utlon

Thedlfferentla|equatlony+zy= chas characterlstlcequatlon


r
z
+2r = r (r+z;= c
wlthdlstlnctrea| roots r, = candr
z
~ -z Becausec

= i . wegetthegenera|
so|utlon
Flgurez i shows severa| dlerent so|utloncurveswlth . = i . a|| appearlng to
approachthe so|utloncurvey,x; = i,wlthc
z
= c;asx +>
Remark: NotethatTheorem5 changes aprob|emlnvo|vlngadlerentla|
equatlonlnto onelnvo|vlngon|y theso|utlonofan./,c/m/.equatlon.
lfthecharacterlstlcequatlonln( I S) has equa|roots r, = r
z
, weget,atnrst)
on|y the slng|e so|utlon y, ,x; ~ c ofEq. , i :; The prob|em ln thls case lsto
producethemlsslngsecondso|utlonofthedlerentla|equatlon
Adoub|erootr = r, wl||occurpreclse|y when thecharacterlstlcequatlonls
aconstantmu|tlp|eoftheequatlon
Anydlerentla|equatlonwlththlscharacterlstlcequatlonlsequlva|entto
,zc;
Butltlseasytoverlfybydlrectsubstltutlonthaty = xc lsasecondso|utlonof
Eq. ,zc; ltlscear,butyoushou|dverlfy)that
are|lnear|y lndependentfunctlons, sothegenera| so|utlonofthedlerentla|equa-
tlonln,zc;ls
cXump| e
x
FIGUR 2. 1. 10. Sol utions
j(x) = + .of
j
o
+ 2y' + y = 0 with diferent
values of Cl '
Problems
Z. I I ntroducti on: Second-Order Li near Equati ons 1 1 1
THEOREM Repeated Roots
Ifthe characterlstlcequatlonln( 1 8) hasequa| (necessarl|yrea|)roots, =

.
then
,zi ;
l sagenera|so|utlonofEq.( I 6) .
To so|vethelnltla|va|ueprob|em
y

+zy+y=c,
y,c;=. y
,c;=-.
wenotenrstthatthecharacterlstlcequatlon

+z+i= ,+i ;

=c
has equa| roots , =

= -i Hencethe genera| so|utlonprovldedbyTheorem6


ls
Dlfferentlatlonyle|ds
sothelnltla|condltlonsyle|dtheequatlons
y,c; = ., .
y ,c;= -., +.

=-.
whlch lmp|ythat., = and .

= z Thus the deslredpartlcu|ar so|utlonofthe


lnltla|va|ueprob|emls
y,x;=c
-
+zxc
-

Thls partlcu|arso|utlon, togetherwlth severa| others oftheformy,x; = . c

+
zxc
-
.ls l||ustratedlnFlg.z i ic
Thecharacterlstlcequatlonln( 1 8) mayhaveeltherrea|orcomp|exroots. The
caseofcomp|exroots wl||bedlscussedlnSectlonz
..1 ....16, ...........
.., ..,........

........,.
..........,......
......, ..,.........,...
....= Cl + .........
...............,x.
2. y" - 9y = 0; l =

yeO) = -1 , y' (0) = 1 5


3. y" + 4y = 0;
= cos 2x, = sin 2x; yeO) = 3,
y' (O) =
4. y" + 25y = 0;
= cos 5x, .= sin 5x; yeO) = 1 0,
y' (0) = -1 0
5 . y" -3y' +2y = 0; = =

yeO) = 1 , y' (O) = 0


6. y" + y' - 6y = 0; l =

yeO) =
y' (0) = -1
1. y" - y = 0; = = y eO) = 0, y' (0) = 5
1 1 2 Chapter Z Li near Equati ons of Hi gher Order
7. 0; YI I , Y
2

8. 0; YI I , Y
2

.
9. 0; YI Y
2
x ;
-I
10. 0; YI

Y
2
.


=
1 1 . 0; YI cos x, Y
2
sin x; 0,

12. 0; YI

cos 2x, Y
2

sin 2x;
0
13. .

. 0; YI X, Y
2
x
2
;
I
14. .

. 0; YI x
2
, Y
2
x
-3
; 0,

15. .

. 0; YI X, Y
2
x ln x;

16. .

. 0; YI cos(1n x) , Y
2
sin(1n x) ;

......,..........,
,..,.,............,..
..
17. Show that I /x is a solution of

0, but that if
.0 and .I , then / x is not a solution.
18. Show that x
3
is a solution of 6x
4
, but that if

. then .

is not a solution.
19. Show that YI = I and Y
2
. are solutions of

0, but that their sum YI Y


2
is not a solution.
......,........20
....26 .....,......,....
...
20. . .. cos
2
X sin
2
x
21. f(x) x
3
, g(x) x
2
1 x l
22. f(x) I x, g(x) I I x l
23. f(x) . .. I x
24. f(x) sin
2
x, g(x) cos 2x
25. . sin x, g(x) cos x
26. . 2 cos x sin x, g(x) cos x sin x
27. Let

be a particular solution of the nonhomogeneous


equation ,,. and let Yc be a solu
tion of its associated homogeneous equation. Show that
Yc Y
P
is a solution of the given nonhomogeneous
equation.
28. With Y
P
I and Yc = CI COS X C
2
sin x in the notation
of Problem fnd a solution of I satisfying the
initial conditions
29. Show that YI x
2
and Y
2
x
3
are two different solu
tions of .

..0, both satisfying the initial


conditions 0 Explain why these facts do
not contradict Theorem (with respect to the guaranteed
uniqueness).
30. (a) Show that YI x
3
and Y
2
x
3
are linearly
independent solutions on the real line of the equation
.

. (b) Verify that W (YI


, Y
2
) is iden
tically zero. Why do these facts not contradict Theorem
3?
31. Show that YI sin x
2
and Y
2
cos x
2
are linearly in
dependent functions, but that their Wronskian vanishes at
x Why does this imply that there is .diferential
equation of the form ,. ,. 0, with both
,and ,continuous everywhere, having both YI and Y
2
as
solutions?
32. Let YI and Y
2
be two solutions of .. ..
. 0 on an open interval where . B, and L
are continuous and .. is never zero. (a) Let W
W(YI , Y
2
) . Show that
Then substitute for .and .from the original difer
ential equation to show that
.
.. -B(x) W(x) .
..
(b) Solve this frst-order equation to deduce Abel's for
mula
W(x) K exp
B(x)
.X ,
..
where K is a constant. (c) Why does Abel
'
s formula
imply that the Wronskian W (YI , Y
2
) is either zero every
where or nonzero everywhere (as stated in Theorem 3)?
.,,...5 ...6 ...........
..,........33 ....42. .
........,x.
33. 0 34. 0
35. 0 36. 0
37. 0 38. . 0
39. .. 0 40. .0
41. 0 42. 0
....43 ....48 .......
. ..........., ..,..
.0 ......,.........
,...
43. . CI C
2
e-
1
0
x
45. .=

c
2
xe-
I
Ox
46. .

C
2
e
l
OOx 47. . CI
C
2
X
48. .

,
.49 ...50 ..........
0 .......2. 1. 6 ...2. 1. 7.
49. Find the highest point on the solution curve with I
and (0) in Fig.
50. Find the third-quadrant point of intersection of the solu
tion curves shown in Fig.
Z. Z General Sol uti ons of Li near Equations 1 1 3
51. A second-order Euler equation is one of the fonn conclude that a general solution of the Euler equation in
(22) is . = \
r
(
+ \
r
,
,
..

+ .+ = 0 (22)
where . are constants. (a) Show that if x 0,
then the substitution U = ln x transfons Eq. (22) into
the constant-coeffcient linear equation
........U = In x 51 ....
....x 0) ...,......52-
56.
.

.
.
dv
2
+ .
.
+ = 0 (23)
52. x
2
y
"
+ . = 0 53. .

+ . 1 2 0
with independent variable U. (b) If the roots and r of
the characteristic equation of Eq. (2) are real and distinct,
54. ..

+ . = 0 55. x
2
y
"
+ .= 0
56. .

. + .= 0
Exampl e 1
We now show that our dlscusslon ln Sectlon 2. I ofsecond-order |lnear equatlons
genera|lzeslnavery natura| waytothegenera| nth-order linear dlerentla|equa-
tlonoftheform
Po (x) y(
n
) +PI ,x; ,

-
-
,

+

+P
n
-
I (X) Y' +P
n
(x) y = F(x) . ( l )
Un|essotherwlsenoted,wewl||a|waysassumethatthecoemclentfunctlons i

(x)
and F(x) are contlnuous on someopen lnterva| l (perhapsunbounded) wherewe
wlshtoso|vetheequatlon. Undertheaddltlona|assumptlonthat io (x) =cateach
polntofl , wecan dlvldeeach termlnEq. ( I ) by P, (x) toobtalnanequatlonwlth
|eadlngcoefnclentI , oftheform
y(
n
) +PI
,x;,

-
-
,

+

+P
n - I (X) Y' +P
n
(x)y = ](x) . (2)
Thehomogeneous |lnearequatlonassociated with Eq.(2)ls
y(
n
) +PI ,x; ,

-
-
,

+

+P
n
-
I (X)y' +P
n
(x) y = c ,;
Justaslnthesecond-ordercase, a/o-o,ccosnth-order|lneardlerentla|equa-
tlon hastheva|uab|epropertythatanysuperposltlon,or//c..o-//./o ofso-
|utlons ofthe equatlon ls agaln a so|utlon. Theproofofthefo||owlng theorem ls
essentla||ythesamearoutlneverlncatlonasthatofTheorem l ofSectlon2. l
THEOREM 1 Pri nci pl e of Superposition for Homogeneous
Equations
LetYl , Y2 , . . . , Y
n
beso|utlonsofthehomogeneous|lnearequatlonln,;on
thelnterva| iIfCl , C2 , . . . ,c
,
areconstants, thenthe|lnearcomblnatlon
Y ~ CI YI +C2Y2 +

+C
n
Y
n
ls a|soaso|utlonofEq. ,;onl .
Itl s easytoverlfythatthethreefunctlons
YI (X) = c
-
3
x , Y2 (X) = cos 2x, and Y
3
(X) =sln 2x
(4)
1 1 4 Chapter Z Li near Equati ons of Hi gher Order

4
Z
-Z
-4
-
o Z 4 & I 0
x
FIGURE 2.2. 1. The particular
solution y(x) =

+ cos 2x - 2 sin 2x.


cXump| e 1
COlI|lUCO
area||so|utlonsofthehomogeneousthlrd-orderequatlon
Y (
3
) +3 y" +4 y' +1 2 y =c
on the entlre rea| |lne. Theorem 1 te||s us that any |lnear comblnatlon ofthese
so|utlons, suchas
y (x) =-3YI (x) +3Y2 (X) - 2Y
3
(X) =
.
3c
3

+3 cos2x - 2 sin 2x,


ls a|so a so|utlonon the entlrerea| |lne. We wl|| see that, converse|y, every so|u-
tlonofthedlerentla|equatlonofthlsexamp|elsa|lnearcomblnatlonofthethree
partlcu|arso|utlonsYJ , Y2 , andY
3
. Thus agenera| so|utlonls glven by

Existence and Uniqueness of Solutions


We sawlnSectlon2. 1 thatapartlcu|arso|utlonofasc.oaoac|lneardlerentla|
equatlon ls determlnedbyolnltla| condltlons. Slml|ar|y, apartlcu|arso|utlonof
an nth-order|lneardlfferentla| equatlon lsdetermlnedbylnltla|condltlons. The
fo||owlngtheorem,provedlntheAppendlx,lsthenatura|genera|lzatlonofTheorem
2 ofSectlon2. 1 .
THEOREM Z Existence and Uniqueness for Linear Equations
Suppose thatthemnctlons
;, , ;
:
, , ;, ,and] are contlnuous ontheopen
lnterva| icontalnlngthepolnt . Then, glvennumbersOg, O , . . . , O

, the
nth-order|lnearequatlon,Eq. (2
y(
n
) +;I (x) y(
n
-
I
)
+
. . .
+;,
-
,(x) y' +;, ,x; y=](x;
has a unlque ,that l s, one and on|y one) so|utlon onthe entlre lntera| i that
satlsnesthelnltla|condltlons
y,c) = bo,
y ,c;=bI , (5)
Equatlon (2) andthecondltlons ln (5) constltute annth-order initial value
problem. Theorem 2 te||sus thatany such lnltla| va|ue prob|em has aunlque so-
|utlononthe/o/clnterva| iwherethecoefnclentfunctlonsln(2) arecontlnuous.
Itte||susnothlng,however, abouthowtonndthls so|utlon. InSectlon2. 3 wewl||
see how to construct exp|lclt so|utlons oflnltla| va|ueprob|ems ln the .os.
coefnclentcasethatoccursoftenlnapp|lcatlons.
We sawear|lerthat
y(x) =
.
3c
-3

+3 cos2x - 2 sln2x
ls aso|utlonof
y<
3
) +3y" +4y' +1 2y =c
ontheentlrerea||lne. Thlspartlcu|arso|utlonhaslnltla|va|uesy ,c;=c,y' ,c;=5,
andy" (O) = -39, andTheorem2 lmp|lesthattherelsnootherso|utlonwlththese
samelnltla| va|ues. Notethatltsgraph ,ln Flg. 2. 2. 1 ) |ooks perlodlcontherlght.
Indeed, becauseofthe negatlve exponent, weseethat y (x) 3 cos2x - 2 sln2x
for|argeposltlvex
.

a
4
J
Z
0
- I
- Z
- J
-4
- a
-! 0 Z
x
Z. Z General Sol uti ons of Li near Equati ons 1 1 5
Remark: Becauseltsgenera|so|utlonlnvo|vesthethreearbltraryconstants
., ..

. and .

, the thlrd-orderequatlon ln Examp|e 1 has a threefo|d lnnnlty of


so|utlons, lnc|udlngthreefaml|lesofespecla||yslmp|eso|utlons.
y(x) =., c
-

(obtalnedfromthegenera| so|utlonwlth.

=.

=c;.
y(x) =.

cos2x (wlth., =.

=c;. and
y(x) =.

sln2x (wlth., =.

=c;
A|tematlve|y, Theorem 2 suggestsathreefo|dlnnnltyofpartlcu|arso|utlonscorre-
spondlngtolndependentcholcesofthethreelnltla|va|uesy ,c; =/,.y' ,c; =/, .
andy"(O) =/

Flgures2. 2. 2 through 2. 2. 4 l||ustratethreecorrespondlngfaml|les


ofso|utlonsforeachofwhlch, twoofthesethreelnltla|va|uesarezero.
Z
- I
-Z
J 4
- J
-Z - ! 0
,
= J

`
,
= -J
Z
x
I
0. o
0.
0, 4
0. Z
0
- 0. Z
-0, 4
- 0,
- 0. o
J 4
- !
-I 0
,= J
,
= -J

Z J 4 a
FIGURE 2.2.2. Solutions of
y(
3
) + 3y
"
+ 4y' + 1 2y 0 with
y' (0) y
"
(0) 0 but with
diferent values for yeO) .
FIGURE 2.2.3. Solutions of
y(
3
) + 3y
"
+ 4y' + 1 2y 0 with
yeO) y
"
(O) 0 but with
diferent values for y' (O) .
FIGURE 2.2.4. Solutions of
y(
3
) + 3y" + 4y' + 1 2y 0 with
yeO) y' (O) 0 but with
different values for y
"
(O) .
cXump| eZ
Note that Theorem 2 lmp|les that the /././ so|utlon y(x) = cls the on|y
so|utlonofthe/o-o,ccosequatlon
y(
n
) +PI (x) y(
n
-
I
) +

+P
n
-
I (X) y' +P
n
(x) y =c (3)
thatsatlsnesthe/././lnltla|condltlons
y(a) =y' (a) = =y<
n
-
I
) (a) =O.
" " - - " " " - - " - --- - _ . .- - - -
Itlseasytoverlfythat
aretwodlfferentso|utlonsof
x
2
y" - 4xy' +6y =c.
andthatboth satlsfy the lnltla| condltlons y(O) = y' (O) = O. Whydoes thlsnot
contradlcttheunlquenesspartofTheorem2? ltlsbecausethe|eadlngcoefnclentln
thlsdlerentla|equatlonvanlshesatx =c.sothlsequatloncannotbewrlttenlnthe
formof Eq.(3) wlthcoemclentfunctlons.o/osonanopen lnterva|contalnlng
thepolntx =O.
1 1 6 Chapter Z Li near Equati ons of Hi gher Order
Linearly Independent Solutions
Onthebaslsofourknow|edgeofgenera|so|utlonsofsecond-order|lnearequatlons,
weantlclpatethatagenera|so|utlonofthe/o-o,ccosnth-order|lnearequatlon
y<
n
) +PI (X) y(
n
-
I
) +

+P
n
- I (X) y' +P
n
(X) Y = c
(3)
wl||bea|lnearcomblnatlon
(4)
whereYI , Y2 , . . . , Y
n
arepartlcu|arso|utlonsofEq.(3). Butthesepartlcu|arso|u-
tlonsmustbesufnclent|ylndependentthatwecana|wayschoosethecoemclents
. C2 , . . . , C
n
ln (4) to satlsfy arbltrary lnltla| condltlons ofthe form ln (5). The
questlon ls thls. What shou|d be meant by /acpcac.cofthreeormore tunc-
tlons!
Reca||thatofunctlonsII andare|lnear|yacpcaclfonelsaconstant
mu|tlp|eoftheother, thatls, lfelther ] =kh or=kil for someconstantk. lf
wewrltetheseequatlonsas
, I ) )
|
+( -k) h =c or
,k) II + ,i ; =c
weseethatthe|lneardependenceofII andlmp|lesthatthereexlsttwoconstants
CI andC2 o/o/,casuchthat
,6)
Converse|y,lfCI andC2 arenotboth zero, thenEq. ,6) certaln|ylmp|lesthatII and
are|lnear|ydependent.
lnana|ogy wlth Eq. ,6), we say thatfunctlons I" . . . . , I
n
are//c./y
acpcacprovldedthatsomeo/././|lnearcomblnatlon
ofthemvanlshesldentlca||y, o/././meansthato.//ofthecoemclentsCI , C2 ,
, C
n
arezero,a|thoughsomeof themmaybezero).
DEFI NITI ON Linear Dependence of Functions
Thefunctlons ], ,
" ,
In
aresaldtobelinearly dependent onthelnterva|
iprovldedthatthereexlstconstantsCl , C2 ,
.
. . , C
n
nota||zero suchthat
,:;
oni,thatls,
fora||xlnl
Ifnot a|| thecoefnclents lnEq. ,:;are zero, thenc|ear|y wecan so|veforat |east
oneofthefunctlonsasa|lnearcomblnatlonoftheothers, andconverse|y. Thusthe
functlonsII , .. . .
, I
n
are|lnear|ydependentlfandon|ylfat|eastoneofthemls
a|lnearcomblnatlonoftheothers.
cXump| e
Z. Z General Sol uti ons of Li near Equati ons 1 1 7
Thefunctlons
),
(x) = sln 2x, _(x) = sln x cos x, and ](x) =e

are|lnear|ydependentontherea||lnebecause
, i ; ), +( -2) ]+(0) ] = 0
(bythefaml|lartrlgonometrlcldentlty sln2x = 2slnxcosx) .
Thefunctlons),

],
" " "
.)-
areca||edlinearly independent onthelnterva|
l provldedthattheyarenot|lnear|ydependentthere. Equlva|ent|y,theyare|lnear|y
lndependentoniprovldedthattheldentlty
.
,,+ .

J
+ + .
- )-
= 0
ho|dsonion|ylnthetrlvla|case
.,
-
.

=
. . .
= .
-
= 0,
:;
that l s, onontrlvla| |lnearcomblnatlon ofthesefunctlons vanlsheson l . Putyet
anotherway,thefunctlons),

.. . . .)-
are|lnear|ylndependentlfnooneofthem
ls a|lnearcomblnatlonoftheothers. (Why!)
Sometlmes one can show that glven functlons are|lnear|y dependent by
nndlng,aslnExamp|e3, nontrlvla|va|uesofthecoemclentssothatEq. (7) ho|ds
Butlnorderto showthatglvenfunctlonsare|lnear|ylndependent,wemustprove
thatnontrlvla|va|uesofthecoefnclents..obefound,andthlsls se|domeasyto
do ln any dlrectorobvlousmanner.
Fortunate|y,lnthecaseofso|utlonsofahomogeneousnth-order|lnearequa-
tlon, therelsatoo|thatmakesthedetermlnatlonofthelr|lneardependenceorlnde-
pendencearoutlnematterlnmanyexamp|es. Thls too| ls theWronsklandeterml-
nant, whlchwe lntroduced(forthe case = 2) lnSectlon2. 1 . Supposethatthe
functlons),

], . .
"
.)-
areeach- 1 tlmesdlerentlab|e. ThenthelrWronskian
lsthe Z determlnant
u =
]
)
)
-
)
(8)
We wrlte u, ), .],
"
. . .)
-
; or u,x; . dependlng on whetherwewlshtoempha-
slzethefunctlonsorthepolntx at whlch thelrWronsklan ls tobe eva|uated. The
WronsklanlsnamedafterthePo|lshmathematlclanJ. M. HWronskl( l 778-1 853).
WesawlnSectlon2. I thattheWronsklanof two|lnear|ydependentfunctlons
vanlshesldentlca||y. Moregenera||y, /eus//.o)//e./,aepeae].
/os], ], . . . .)-
/s/ae/..//,,e Toprovethls, assumethatEq. (7) ho|dson
the lnterva| l for somecholceoftheconstants., ..

, , .
-
nota|| zero. Wethen
dlfferentlatethlsequatlon- I tlmeslnsuccesslon,obtalnlngtheequatlons
., ,x;+.

J,x;
._ ,x;+.

) ,x;
+

+
+

+
.
- )-
,x;=0,
.
- ) ,x;
-
0,
(9)
1 1 8 Chapter Z Li near Equati ons of Hi gher Order
Exampl e 4
whlch ho|d for a|| x ln l . We reca|| from |lnear a|gebra that a system of |ln-
ear /o-o,eeosequatlons ln unknowns has a nontrlvla| so|utlon lfand on|y
lfthedetermlnantofcoefnclents vanlshes. InEq. (9) the unknowns arethe con-
stants ., . .

. , c,
and the determlnant ofcoefnclents ls slmp|y theWronsklan
w,), .. .]
-
;eva|uatedatthetyplca|polntx ofi Becauseweknowthatthe
.

arenota||zero,ltfo||owsthat w(x) = 0,aswewantedtoprove.


Therefore, to show thatthefunctlons ], . . . ]
-
are//e./,/aepeae
onthelnterva|I,ltsumcestoshowthatthelrWronsklanlsnonzeroatustonepolnt
ofl .
Show that the functlons
y, ,x; = e
-

, ,(x) = cos 2x, and y

,x; = sln 2x (of


Examp|e 1 ) are|lnear|ylndependent.
Sol uti on ThelrWronsklanls
Exampl e
cos 2x
w = .
e

-2 sln 2x
sln 2x
2 cos 2x
,
e
-

-4 cos2x -4 sln2x
-2sln2x 2 cos 2x
-4 cos2x -4 sln2x
+e
-

cos 2x sln 2x
-4 cos2x -4 sln2x
cos 2x sln 2x
= z:e
-

= O
.
-2sln2x 2cos2x
Because w = 0everywhere, lt fo||owsthat y, .y

.andy

are|lnear|ylndependent
onanyopenlnterva|(lnc|udlngtheentlrerea||lne) .
Shownrstthatthethreeso|utlons
ofthethlrd-orderequatlon
x

- x

+2x,- 2,= 0 ( I 0)
are|lnear|ylndependentontheopen lnterva|x > O. Then nnd apartlcu|arso|utlon
ofEq.( I 0)thatsatlsnesthelnltla|condltlons
,, i ; = . , , i ; =2, ,

, i ; = I . ( I I )
Sol ution Notethatforx > 0,wecou|ddlvldeeachtermln( I 0) byx

toobtalnahomoge-
neous|lnearequatlonofthestandardformln,; WhenwecomputetheWronsklan
ofthethreeglvenso|utlons, wenndthat
x x|n x x

w =
I +|n x 2x
= x.
0
1
2
x
Z. Z General Sol uti ons of Li near Equati ons 1 1 9
Thus W (x) ,cforx > c,soYl , Y2 , andY
3
are|lnear|ylndependentonthelnterva|
x > O. Tonndthe deslredpartlcu|arso|utlon, welmposethe lnltla|condltlons ln
( 1 1 ) on
y(x) =Cl X + C2X In x
y (x) =c
Thl syle|dstheslmu|taneousequatlons
y( 1 ) =Cl
+ C
3
=3,
y' ( 1 ) =Cl + C2 + 2C
3
=2,
y , i ; =
we so|ve to nndCl = 1 , C2 = -3, and C
3
= 2. Thus the partlcu|ar so|utlon ln
questlonl s
y (x) =x - 3x In x + 2x
2
.
Provlded that W(Yl , Y2 , . . . , Yn
) ,c,lt tums out (Theorem 4) that we can
a|waysnndva|uesofthecoefnclentslnthe|lnearcomblnatlon
Y =Cl YI + C2Y2 + . .
.
+ cn
Y
n
thatsatlsfy anyglvenlnltla|condltlonsoftheformln(5). Theorem3 provldesthe
necessarynonvanlshlngofW lnthecase of|lnear|ylndependentso|utlons.
THEOREM 3 Wronskians of Sol utions
SupposethatYl , Y2, . . . , Y
n
aren so|utlonsofthehomogeneousnth-order|lnear
equatlon
y(
n
)
+ pj(x) /
n
-
l
)
+
. . .
+P
n
-
I
(X) Y' + P
n
(x) y =c (3)
onanopenlnterva|I, whereeachPi lscontlnuous. Let
u= W(Yl , Y2 ,

, Y
n
) .
(a) IfYl , Y2, . . . , Yn
e|lnear|ydependent,then W = conI.
(b) IfYl , Y2 , . . . , Y
n
are|lnear|ylndependent,then W ,cateachpolntofI.
Thusthereareusttwoposslbl|ltles. Elther W =ceverywhereonI, or w = c
everywhereonI .
Proof: We have a|ready proven part (a) . To prove part (b), ltl s sumclent
toassumethat W ,.; = cat somepolntofI , and showthls lmp|lesthattheso|u-
tlons Yl , Y2 , . . . , Y
n
are|lnear|ydependent. Butw(a) ls slmp|ythedetermlnantof
1 20 Chapter Z Li near Equati ons of Hi gher Order
coemclentsofthesystemofhomogeneous|lnearequatlons
CI Yl (a) +
CI Y(a) +
C2Y2 (a) +
. . .
+
.2y (a;+
. . .
+ .
-
y ,.;~ c,
CI y
-

(a;+ .
:
y
-
-

(a) +
. . .
+ .
-
y
-
-
,

,.;~ c
( 1 2)
ln theunknownsCl , C2 , ..
-
Because W(a) ~ c,thebaslcfact from |lnear
a|gebra quoted ustafter (9) lmp|les that the equatlons ln ( I 2) have a nontnvla|
so|utlon. Thatls, thenumbers.
, .C2 , ..
-
arenota||zero.
We nowusetheseva|uestodennethepartlcu|arso|utlon
( 1 3)
of Eq. (3). The equatlons ln ( 1 2) then lmp|y that I satlsnes the tnvla| lnltla|
condltlons
Y(a) ~ Y' (a) ~
. . .
= y
(
n
-
l
) (a) ~ O.
Theorem 2 (unlqueness) therefore lmp|les that I(x; = con I. In vlew of( 1 3)
andthe factthat Cl , C2 , ..
-
are nota|| zero, thls ls the deslredconc|uslonthat
the so|utlons YI , Y2, . y-
are |lnear|y dependent. Thls comp|etes the proofof
Theorem3.
General Solutions
We cannowshow that, glvenanynxedsetof|lnear|y lndependentso|utlonsof
a /o-o,eeosnth-order equatlon, e.e(other) so|utlon oftheequatlon canbe
expressed as a |lnear comblnatlon ofthose partlcu|ar so|utlons. Uslngthefact
fromTheorem3 thattheWronsklanof|lnear|ylndependentso|utlonsls nonzero,
theproofofthefo||owlngtheoremlsessentla||ythesameastheproofofTheorem
:ofSectlon2. 1 (thecase~ 2).
THEOREM 4 General Sol utions of Homogeneous Equations
LetYl . Y2 , .y-
be|lnear|ylndependentso|utlonsof thehomogeneousequa-
tlon
y

+;,,x; y

+
.
.
.
+
;-

, ,x; y
+
;-
,x; y c ,;
onanopenlnterva| wherethe
;,arecontlnuous. IfI lsanyso|utlonwhatsoever
ofEq. (3), thenthereexlstnumbers., .C2 , ..
-
suchthat
fora||xln I .
Thus e.eso|utlonofahomogeneousnth-order|lneardlerentla| equatlon
lsa|lnearcomblnatlon
Y ~ C
I YI + C2Y2 +

+ .
- y-
ofany glven |lnear|y lndependentso|utlons. Onthlsbaslswe ca|| sucha |lnear
comblnatlonageneral solution ofthedlerentla|equatlon.
Exampl e
Z. Z General Sol uti ons of Li near Equati ons 1 21
Accordlng to Examp|e1,the partlcu|arso|utlons y,
,x; = e
-

, )
:
,x; = coszx,
andy

,x;=slnzxofthe|lneardlerentla|equatlony

+y+:y+i zy=care
|lnear|ylndependent. Now Theoremzsaysthatglvenb,,b, , andb
:
thereexlsts
a partlcu|ar so|utlon y,x; satlsfylng the lnltla|condltlons y,c; = b,, y ,c; = b, ,
andy,c; = b
:
Hence Theorem :lmp|lesthat //spartlcu|arso|utlon ls a|lnear
comblnatlon of), , )
:
, and y

Thatls, there exlstcoemclentsc, , c


:
, andc

such
that
y,x; =., e
-

+c
:
coszx+c

slnzx
Upon successlvedlerentlatlonandsubstltutlonofx =c.wedlscoverthattonnd
thesecoemclents,weneedon|yso|vethethree|lnearequatlons
., + c
:
=b,,
-., +z.

= b, ,
., - 4c
:
=b
:
.
(Seetheapp|lcatlonforthls sectlon. )
Nonhomogeneous Equations
We nowconsldertheo/o-o,eeosnth-order|lneardlerentla|equatlon
y

+;, ,x; y

-
-
,

+

+;-
-
, ,x; y
+;-
,x; y=),x;
wlthassoclatedhomogeneousequatlon
y

+;, ,x; y

-
-
,

+

+;-
-
, ,x; y+;-
,x; y=c

,z;
,;
Suppose that a slng|e nxed partlcu|ar so|utlon )
o
ofthe nonhomogeneous
equatlonln,z;lsknown,andthatrlsanyotherso|utlonof Eq.,z;Ify,=r- y

,
thensubsltutlonofy,lnthedlerentla|equatlonglves(uslngthe|lnearltyofdler-
entlatlon)
y
-

+;, y
-
-
,

+

+;-
-
, y
+;- y,
= ,
,
-

+;,
,
-
-
,

+

+;-
-
, r

+;-
r]
- ,y
-

+;, y
-
-
,

+

+;-
-
, y
+;-y

]
=),x; ),x;=0
Thusy,=r-)
o
lsaso|utlonoftheassoclatedhomogeneousequatlonln,; Then
r=y,+y

. , i :;
andltfo||owsfromTheorem:that
, i ;
where ), , )
:
, . . , y-
are |lnear|y lndependent so|utlons ofthe assoclated /oo
,eeosequatlon. Weca|| y,acomplementary function ofthenonhomogeneous
equatlonandhavethusprovedthata,eem/so//oofthenonhomogeneousequa-
tlonln,z;ls the sumofltscomp|ementaryfunctlony,andaslng|epartlcu|arso|u-
tlon)
o
ofEq.,z;
1 22 Chapter Z Li near Equati ons of Hi gher Order
THEOREM b Sol uti ons of Nonhomogeneous Equations
LetY
p
be a particu|ar so|ution ofme nonhomogeneous equatlon ln ,z;on an
openinterva| I wherethefunctlons, and] arecontlnuous. LetYI , Y
2
, .
.
. , Yn
be|lnear|ylndependentso|utlonsoftheassociatedhomogeneousequatlonin(3).
IfI l sany so|utlonwhatsoeverofEq. (2)on I, thenthereexlstnumbersCI , C2 ,
g C
n
suchthat
fora||x in I .
cXump| e Itls evldentthatY
p
=3x ls apartlcu|arso|utlonoftheequatlon
Y
"
+ 4y = I zx, ( l 7)
andthatYc (x) =CI coszx +C2 sln2x ls ltscomp|ementaryso|utlon Flndaso|utlon
ofEq. ( I 7)thatsatlsnesthelnltla|condltlonsy (O) =5, Y' ,c;=7.
Sol uti on Thegenera|so|utlonofEq. ( I 7)ls
y(x) =CI coszx + C2 slnzx + 3x .
Now
y' (x) =-z.,slnzx + 2C2 coszx + 3.
Hencethelnltla|condltlonsglve
y (O) = CI
=5 ,
y' ,c;=zC2 + 3 =7.
WenndthatCI =5 andC2 =zThusthedeslredso|utlonls
.1 ....6, ..........
...,.........
.............
..
1. f(x) = 2x, g(x) = 3x
2
, .ex) 5x - 8x
2
2. f(x) = 5, g(x) = - 3x
2
, .ex) = 1 0 1 5x
2
3. f(x) = 0, g(x) = sin x, .ex) =
4. f(x) = 1 7, g(x) = 2 sin
2
x, .ex) = 3 cos
2
x
5. f(x) = 1 7, g(x) cos
2
x, .ex) = cos 2x
6. f(x) = g(x) cosh x, .ex) = sinh x
.....12, ......,...
......,.....
.
y (x) =5 coszx + zslnzx + 3x
.

7. f(x) = 1 , g(x) = x, .ex) = x


2
; !he real line
8. f(x) = g(x) =

.ex) =

!he real line


9. f(x) = g(x) cos x, .ex) = sin x; the real line
10. f(x) = g(x) = x-
2
, .ex) = x
-2
ln x; x > 0
11. f(x) = x, g(x) = . .ex) = .

the real line


12. f(x) = x, g(x) = cos (ln x) , .ex) = sin (In x) ; x > 0
.13 ....20, .........
,.......,......
..,........,.....
.
13.

= 0; yeO) = 1 , leO) = (O) = 0;


YI = Y
2
=

Y
3

-2

14. y(
3
)
- 6y" + l l y' - 6y = 0; yeO) = 0, y' (O) = 0,
y" (O) = 3 ; YI = Y
2
=

Y
3
=

15. y<
3
)
- 3y" +3y' - y = 0; y eO) = 2, y' (O) = 0, y" (O) = 0;
YI = Y
2
= . Y
3
= .

16. y(
3
)
_5y" +8y' -4y = 0; yeO) = 1 , y' (O) = 4, y" (O) = 0;
YI = Y
2
=

Y
3
= .

17. y(
3
) + 9y' = 0; yeO) = 3, y' (O) = -1 , y" (O) = 2; YI = 1 ,
Y
2
= cos 3x, Y
3
= sin 3x
18. y(
3
) -3y" +4y' -2y = 0; y eO) = 1 , y' (O) = 0, y" (O) = 0;
YI = Y
2
= cos x, Y
3
= sin x.
19. x
3
y(
3
)
- 3x
2
y" + 6xy' - 6y = 0; y( l ) = 6, y' ( 1 ) = 1 4,
y" ( 1 ) = 22; YI = X, Y
2
= x
2
, Y
3
= x
3
20. x
3
y(
3
) + 6x
2
y" + 4xy' - 4y = 0; y( 1 ) = 1 , y' ( 1 ) = 5,
y" ( I ) = - 1 1 ; YI = X, Y
2
= X
-2
, Y
3
= x
-2
In x
.21 ....24, ......, .
,...,...Yo ...,....
.yp .........,....
..
21. y" + y = 3x; yeO) = 2, y' (O) = -2;
Yc = C
I
cos x + C
2
sin x; YP = 3x
22. y" - 4y = 1 2; yeO) = 0, y' (O) = 1 0;
Yc =

YP = -3
23. y" - 2y' - 3y = 6; y eO) = 3, y' (O) = 1 1 ;
Yc
=

YP = -2
24. y" - 2y' + 2y = 2x; yeO) = 4, y' (O) = 8;
Yc = cos x +

sin x; YP = x + 1
25. Let Ly = y" + ,+ , Suppose that YI and Y
2
are two
functions such that
LYI
= f(x) and LY
2
= g(x) .
Show that their sum y = YI + Y
2
sati sfes the nonhomoge
neous equation Ly = f(x) + ..
26. (a) Fi nd by inspection particular solutions of the two non
homogeneous equations
y" + 2y = 4 and y" + 2y = 6x.
(b) Use the method of Problem 25 to fnd a particular so
lution of the diferential equation y" + 2y = 6x + 4.
27. Prove directly that the functions
fl (x) = 1 , hex) = x, and hex) = x
2
are linearly independent on the whole real line. ....
Assume that C
I
+ C
2
X + C
3
x
2
= O. Diferentiate this
equation twice, and conclude from the equations you get
that C
I
= C
2
= C
3
= 0. )
28. Generalize the method of Problem 27 to prove directly that
the functions
are linearly independent on the real line.
29. Use the result of Problem 28 and the defnition of linear
independence to prove directly that, for any constant the
functions
fo (x) = !ex) = xe'x
'
are linearly independent on the whole real line.
Z. Z General Sol uti ons of Li near Equati ons 1 23
30. Verify that YI
= x and Y
2
= x
2
are linearly independent
solutions on the entire real line of the equation
x
2
y" - 2xy' + 2y = 0,
but that (x, x
2
) vanishes at x = O. Why do these obser
vations not contradict part (b) of Theorem 3?
31. This problem indicates why we can impose initial
conditions on a solution of an nth-order linear differential
equation. (a) Given the equation
32.
y" + ,+ ,= 0,
explain why the value of .is determined by the values
of .and . (b) Prove that the equation
y" - 2y' - 5y = 0
has a solution satisfying the conditions
y eO) = 1 , y' (O) = 0, and y"(O) = C
if and only if = 5.
Prove that an nth-order homogeneous linear diferential
equation satisfying the hypotheses of Theorem 2 has lin
early independent solutions YI
, Y
2
, . . . , Yn . ....
Let Yj be the unique solution such that
(
i
-
I
)
( )
-
1 Yj
.

and

.= 0 if k


33. Suppose that the three numbers

and

are dis
tinct. Show that the three functions exp(rl x) , exp(r
2
x),
and exp(r
3
x) are linearly independent by showing that
their Wronskian
is nonzero for all x.
34. Assume as known that the Vandermonde determinant

n
v =

n
n
-
I

n
-
I

-
I
is nonzero if the numbers l ,

. . . , r
n
are distinct. Prove
by the method of Problem 33 that the functions
f
(x) = exp(rj x) , 1
are linearly independent.
1 24 Chapter Z Li near Equati ons of Hi gher Order
35. According t o Problem 32 of Section 2. 1 , the Wronskian
(YI , Y
2
) of two solutions of the second-order equation
37. Before applying Eq. ( 1 9) with a given homogeneous
second-order linear diferential equation and a known so
lution Yl (x) , the equation must frst be written in the form
of ( 1 8) with leading coefcient I in order to correctly
determine the coefcient function p(x) . Frequently it is
more convenient to simply substitute Y = V(X) Yl (X) in
the given diferential equation and then proceed directly
to fnd v ex) . Thus, starting with the readily verifed solu
tion YI (x) = x
3
of the equation
36.
Y" +
PI (X) Y
'
+
P2
(X) Y = 0
is given by Abel ' s' s formula
W(x) = K exp -
PI
(X) dX
for some constant K. It can be shown that the Wronskian
of solutions YI > Y
2
, . . . , Y
n
of the nth-order equation
y(
n
)
+
P
I (x) y
<n
-
I
)
+
. . .
+
P
n
-
I (X) Y
'
+
P
n
(x) y = 0
satisfes the same identity. Prove this for the case = 3
as follows: (a) The derivative of a determinant of func
tions is the sum of the determinants obtained by separately
diferentiating the rows of the original determinant. Con
clude that
YI Y
2
=
Y
;
y y

yl
3
)
y2
)
yj
3
)
(b) Substitute for Y
)
, yi
3
)
, and yj
3
)
from the equation
y
(
3
)
+
P
I Y" +
P
2
Y' +
P
3
Y = 0,
and then show that = -
P
I Integration now gives
Abel ' s formula.
Suppose that one solution YI (x) of the homogeneous
second-order linear diferential equation
Y" + p(x) y' + q (x) y = 0 ( 1 8)
is known (on an interval where
P
and ,are continuous
functions). The method of reduction of order consists
of substituting Y
2
(X) = V(X) Y
I
(x) in ( 1 8) and attempting
to determine the function v ex) so that Y
2
(X) is a second
linearly independent solution of ( 1 8) . Afer substituting
y = V(X) YI (x) in Eq. ( 1 8), use the fact that Yl (x) is a
solution to deduce that
( 1 9)
If YI (x) is known, then ( 1 9) is a separable equation that
is readily solved for the derivative v' (x) of vex) . Integra
tion of v' (x) then gives the desired (nonconstant) function
vex) .
x
2
y" - 5xy' + 9y = 0 (x > 0) ,
substitute Y = vx
3
and deduce that xv" + v' = O.
Thence
solve for v ex) = In x, and thereby obtain (with = 1 )
the second solution Y
2
(X) = x
3
ln x.
.......42, .., .,....
..Yl .........
........,...
Y
2
38. x
2
y" + xy' - 9y = 0 (x > 0) ; YI
(x) = x
3
39. 4y" - 4y' + Y = 0; Yl (x) = e
x
/
2
40. x
2
y" - x(x + 2) y' + (x + 2)y = 0 (x > 0) ;
YI (x) = x
41. (x + l ) y" - (x + 2) y' + y = 0 (x > -1 ) ; Yl (x) =
x
42. ( 1 - x
2
) y" + 2xy' - 2y = 0 ( -1 x 1 ) ; Yl (X) = x
43. First note that YI (x) = x is one solution of Legendre' s
equation of order 1 ,
( 1 - x
2
) y" - 2xy' + 2y = O.
Then use the method of reduction of order to derive the
second solution
x 1 + x
Y
2
(X) = 1 - - In (for -l x 1 ) .
2 I - x
44. First verify by substitution that Yl (x) = x -1
/
2
cos x is one
solution (for x > 0) of Bessel ' s equation of order ,
Then derive by reduction of order the second solution
Y
2
(X) = X-1
/
2
sin x.
Homogeneous Eq!:atims with Constant Coeficents
In Sectlon 2. 2 we saw thatagenera| so|utlonofannth-orderhomogeneous|lnear
equatlonls a|lnearcomblnatlonofn |lnear|ylndependentpartlcu|arso|utlons, but
we sald|ltt|e about how actua||yto nnd even a slng|e so|utlon. The so|utlonofa
|lneardlerentla|equatlon wlth .././/ecoefnclents ordlnarl|yrequlresnumerlca|
methods (Chapter6) orlnnnlteserlesmethods(Chapter3) . Butwecannowshow
how to nnd, exp|lclt|yandlnaratherstralghtforward way, n |lnear|y lndependent
so|utlons ofa glven nth-order |lnear equatlon lflthas .os.coefnclents. The
Z. Homogeneous Equati ons wi th Constant Coeffci ents 1 25
genera|suchequatlonmaybewrlttenlntheform
( I )
wherethecoefnclents.,.., ,.

. ..
-
arerea|constantswlth.
-
=O
.
The Characteristic Equation
We nrst|ookforas/,/eso|utlonofEq.( 1 ), andbeglnwlththeobservatlonthat
,2)
so any derlvatlve of e ls a constant mu|tlp|e ofe Hence, lfwe substltuted
,= elnEq. ( 1 ), eachtermwou|dbeaconstantmu|tlp|eofe .wlththeconstant
coefnclentsdependlngonandthecoefnclents.

Thl ssuggeststhatwetrytonnd
sothata||thesemu|tlp|esofe
wl||havesumzero,lnwhlchcase,= ewl||be
aso|utlonof Eq.( I ).
Forexamp|e,lnSectlon2. I wesubstltuted,= e
lnthesecond-orderequa
tlon
.,

+/,+.,= c
toderlvethecharacterlstlcequatlon
.

+ /+ . = c
thatmustsatlsfy.
Tocarryoutthlstechnlquelnthegenera|case,wesubstltute,= elnEq.( I ),
and wlththealdofEq. (2) wenndtheresu|ttobe
thatl s,
e
.
-

-
+.
-
,
-
-

+

+.

+., +.,)
= O.
Becauseel sneverzero,weseethat,= ewl||beaso|utlonof Eq.( I ) preclse|y
whenlsarootoftheequatlon
,3)
Thls equatlon ls ca||ed the characteristic equation orauxiliary equation ofthe
dlfferentla| equatlon ln ( 1 ). Our prob|em, then, ls

reduced to the so|utlon ofthls


pure|ya|gebralcequatlon.
Accordlng to the fundamenta| theorem of a|gebra, every nth-degree po|y-
nomla|such as the one ln Eq. (3)has n zeros, thoughnot necessarl|y dlstlnct
andnotnecessarl|y rea|. Flndlng the exact va|ues ofthese zeros maybedlmcu|t
orevenlmposslb|e,thequadratlcformu|alssufnclentforsecond-degreeequatlons,
but for equatlons ofhlgher degree we may need elther to spot a fortultous fac
torlzatlon or to app|y a numerlca| technlque such as Newton' s method (or use a
ca|cu|atorJcomputersolve command) .
1 26 Chapter Z Li near Equati ons of Hi gher Order
cXump| e 1
Distinct Real Roots
Whateverthemethodweuse,|etussupposethatwehaveso|vedthecharacterlstlc
equatlon. Thenwecan a|ways wrlteagenera| so|utlonofthedlerentla|equatlon.
Thesltuatlonl ss|lght|ymorecomp|lcatedlnthecaseofrepeatedrootsorcomp|ex
rootsofEq.(3), so|etusnrstexamlnetheslmp|estcaselnwhlchthecharacterlstlc
equatlonhasdlstlnct(notwoequa|)e./roots
, .

. .
-
Thenthefunctlons
area||so|utlonsofEq.( I ),and(byProb|em34ofSectlonz z;theseso|utlonsare
|lnear|ylndependentontheentlrerea||lne.Insummary,wehaveprovedTheoremI .
THEOREM 1 Disti nct Real Roots
If theroots, .

. ,
-
ofthecharacteristlcequatlonln,;arerea|anddlstlnct,
then
lsa genera|so|utlonof Eq. ( 1 ) .
.........
So|vethelnltla|va|ueprob|em
,

+ 3,

- i c,=c,
,,c;=:. , ,c;=c. ,

,c;=:c
(4)
Sol uti on Thecharacterlstlcequatlonoftheglvendlerentla|equatlonl s

- i c=c
We so|vebyfactorlng.
,

+ - i c;= ,+ ; ,- z;=c.
andsothecharacterlstlcequatlonhasthe three dlstlnctrea|roots = c. ~ -.
and=zBecausee

= I , Theorem I glvesthegenera|so|utlon
Thentheglvenlnltla|condltlonsyle|dthe|lnearequatlons
,,c;=., +
, ,c;=
, ,c;=
.

+ .

= :.
.

+ z.

= c.
z.

+ 4.

=:c
ln the coefnclents ., . .

. and.

The |asttwoequatlons glve , ,c; - z, ,c; =


.

=:c. so.

=z Then the secondequatlonglves.

=. andnna||ythenrst
equatlon glves., =c Thusthedeslredpartlcu|arso|utlonl s

Z. Homogeneous Equati ons with Constant Coeffci ents 1 27


Polyomial Diferential Operators
Iftherootsofthecharacterlstlcequatlonln(3)arenotdlstlncttherearerepeated
rootsthen wecannotproducen |lnear|y lndependent so|utlonsofEq. ( I ) by the
methodofTheorem I . Forexamp|e,lftherootsare I , 2, 2, and 2, weobtalnon|y
theofunctlonseande

Theprob|em,then,ls toproducethemlsslng|lnear|y
lndependentso|utlons Forthlspurpose,ltlsconvenlenttoadoptoperatornotatlon
andwrlteEq. ( I ) lntheformL,=c. wheretheoperator
ope.esonthen-tlmesdlerentlab|efunctlon,(x) toproducethe|lnearcomblna
tlon
L,=.
-
,

+ .
-
, ,

-
-
,

+
. . .
+ .

+ ., ,
+ .,,
of, andl ts nrst n derlvatlves. We a|so denote by n = a,ax the operatlon of
dlerentlatlonwlthrespecttox. so that
andsoon Intermsofn.theoperatorL ln(5) maybewrltten
(6)
andwewl||nndltusefu|tothlnkoftherlght-handsldelnEq.(6) asa(forma|)nth
degreepo|ynomla|lnthe varlab|e n, ltlsapolynomial diferential operator.
A nrst-degree po|ynomla| operator wlth |eadlng coemclent I has the forn
n .. where.ls area|number. Itoperateson afunctlon,= ,(x) toproduce
,n- .; ,=n,- .,=,- .,
Thelmportantfactaboutsuchoperatorsls thatanytwoofthem.oe
, n- .; , n /;,=, n- /; , n- .; , (7)
foranytwlcedlerentlab|efunctlon, = ,,x; Theproofoftheformu|aln(7) ls
thefo||owlngcomputatlon.
, n- .; , n- /;,= , n- .; ,,- /,;
=n,,- /,;- . ,,- /,;
=,
- ,/+ .; ,
+ ./,=,- ,.+ /;,
+ /.,
=n,,- .,;- /,,- .,;
=, n- /; ,,- .,;=, n- /; , n- .; ,
We seeherea|sothat , n- .; , n- /; = n

- , .+ /; n+ ./

Slml|ar|y, ltcan
be shown by lnductlon on the number of factors that an operator product of the
form , n- ., n- .
:
; ,n .
-
;expandsbymu|tlp|ylngoutandco||ectlng
coefnclentsln the same way as does an ordlnary product ,x - ., ; ,x - .

;
,x- .
-
;of|lnearfactors,wlthxdenotlngarea|varlab|e. Consequent|y,thea|gebra
ofpo|ynomla|dlerentla|operatorsc|ose|yresemb|esthe a|gebra ofordlnaryrea|
po|ynomla|s
1 28 Chapter Z Li near Equati ons of Hi gher Order
Repeated Real Roots
Letusnowconsldertheposslbl|ltythatthecharacterlstlcequatlon
.
-

-
+
.
-
,
-
-

+ .
.
. +
.
:

:
+ .,
+
.,=0 (3)
hasepe.earoots. Forexamp|e,supposethatEq.(3)hason|ytwodlstlnctroots,,
ofmu|tlp|lclty I and, ofmu|tlp|lcltyk =- I > I Then,afterdlvldlngby .
-
;
Eq. (3)canberewrlttenlntheform
,- ,- ,;=,- ,; ,-

=c
Slml|ar|y,thecorrespondlngoperatorL ln (6) canbewrltten as
L = ,n- ,n- ,;= ,n- ,; ,n .
theorderofthefactorsmaklngnodlerencebecauseoftheformu|aln(7) .
, s;
(9)
Twoso|utlonsofthedlfferentla|equatloniy=0arecertaln|yy,=e

and
y, = e
Thls l s, however, not sufnclent, we need k + 1 |lnear|y lndependent
so|utlons lnordertoconstructagenera| so|utlon, because the equatlonlsoforder
k
+
I . Tonndthemlsslngk- I so|utlons,wenotethat
iy= ,n- ,; , n- y}=c
Consequent|y, e.eso|utlonofthekth-orderequatlon
,n - y = o , I 0)
wl|| a|so be a so|utlon ofthe orlglna| equatlon iy = c Hence our prob|em ls
reducedtothatofnndlngagenera|so|utlonofthedlerentla|equatlonln, I 0) .
The fact that y, = e
ls one so|utlonofEq. , I 0) suggests thatwetry the
substltutlon
( 1 1 )
where,x;l s afunctlonyettobedetermlned. Observethat
,n-
, ;e

]= , n; e
+
,,
e

;- ,,e

;=,n; e

, I 2)
Upon kapp|lcatlonsofthls fact,ltfo||owsthat
( I 3)
foranysufnclent|ydlerentlab|efunctlon,x; Hencey=e wl||beaso|utlon
ofEq. , I 0)lfandon|ylfn

=cButthls lssolfandon|ylf
,x; =.,+.
:
x
+
.

x
:
+
. . . + ., x

-
,
.
apo|ynomla|ofdegreeatmostk - I . Henceourdeslredso|utlonofEq. , I 0)ls
y,x; =e

=,.,
+
.
:
x
+
.

x
:
+ . . .
+
.

;e

Inpartlcu|ar, weseeheretheaddltlona|so|utlonsxe .x
:
e . .x

e ofthe
orlglna|dlerentla|equatlon iy=c
Theprecedlngana|yslscan be carrled outwlththe operator n- , rep|aced
wlthanarbltrarypo|ynomla|operator.Whenthlslsdone, theresu|tlsaproofofthe
fo||owlngtheorem.
cXump| eZ
Z. Homogeneous Equati ons wi th Constant Coeffi ci ents 1 29
THEOREM Z Repeated Roots
Ifthecharacterlstic equationin (3) has arepeatedrootofmu|tip|icltyk, then
thepartofagenera|so|utionofthedierentia|equationin( I ) conespondingto
is oftheform
, I4)
We may observe that, according to Prob|em zofSectlon 2 2, the /tunc-
tlonse
.xe
.x

e . , andx

elnvo|vedln, I 4)are|lnear|ylndependenton
the rea| |lne. Thus a root ofmu|tlp|lclty /corresponds to /|lnear|y lndependent
so|utlonsofthedlerentla|equatlon.
. .. .. ... ... . . ..
Flndagenera|so|utlonofthenfth-orderdlerentla|equatlon

+
,

= 0.
Sol uti on Thecharacterlstlcequatlonls

- :
-
+

- :+ I) =

,- i ;

= 0.
It has the trlp|e root = 0 and the doub|e root = The trlp|e root * 0
contrlbutes
.,
e
,
+
.

xe
,
+ .


= ,
,
+ ,

x+
,

totheso|utlon,whl|ethedoub|eroot = contrlbutes,
-
e

+,

xe

Hencea
genera|so|utlonofthegiven dlerentla| equatlonls

Complex-Valued Functions and Euler


'
s Formula
Becausewehave assumed that the coefncients ofthedlerentla|equatlon andlts
characterlstlcequatlonarerea|, anycomp|ex(nonrea|)rootswl||occurlnconp|ex
conugatepalrsO +//whereO and/arerea|and/ = ,. Thlsralsesthequestlon
ofwhatmlghtbemeantbyanexponentia|suchase

To answer thls questlon, we reca|| from e|ementary ca|cu|us the Tay|or(or


MacLaurln)serlesfortheexponentla|functlon

-
e

= _ = i + + + + +
-
n ' 2 ' 3 ' 4'
Ifwesubstltute = /-lnthlsserlesandreca||that/

= -I , /

= -/ . /
-
= I , and
soon, weget
e

= _
,/ ;
-
-
n'
-

-
-
/

= i + / - - + + - "
2' 3' 4' 5'
=

+
-
-


+/

+
-
-

.
2' 4' 3 ' 5 '
1 30 Chapter Z Li near Equati ons of Hi gher Order
Becausethetworea| serleslnthe|ast|lnearetheTay|orserlesforcos-andsl n-,
respectlve|y,thlslmp|lesthat
e

-
= cos 0 +/sln 0. ( I 5)
Thlsresu|tls knownasEuler's formula. Becauseoflt, weae]etheexponentla|
functlone , for = x+/,anarbltrarycomp|exnumber,tobe
( I 6)
Thus lt appearsthatcomp|ex roots ofthecharacterlstlcequatlonwl|||eadto
comp|ex-va|uedso|utlonsofthedlerentla|equatlon. Acomplex-valued function
ioftherea|varlab|exassoclateswltheachrea|numberx(lnltsdomalnofdennl-
tlon)thecomp|exnumber
i,x; = ),x;+/ ,,x; ( I 7)
Therea|-va|uedfunctlons)and,areca||edthereal andimaginary parts, respec-
tlve|y,ofiIftheyaredlfferentlab|e,wedennethederivative iofiby
i ,x;= )
,x;+/ ,
,x; ( I S)
Thus weslmp|ydlerentlatetherea|andlmaglnarypartsofiseparate|y.
Wesaythatthecomp|ex-va|uedfunctloni,x; satlsnesthehomogeneous|ln-
eardlerentla|equatloni i,x; } = 0provldedthatltsrea|andlmaglnarypartsln
( I 7)separate|ysatlsfythlsequatlonsoi i,x; } = i ),x; } +/ i,,x; } = O.
The partlcu|ar comp|ex-va|ued functlons of lnterest here are of the form
i,x; = e
.where= .+// We notefromEu|er' sformu|athat
( I 9a)
and
( I 9b)
Themostlmportantpropertyofelsthat
(20)
lfl sacomp|exnumber.Theproofofthlsassertlonlsastralghtforwardcomputa-
tlonbasedonthedennltlonsandformu|asglvenear|ler.
n

,e
;= n

,e

cos /x) +/ n

,e

|a /x;
= .e

.a /x- /e

sln bx+/.e

sln bx +/e

cos /x
= ,.+// ; ,e

cos bx+/ e

sln /x) = e

Exampl e
Exampl e 4
Z. Homogeneous Equati ons wi th Constant Coeffi ci ents 1 31
Complex Roots
Itfo||owsfromEq. (20)thatwhenlscomp|exustaswhenlsrea|), e

wl||bea
so|utlonofthedlfferentla|equatlonln( I ) lfandon|ylflsaroototltscharacterlstlc
equatlon. Ifthecomp|ex conugate palrofroots, = O +//and

= O - //are
slmp|e(nonrepeated), thenthecorrespondlngpartofagenera|so|utlonofEq.( I ) ls
,(x) = c, e
+c

e
= c, e

-

+c

-
-

= c,
e

(cos bx+/sln bx) +c

(cos bx - /sln bx)


,(x) = ,c, +c

; e

cos bx +/ .c, - c

; e

sln bx,
where thearbltrary constants c, and c

canbecomp|ex. Forlnstance, thecholce


c, = c

= glvestherea|-va|ued so|utlon y,(x) = e

cosbx, whl|ethe cholce


c, = - / .c

= /glvesthelndependentrea|-va|uedso|utlony

,x; = e

sln/x.
Thlsyle|dsthefo||owlngresu|t.
THEOREM 3 Compl ex Roots
Ifthecharacterlstlcequatlonln(3)hasanunrepeatedpalrofcomp|exconugate
roots O +// (wlthb = 0) , then thecorrespondlngpartofagenera| so|utlonof
Eq. ( 1 ) hasthefom
e

,., cos bx+.


:
sln bx) .
Thecharacterlstlcequatlonof
y

+b
:
, = 0 (b > 0)
(2I )
l s

+/

= 0, wlthroots = +// SoTheorem3 (wlthO = 0) glvesthegenera|


so|utlon
,(x) = .,cos bx+.

sln bx.
Flndthepartlcu|arso|utlonof
, - :,'
+5, = 0
forwhlch,(0) = I and,' (0) = 5.
Sol uti on Comp|etlonofthesquarelnthecharacterlstlcequatlonyle|ds

- :+5 = ,- 2)

+ I = 0,
so- 2 = +,= +/ Thusweobtalnthecomp|exconugateroots2+/(whlch
cou|d a|so befound dlrect|yuslngthequadratlcformu|a) . HenceTheoren3 wlth
O = 2andb = I glvesthegenera|so|utlon
,(x) = e

,.,cos x+.

sln x) .
1 32 Chapter Z Li near Equati ons of Hi gher Order
Y
| Y
FIGURE 2.3. 1. Modulus and
argument of the complex number

cXump| e

..
...,.
,(0) = ., = i .. ,'
(0) = z.,.

= 5 .
....

= . .. ...,....
,x= e

,.a x .
...,5...,,.polar form
= x/ , = e

,zz;
..,... ....... ....,.,
= +/ ,= / = e

..modulus = , x

> 0... ..argument 0


..., z i ....,.,../ ......i ..
.,.. }z./ = e

...,-/ = e

.,...
....,.. = e

....,..
,z;
........,.....,.,..
......
..,..

4,= 0.
Sol ution .....,..
....+,+z/ ./ = e

..-/ = e

.....
..
,
-z/= ze

=
_
e

-
=
_

= -i /
..........,...= +,+i+/ ; ..
.,..,, .,.i+/..-i J /.,..,..
....,..

4, = 0.
Exampl e
Z. Homogeneous Equati ons wi th Constant Coeffi ci ents 1 33
Repeated Complex Roots
Theorem zho|dsforrepeated comp|exroots. IftheconugatepalrO +//hasnu|-
tlp|lclty/.thenthecorrespondlngpartofthegenera|so|utlonhastheform
,~,+~

x+

; e

-

+ , s, + s

-'
;e

-
-

= _x

,.

cos/x+a

sln/x; ,z:;

Itcanbeshownthatthez/functlons
x

.a /x. x

|a /x. 0 /- 1
thatappearlnEq. ,z:;are|lnear|ylndependent.
Flndagenera|so|utlonof( D
2
+6D + 1 3)
2
y = O.
Sol uti on Bycomp|etlngthesquare,weseethatthecharacterlstlcequatlon
Exampl e
(r
2

6r + 1 3)
2
= [ (r

3)
2

:}= 0
has as lts roots theconugate palr -3 +z/ofmu|tlp|lclty / = z Hence Eq. ,z:;
glvesthegenera|so|utlon
,,x; = e
-

,., .azx+a,|a zx;

xe
-

,.2 coszx+a2 slnzx;

In app|lcatlons we are se|dom presented ln advance wlth a factorlzatlon as


convenlentastheonelnExamp|e6. Often themostdlfncu|tpartofso|vlngahono-
geneous |lnearequatlonls nndlng therootsofltscharacterlstlcequatlon. Exanp|e
7l||ustratesanapproachthatmaysucceedwhenarootofthecharacterlstlcequatlon
canbefoundbylnspectlon.
. .
Thecharacterlstlcequatlonofthedlerentla|equatlon
,

+y' - l Oy = 0
ls thecublcequatlon

r - 10 = O.
Bya standard theorem ofe|ementary a|gebra, the on|y posslb|eratlona| roots are
the factors +1 , +z. +. and +1 0 ofthe constantterm 1 0. By trla| and error (ltnot
by lnspectlon) we dlscoverthe root z The factor theorem ofe|ementary a|gebra
lmp|lesthatr - zlsafactorof

r - 1 0, anddlvlslonofthefornerlntothe|atter
producesasquotlentthequadratlcpo|ynomla|
r
2

zr

= (r + 1 )
2
+:
Therootsofthlsquotlentarethecomp|exconugates -I +z/ Thethreerootswc
havefoundnowyle|dthegenera|so|utlon

1 34 Chapter Z Li near Equati ons of Hi gher Order


Exampl e d Theroots ofthecharacterlstlcequatlonofacertaln dlerentla|equatlonare . -5,
0, 0, 0, 0, -5, 2 +/ . and 2 +/ Wrlte a genera| so|utlonofthlshomogeneous
dlerentla|equatlon.
Sol uti on Theso|utloncanbereaddlrect|yfromthe|lstofroots. Itls
,,x; =., +.

x+.

+.
-
x

+.

+.

e
-
+.

xe

+e

(cs cosx+.
.
slnx;+xe

,.,

cosx+., , slnx;
Probles
........, .,...
.....
1. .=
3. =
5. =
7. . =
9. =
11.

=
12.

=
2. =
4. =
6. =
8.
10.

=
13.

. = ..

.=
15.

= 16.

=
17.

.= 18.

=
19.

=
20.

....Expand


....,.......

21. . = = =
22. . = .
23. = = =
24.

= = =
25.

= = = =
26.

= = . =
.......,........
. ..........,..
.,....
27.

.
28.

=
29.

=
30.

=
31.

. =
.

=
........., .
,..........
33.

.= ] =

34.

= = e
2
xf3

. ] = cos .
36.

. .= = sin x
37. Find a function . such that

.=

.for all .
and = = = and

=
38. Solve the initial value problem

=
= = =
given that ]j .=

i s one particular solution of the


differential equation.
............
..,,..........
39. . = ...Cx
2
) e
2
x
40. . = Ae
2
x .cos .sin .
.. . = .cos ..sin .cosh .sinh .
42. . = . .. .
2
cos . .. .
2
sin .
.....,....., .
,.....,.,.
43. (a) Use Euler' s formula to show that every complex num
ber can be written in the form re
i
8 , where and
d (b) Express the numbers .
and

i n the form re
i
8 (c) The two square
roots of re
i
8 are .e
i
8/
2
Find the square roots of the
numbers

and

44. Use the quadratic formula to solve the following equa


tions. Note in each case that the roots are not complex
conjugates.
(a) .
2
.= (b) .
2
.=
45. Find a general solution of =
46. Find a general solution of =
47. Find a general solution of =

,
48. Solve the initial value problem

=
=
= =
....Impose the given initial conditions on the
general solution
.= ..
where O and f are the complex conjugate roots of

- 1 =
to discover that
1 .

. = _ + 2`'

cos -
2
-
is a solution. )
49. Solve the initial value problem
/
4
) =

+ + + 2;
=

30.
50. The diferential equation
+ sgn . =
has the discontinuous coeffcient function
+1 if .
sgn x
- 1 if .
(25)
Show that Eq. (25) nevertheless has two linearly indepen
dent solutions .and

. defned for all .such that


Each satisfes Eq. (25) at each point .=
Each has a continuous derivative at .=
1 and

= =
.... Each .will be defned by one formula
for . and by another for . The graphs of these
two solutions are shown in Fig. 2. 3. 2.
51.
Z. 4 Mechani cal Vi brati ons 1 35
FIGUR 2.3.2. Graphs of
.and

.in Problem 50.


According to Problem 5 1 in Section 2. 1 , the substitution
In ..transforms the second-order Euler equa
tion ..

+ .+ = to a constant-coeffcient ho
mogeneous linear equation. Show similarly that this same
substitution transforms the third-order Euler equation
..

+ .

+ .+ .=
(where . .are constants) into the constant
coeffcient equation

. + 3. + . .
.

.
.......= In ....
......,....
....
52. .

+ .+ =
53. .

+ . + =
54. .

+ .

+ ..=
55. .

+ . =
56. .

+ .

+ . =
57. .

+ .=
58. .

+ .

+ .+ =
m
WWWW W


C
The motlon ofa mass attached to a sprlng serves as a re|atlve|y slmp|e exanp|e
ofthevlbratlonsthatoccurln morecomp|exmechanlca| systems. Fornany such
systems,theana|yslsofthesevlbratlonslsaprob|emlntheso|utlonof|lneardler-
entla|equatlonswlthconstantcoemclents.

'
'
La|||ot|am
os|t|oa
FIGUR 2.4. 1. A mass
spring-dashpot system.
We conslderabodyofmassm attachedtooneendofanordlnarysprlngthat
reslstscompresslonaswe||asstretchlng, theotherend ofthesprlng ls attachedto
a nxed wa||, as shown ln Flg. 2. 4. I . Assumethat the body rests on afrlctlon|ess
horlzonta| p|ane, sothatltcanmoveon|ybackandforth asthesprlngconpresses
andstretches. Denotebyxthedlstanceofthebodyfromltsequilibrium position
lts posltlon when the sprlng ls unstretched. We take x > 0 when the sprlng ls
stretched,andthusx < 0whenltlscompressed.
Accordlng toHooke' s|aw, therestoratlve force ithatthesprlng exerts on
themassl sproportlona|tothedlstancexthatthesprlnghasbeenstretchedorcon-
pressed. Because thls ls the same as the dlsp|acementx ofthe nass m fron lts
1 36 Chapter Z Li near Equati ons of Hi gher Order
Unstretched
spring
Static
equilibrium
Y
System
i nmotion "
FIGURE 2.4.2. A mass
suspended vertically from a spring.
equl|lbrlumposltlon,ltfo||owsthat
i=-/x ( I )
Theposltlveconstantof proportlona|lty/l s ca||edthespring constant. Notethat
iandxhaveopposlteslgns. i< 0whenx > 0,i> 0whenx < 0
Flgure 2. 4. I shows the mass attached to a dashpota devlce, |lke a shock
absorber, thatprovldes aforcedlrected opposlte to the lnstantaneous dlrectlonof
motlonofthemass- Weassumethedashpotls so deslgnedthat thls forcei,ls
proportlona|totheve|ocltyu =ax,aofthemass, thatls,
ax
i,= -cc=-c-
a
(2)
Theposltlveconstantcl sthedamping constant ofthe dashpot. Moregenera||y,
wemayregardEq. (2) as speclfylngfrlctlona| forces lnoursystem (lnc|udlngalr
reslstancetothemotlonof-;
If, lnaddltlont otheforces iand i, .themasslssubectedt oaglvenexter
nal force i, = i, ; . thenthetota|forceactlngonthemassls i = i+i,+i,
UslngNewton' s|aw
a

x
i = -. = --= -x
a

weobtalnthesecond-order|lneardlerentla|equatlon
-x+cx+/x=i,;
thatgovemsthemotlonofthemass.
(3)
Iftherel snodashpot(andwelgnorea||frlctlona|forces), thenwesetc=0
ln Eq. (3)andca|| themotlonundamped; ltls damped motlonlfc> 0. Iftherels
noextema|force, werep|acei,; wlth0lnEq. (3). We refertothemotlonasfree
ln thlscaseand forced lnthe casei, ; = 0. Thusthehomogeneousequatlon
-x

+cx+/x=0 (4)
descrlbesfreemotlonofamassonasprlngwlthdashpotbutwlthnoexterna|forces
app|led. Wewl||deferdlscusslonofforcedmotlonuntl|Sectlon2. 6.
Forana|ternatlve examp|e, we mlghtattach the masstothe |owerendofa
sprlngthatls suspendedvertlca||yfromanxedsupport,aslnFlg. 2. 4. 2. Inthlscase
thewelghtw =-,ofthemasswou|dstretchthesprlngadlstances,deternlned
byEq. ( I ) wlthi= -w andx =s, Thatls, -,=/s,. sothats,=-,,/ Thls
glves thestatic equl|lbrlumposltlonofthemass. If,denotesthedlsp|acementof
themass ln motlon, measured downward from lts statlc equl|lbrlumposltlon, then
weaskyoutoshowlnProb|em9that,satlsnesEq. (3), speclnca||y,that
-,

+c,+/,=i, ; (5)
lfwelnc|udedamplngandexterna|forces(neanlngthoseotherthangravlty) .
FIGURE 2.4.3. The simple
pendulum.
Z.4 Mechani cal Vi brati ons 1 37
The Simple Pendulum
Thelmportanceofthedlerentla| equatlonthat appears lnEqs. (3)and,5) stens
fromthefactthatltdescrlbesthemotlonofmanyotherslnp|emechanlca|systens
Forexamp|e,asimple pendulum conslstsofanass-swlnglngbackandforthon
the end ofa strlng (or better, a-.ss/essa;of|ength i. as shownln Flg. 2.4. 3.
We may speclfy theposltlonofthemassattlmebyglvlngthecounterc|ockwlse
ang|e = ,; thatthe strlngorrodmakeswlththevertlca| attlne To ana|yze
themotlonofthemass-. wewl|| app|ythe|awofthe.ose../oo)-e././../
ee,,.accordlngtowhlchthe sum oftheklnetlcenergy andthepotentla|energy
of-remalnsconstant.
Thedlstancea|ongtheclrcu|ararcfrom0 to-lsS = i. sotheve|ocltyof
themassls c= as,a= i,a,a ; . andthereforeltsklnetlc energyls
=

-c

-
as

-i

2 2 a 2 a
We next choose as reference polnt the |owest polnt U reached by the nass (see
Flg. 2. 4. 3). Then lts potentla| energy \ ls the product of lts welght -, and lts
vertlca|helght/= i, i - cos; above U, so
\ = -,i, i - cos;
Thefactthatthesumofand \ ls aconstantC thereforeglves
I a

--i

- + -,i, i - .a ; = c
2 a
Wedlerentlatebothsldesofthls ldentltywlthrespecttotoobtaln
so
a a

a
-i

- -

+-,i,m; - = 0,
a a a
(6)
after remova| ofthe common factor -i

,a,a; Thls dlerentla| equatlon can


bederlved ln a seemlng|y more e|ementary manneruslngthefanl|larsecond|aw
i = -. of Newton (app|led to tangentla| components ofthe acce|eratlon otthe
mass and the force actlng on lt) . However, derlvatlons ofdlerentla| equatlons
based onconservatlonofenergy are oen seen lnmore conp|ex sltuatlons where
Newton' s|awlsnotsodlrect|yapp|lcab|e,andltmaybelnstructlvetoseetheenergy
methodln aslmp|erapp|lcatlon|lkethependu|um.
Nowreca||thatlfl ssma||,thensln (thlsapproxlnatlonobtalnedby
retalnlngustthe nrstterm ln the Tay|orserlesforsln; Infact, slnand agree
totwodeclma|p|aceswhen ls atmost-,I 2(thatls, I 5 ) . Inatyplca|pendu|un
c|ock, forexamp|e, wou|d never exceed I 5 . It therefore seems reasonab|e to
slmp|lfyourmathematlca|mode|oftheslmp|ependu|umbyrep|aclng slnwlth-
ln Eq. (6). Ifwea|solnsertaterm.toaccountforthefrlctlona|reslstanceotthe
surroundlngmedlum,theresu|tls anequatlonlnthefornofEq.(4) .
+. +/= 0, ,7)
1 38 Chapter Z Li near Equati ons of Hi gher Order
A
FIGURE 2.4.4. The angle O.
where/ = ,}i Notethatthl s equatlonl slndependentofthe mass-ontheend
oftherod. Wemlght,however, expecttheeectsofthedlscrepancybetween0 and
sln0 toaccumu|ateoveraperlodoftlme,sothatEq. (7) wl||probab|y notdescrlbe
accurate|ytheactua|motlonofthependu|umovera|ongperlodoftlne.
Intheremalnderofthlssectlon,wenrstana|yzefree undampedmotlonand
thenfreedampedmotlon.
Free Undamped Motion
Ifwehaveon|yamassonasprlng, wlth neltherdamplngnorextema|force, then
Eq. (3)takestheslmp|erform
Itls convenlenttodenne
andrewrlteEq. (S)as
Thegenera|so|utlonofEq. (S' )ls
-x

+ /x= 0.
" 2
0 X + .
,
x= .
x(t)= ~.a .

+ ssln o,t .
(S)
(9)
(S' )
( I 0)
Toana|yzethemotlondescrlbedbythlsso|utlon,wechooseconstantsC and
o sothat
~
cos o =
C
'
and

s
sin o = -,
C
( I I )
as lndlcatedl nFlg. 2. 4. 4. Notethat,a|thoughtano = s}~.theang|eol s notglven
bytheprlnclpa|branchofthelnversetangentfunctlon(whlchglves va|ueson|yln
the lnterva| --} < x < -}; Instead, o l sthe ang|ebetween0 andz-whose
coslneand slnehavethe slgnsglvenln( I I ), whereelther~or sorbothmay be
negatlve. Thus
tan
,
, s}~;
o = -+tan
,
, s}~;
z-+tan
,
, s}~;
lf~ > 0, s > 0(nrstquadrant),
lf~ < 0(secondorthlrdquadrant),
lf~ > 0,s < 0(fourthquadrant),
wheretan
,
, s}~;lstheang|eln,--}. -};glvenbyaca|cu|atororconputer.
Inanyevent,from( I 0) and ( I I ) weget
x(t)= C cos o,t + sln o,t= C(cos o cos o,t +sln o sln o,t ) .
Wlththealdofthecoslneaddltlonformu|a,wenndthat
x(t) = Ccos (o,t - o) . ( I 2)
Thusthemassoscl||atestoandfrom aboutlts equl|lbrlumposltlonwlth
X
~ I
Z. 4 Mechani cal Vi brati ons 1 39
1. Amplitude
2. Circular frequency
J. Phase angle
C,
-
,, and
o.
Suchmotlonl s ca||edsimple harmonic motion.
Iftlmet ls measuredlnseconds,theclrcu|arfrequency
-
,hasdlmenslonsof
radlans persecond (radJs). Theperiod ofthe motlon ls the tlme requlredforthe
systemtocomp|eteonefu||oscl||atlon,solsglvenby
seconds, ltsfrequency ls
2
I =
-
,
I
-
,
t = =
I 2
( I 3)
( I 4)
ln hertz (Hz), whlch measures the number ofcomp|ete cyc|es per second Note
thatfrequencyls measuredlncyc|espersecond,whereasclrcu|arfrequencyhasthe
dlmenslonsofradlanspersecond.
Atyplca|graphofaslmp|eharmonlcposltlonfunctlon
x (t ) = Ccos (o,t - o) = Ccoso, t -
,
,,= Ccos(
-
, (t -
FGLR

.4.5. Simple
hamonic motion
. ls shown ln Flg. 2. 4. 5, where the geometrlc slgnlncance of the amp|ltude C, the
perlod I, andthetime lag
Exampl e 1
arelndlcated.
o
=
-
,
Ifthe lnltla|posltlonx(0) = x, and lnltla| ve|ocltyx'(0) = u, ofthe mass
areglven,wenrstdetermlnetheva|uesofthecoefnclents~andslnEq. ( I 0), then
nndtheamp|ltude C andphaseang|eo bycarrylng outthetransformatlonofx(t)
totheformlnEq. ( I 2), aslndlcatedprevlous|y.
A

ody ithmassm =

ogram (kg) lsattached to the endofasprlngthatls


stretched2meters(m)byaforceofI 00newtons(N). Itlssetlnmotlonwlthlnltla|
posltlonx, = I (m) and lnltla| ve|oclty u, = -5 (m/s). (Note that these lnltla|
condltlonslndlcatethatthebody ls dlsp|acedtotherlghtandlsmovlngto the|eft
at tlme t = 0. ) Flnd the posltlon functlon ofthe body as we|| as the amp|ltude,
frequency,perlodofoscl||atlon,andtlme|agofltsmotlon.
Sol uti on
Thesprlngconstantls/ = ( I 00N) J(2m) = 50(NJm), soEq. (S)yle|ds x+
50x = 0, thatls,
x+ I 00x = 0.
Consequent|y,theclrcu|arfrequencyoftheresu|tlngslmp|eharmonlcmotlonofthe
body wl||be
-
, =

I 00= I 0(rad/s) . Henceltwl||oscl||atewlthperlod


andwlthfrequency
2 2
I = = 0. 62S3s
-
, I 0
I
-
, I 0
t = = = I . 59 I 5Hz.
I 2 2
1 40 Chapter Z Li near Equati ons of Hi gher Order
Wenowlmposethelnltla|condltlonsx(0)= I andx (0)= -5ontheposltlon
functlon
x ( ) = ~cosl 0t + Bslnl 0t wlth x ,) = -l 0~sln l 0t + l 0Bcos l 0t .
I t fo||owsreadl|ythat~= I ands = - , s otheposltlonfunctlonofthebodyls
x () = cosl 0t - slnl 0t .
Henceltsamp|ltudeofmotlonl s
To nndthetlme|ag,wewrlte

2 I

x ( ) = _

cos l 0t -

sln l 0t = _cos ( l 0t - o) ,
wherethephaseang|eo satlsnes
2 I
cos o =

> 0 and sln o = - < 0.

Henceo lsthefourth-quadrantang|e
-I J

o = 2 +tan
|
2J

= 2 - tan
,
, , ~ 5. SI 95,
andthetlme|agofthemotlonl s
o
= ~ 0. 5S20s.
O
Wlththeamp|ltudeandapproxlmatephaseang|eshownexp|lclt|y,theposltlonfunc-
tlonofthebody takestheform
x ( ) -| . cos ( l 0t - 5. S I 95) ,
andltsgraphlsshownlnFlg. 2. 4. 6.
0
FIGURE 2.4.7. Overdamped
motion: x (t ) cj c
r|
t
+ c_c
rz
t
with
/j 0 and

Solution curves
are graphed with the same initial
position Xo and diferent initial
velocities.
Z. 4 Mechani cal Vi brati ons 1 41
X

I

-|
FIGURE 2.4.6. Graph of the position function
x(t ) cos(wot a) in Example I , with amplitude
1 . 1 1 8, period . 0. 628, and time lag 0. 582.
lreeDampedMotion
3
Wlthdamplngbutnoextema|force,thedlerentla|equatlonwehavebeenstudylng
takestheform-x+.x+/x=c,a|tematlve|y,
2 I

c x + px + .
,
x= .
where.,=,/}-ls thecorrespondlnga.-peaclrcu|arfrequencyand
c
p = - > c
z-
Thecharacterlstlcequatlon

+zp+.=cofEq., I 5)hasroots
, .

= -p+,p

.;

thatdependontheslgnof

.

/ .

- :/-
p - . = - =
,
:-

- :-

, I 5)
, I 6)
, I 7)
, I )
Thecritical damping c
;
ls glvenbyc
;
= , :/-.andwedlstlngulshthreecases,
accordlngasc >c
;
, c =c, orc < c
;
OVERDAMPED CASE: c ~ c,, (c
-
~ 4km) . Becausec l s re|atlve|y|argel nthls
case,weare dea|lngwlthastrongreslstancelncomparlsonwlthare|atlve|yweak
sprlngorasma||mass Then , I 7)glvesdlstlnctrea| roots, and

.bothofwhlch
arenegatlve Theposltlonfunctlonhastheform
, I 9)
Itlseasytoseethatx,;-- cas-- +~andthat thebodysett|estoltsequl|lbrlum
posltlon wlthout any oscl||atlons (Prob|em 29). Flgure 2 4 7 shows some typlca|
graphsoftheposltlonfunctlonfortheoverdampedcase,wechosex,anxedposltlve
numberandl||ustratedtheeectsofchanglngthelnltla|ve|oclty:, Ineverycase
thewou|d-beoscl||atlonsaredampedout.
1 42 Chapter Z Li near Equati ons of Hi gher Order
0
FIGUR 2.4.8. Critically
damped motion:
x(t ) = (C
I
+ c2t)e-pt with ]
Solution curves are graphed with
the same initial position Xo and
diferent initial velocities.
O
_ | j
" x = Ce-pt cos (W j / - a)
x- + Ce-pt
-
0
FIGUR 2.4.9. Underdamped
oscillations:
x(t) = Ce-pt COS(Wl t - a) .
CRITICALLY DAMPED CASE: c = c,, (c
-
= 4km) . Inthls case, ( I 7) glves
equa|roots, =
:
= - ofthecharacterlstlcequatlon,sothegenera| so|utlonls
(20)
Because e
-

> 0 and .,+.


:
has at most one posltlve zero, the body passes
through lts equl|lbrlum posltlon at most once, and lt l s c|ear thatx () -- 0 as
-- +c. Some graphs ofthe motlon ln the crltlca||y damped case appear ln
Flg. 2. 4. S, and they resemb|e those of the overdamped case (Flg. 2. 4. 7). In the
crltlca||y damped case, the reslstanceofthe dashpot lsust|arge enough to damp
out any oscl||atlons, but even a s|lghtreductlon ln reslstance wl|| brlng us to the
remalnlngcase,theonethatshowsthemostdramatlcbehavlor.
UNDERDAMPED CASE: c * c,, (c
-
* 4km) . Thecharacterlstlcequatlonnow
hastwocomp|exconugateroots -p+/, .- p
:
,andthegenera| so|utlonls
(2I )
where
@ : :
@
,
4km - .
:
.,

.
,
- p

2m
(22)
Uslng thecoslneaddltlonformu|aaslnthederlvatlonofEq. ( I 2), wemayrewrlte
Eq. (20)as
so
x,; = ce
-

.a ,., - o) (23)
where
~
cos o =
c

and

s
smo =
c

Theso|utlonln(22)representsexponentla||ydampedoscl||atlonsofthebody
around lts equl|lbrlum posltlon. The graph ofx,; |les between the 'amp|ltude
enve|ope curvesx = -ce
-

andx = ce
-

andtouches themwhen., - o ls
anlntegra|mu|tlp|eof. Themotlonlsnotactua||yperlodlc,butltlsneverthe|ess
usefu|toca||.,ltscircular frequeucy (moreproper|y,ltspseudofrequency), , =
2J., lts pseudoperiod of oscl||atlon, and ce
-

lts time-varying amplitude.


Most ofthese quantltles are shown ln the typlca| graph of underdamped motlon
ln Flg. 2. 4. 9. Note from Eq. (2I ) that ln thls case ., ls |ess than the undamped
clrcu|arfrequency .,, so , ls |argerthan theperlod ofoscl||atlonofthe same
mass wlthoutdamplng onthe same sprlng. Thus the actlon ofthe dashpothas at
|easttwoeects.
1. It exponentla||y damps the oscl||atlons, ln accord wlth the tlme-varylng
amp|ltude.
Z. Its|owsthemotlon, thatls,thedashpotdecreasesthefrequencyofthemotlon.
cXump| eZ
2. 4 Mechani cal Vi brati ons 1 43
Asthefo||owlngexamp|el||ustrates, damplngtyplca||ya|sode|aysthemotloa
furtherthatls, lncreasesthetlme |agas compared wlth undamped motlonwlth
thesamelnltla|condltlons
The mass and sprlng ofExamp|e I are now attached a|so to a dashpot that pro-
vldes I N ofreslstancefor each meter persecondofve|oclty The mass ls setla
motlon wlththe samelnltla|posltlonx(0) I and lnltla| ve|ocltyx' (0) * 5as
ln Examp|e I . Now nndtheposltlonfunctlon ofthe mass, lts new frequency aad
pseudoperlod ofmotlon, lts new tlme |ag, and the tlmes of lts nrstfour passages
throughthelnltla|posltlonx 0
Sol uti on Ratherthanmemorlzlngthevarlousformu|asglvenlntheprecedlngdlscusslon,ltls
betterpractlcelnapartlcu|arcasetosetupthedlfferentla|equatlonandthensolve
ltdlrect|y Reca|| thatm and/ 50, weare nowglven. * I ln mksuaits
HenceEq (4) ls x+x'+50x 0, thatls,
x+2x'+ I 00x 0.
Thecharacterl stlc equatlon
:
+ 2r + I 00 ,+ I )
:
+ 99 0 has roots r
i o

:
= -I +/ . sothe genera| so|utlonls
x(t ) 8
/
,~cos

+ Bsln

; (24)
Consequent|y,thenewclrcu|ar(pseudo)frequencyls.[ 9. 9499(ascom-
paredwlthu I 0lnExamp|e I ). Thenew(pseudo)perlodandfrequencyare
and
2 2
,
*
0. 63 I 5S
.
, 99
I .,
.,
*
I . 5S36Hz
, 2 2
(ascomparedwlth 0. 62S3 <
,
and I I 59I 5> I[ lnExample I ) .
Wenowlmposethelnltla|condltlonsx,0) I andx',0) 5ontheposltloa
functlonln,23)andtheresu|tlngve|ocltyfunctlon
x'
,;-8
/
,~cos

+Bsln

;+

8
/
, -~sln

+Bcos


Itfo||owsthat
x(0) ~I and x' (0) -~+ B

-5,
whencewenndthat~ I and B -4J

Thusthenewposltlonfunctlonof
the bodyls
x(t ) 8
/
cos

- sln

,
Hencelts tlme-varylngamp|ltudeofmotlonls
1 44 Chapter 2 Li near Equati ons of Hi gher Order
We thereforewrlte

x(t ) = e
_

cos

t -

sln

t
. 99 .I I 5 .I I 5

=

e
_
cos (

t - :, ; ,
wherethephaseang|e:, satlsnes

cos:, =

> 0 and
.I I 5
Hence:, ls thefourth-quadrantang|e

4
m:, = -< 0.

-4J

4
'
:, = 2 +tan
,
_ = 2 - tan
,
-5 9009,
99J I I 5 . 99
andthetlme |agofthemotlonls
:,

(
= ~0 593 I s
-
,
(as compared wlth - 0 5S20 <
(
ln Examp|e I ) Wlth the tlme-varylng am-
p|ltudeandapproxlmatephase ang|e shown exp|lclt|y, theposltlonfunctlonofthe
masstakestheform

x(t ) -

e
_
cos (

t - 5 9009) , (25)
andltsgraphlsthedampedexponentla|thatlsshownlnFlg 2 4 I 0(lncomparlson
wlththeundampedoscl||atlonsofExamp|e I )
X

FIGURE 2.4. 10. Graphs of the position function
x (t ) cos(wJ t al ) of Example 2 (damped
oscillations), the position function x (t) = cos (wot a) of
Example I (undamped oscillations), and the envelope curves
x (t) .


2. 4 Mechani cal Vi brati ons 1 45
From(24)weseethatthemasspassesthrough ltsequl|lbrlumposltlonx = 0
when .a ,, - (I ) = 0, andthus when
thatls,when

2
'
3
`
. . . .
Wesee slml|ar|ythattheundampedmassofExamp|e I passesthroughequl|lbrlum
when
3
t = o
a
-

,
2,

2
-
'

.
z,
3

+
2
-

Thefo||owlng tab|e compares the nrst fourva|ues , ,


:
,

,
-
weca|cu|ateforthe
undampedanddampedcases,respectlve|y.
I 2 3 4

-
(undamped) 0. I I 07 0. 4249 0. 7390 I . 0532

-
(damped) 0. I I 95 0. 4352 0. 7509 I 0667
Accordlng|y, ln Flg. 2. 4. I I (where on|y the nrst three equl|lbrlum passages are
shown)we seethedampedoscl||atlons|agglngs|lght|ybehlndthe undampedones.

x
x(t) " C cos(Wo t - O )
-I
FIGURE 2.4. 11. Graphs on the interval 0 :: :: 0. 8
illustrating the additional delay associated with damping.
1. Determine the period and frequency of the simple har
monic motion of a 4-kg mass on the end of a spring with
spring constant 1 6 N/m.
2. Determine the period and frequency of the simple har
monic motion of a body of mass 0. 75 kg on the end of
a spring with spring constant 48 N/m.
3. A mass of 3 kg is attached to the end of a spring that is
stretched 20 cm by a force of 15 N. It is set in motion with
initial position X 0 and initial velocity 0 = -1 0 m/s.
Find the amplitude, period, and frequency of the resulting
motion.
4. A body with mass 250 g is attached to the end of a spring
that is stretched 25 cm by a force of 9 N. At time 0
1 46 Chapter 2 Li near Equati ons of Hi gher Order
the body is pulled 1 m to the right, stretching the spring,
and set in motion with an initial velocity of 5 m/s to the
left. (a) Find .in the form cos(uq- a) . (b) Find
the amplitude and period of motion of the body.
............., .,..
..,,........+ . 0, ..
. / .

..........
.,............
.........
5. Two pendulums are of lengths . and L
2
and-when lo
cated at the respective distances .and .

fom the center


of the earth-have periods PI and P
2
. Show that
. .
_
P
2

6. A certain pendulum keeps perfect time in Paris, where the


radius of the earth is . 3956 (mi). But this clock loses
2 min 40 s per day at a location on the equator. Use the
result of Problem 5 to fnd the amount of the equatorial
bulge of the earth.
7. A pendulum of length 1 00. 1 0 in. , located at a point at
sea level where the radius of the earth is . 3960 (mi),
has the same period as does a pendulum of length 1 00. 00
in. atop a nearby mountain. Use the result of Problem 5 to
fnd the height of the mountain.
8. Most grandfather clocks have pendulums with adj ustable
lengths. One such clock loses 10 min per day when the
length of its pendulum is 30 in. With what length pendu
lum will this clock keep perfect time?
9. Derive Eq. (5) describing the motion of a mass attached to
the bottom of a vertically suspended spring. ....
First denote by .the displacement of the mass below
the unstretched position of the spring; set up the diferen
tial equation for . Then substitute y . Xq in this
diferential equation. )
10. Consider a foating cylindrical buoy with radius height
.and uniform density .0.5 (recall that the density
of water is 1 g/cm
3
). The buoy is initially suspended at
rest with its bottom at the top surface of the water and
is released at time Thereafter it is acted on by
two forces: a downward gravitational force equal to its
weight .

..and (by Archimedes' principle of


buoyancy) an upward force equal to the weight

..of
water displaced, where . . is the depth of the bot
tom of the buoy beneath the surface at time (Fig. 2. 4. 1 2) .
Conclude that the buoy undergoes simple harmonic mo
tion around its equilibrium position X, ..with period
, 2

...Compute ,and the amplitude of the mo


tion if . 0. 5 g/cm
3
, . 200 cm, and . 980 cm/s
2
.
FIGURE 2.4. 12. The buoy of Problem 1 0.
11. A cylindrical buoy weighing 1 00 lb (thus of mass
3. 1 25 slugs in ft-lb-s (fps) units) foats in water with its
axis vertical (as in Problem 1 0). When depressed slightly
and released, it oscillates up and down four times every
1 0 s. Assume that friction is negligible. Find the radius of
the buoy.
12. Assume that the earth is a solid sphere of uniform density,
with mass and radius . 3960 (mi). For a particle of
mass ..the earth at distance from the center of
the earth, the gravitational force attracting toward the
center is Fr

where M
r
is the mass of the
part of the earth within a sphere of radius (a) Show that
Fr .

(b) Now suppose that a small hole is


drilled straight through the center of the earth, thus con
necting two antipodal points on its surface. Let a particle
of mass be dropped at time 0 into thi s hole with ini
tial speed zero, and let be its distance from the center
of the earth at time (Fig. 2. 4. 1 3). Conclude from New
ton' s second law and part (a) that . where
k
2
.

..
FIGURE 2.4. 13. A mass falling down
a hole through the center of the earth
(Problem 1 2).
(c) Take . 32. 2 f/S
2
, and conclude from part (b) that
the particle undergoes simple harmonic motion back and
forth between the ends of the hole, with a period of about
84 min. (d) Look up (or derive) the period of a satellite
that just skims the surface of the earth; compare with the
result in part (c). How do you explain the coincidence?
Or .it a coincidence? (e) With what speed (in miles
13.
14.
per hour) does the particle pass through the center of the
earth? (D Look up (or derive) the orbital velocity of a
satellite that just skims the surface of the earth; compare
with the result in part (e). How do you explain the coinci
dence? Or .it a coincidence?
Suppose that the mass in a mass-spring-dashpot system
with 9, and . is set in motion with
. and . (a) Find the position func
tion . and show that its graph looks as indicated in
Fig. . . (b) Find how far the mass moves to the
right before starting back toward the origin.
: -
4
5
.

- I

.
.
FIGURE 2.4. 14. The position function
. of Problem
Suppose that the mass i n a mass-spring-dashpot system
with and . is set in motion
with . and . . (a) Find the position
function . and show that its graph looks as indicated in
Fig. . (b) Find the pseudoperiod of the oscillations
and the equations of the "envelope curves" that are dashed
in the fgure.
.
|


- |
Z
Z
- .
r
: | | : .
FIGURE 2.4. 15. The position function
.of Problem .
....,. .........,.
..... .........
...,.....,...........
,.....,... .......
...,..... .
.,..........
...,....,... ..,.
.....,...,...
.

.. . .......,.,.
2. 4 Mechani cal Vi brati ons 1 47
.. ... ........q....
..,.........,.
..........,....
............,..,.
,....,......
15. , . ..
16. . . ,
17. I , . .
18. . . -
19. . . 9. .
20. . .. .
21. I , . .
22. A 1 2-lb weight (mass slugs in fps units)
is attached both to a vertically suspended spring that it
stretches in. and to a dashpot that provides Ib of re
sistance for every foot per second of velocity. (a) If the
weight is pulled down I f below its static equilibrium po
sition and then released from rest at time fnd its po
sition function . (b) Find the frequency, time-varying
amplitude, and phase angle of the motion.
23. This problem deals with a highly simplifed model of a car
of weight Ib (mass slugs in fps units). As
sume that the suspension system acts like a single spring
and its shock absorbers like a single dashpot, so that its
vertical vibrations satisfy Eq. .with appropriate values
of the coeffcients. (a) Find the stifness coeffcient .
of the spring if the car undergoes free vibrations at cy
cles per minute (cycles/min) when its shock absorbers are
disconnected. (b) With the shock absorbers connected,
the car is set into vibration by driving it over a bump, and
the resulting damped vibrations have a frequency of
cycles/min. Afer how long will the time-varying ampli
tude be I % of its initial value?
..............,.....,..
...,......,.,.
. ... . .....,
u . ..u u ,

......
..,...,... ..,....,...
,
24. (Critically damped) Show in this case that
. .+ + ,.

25. (Critically damped) Deduce from Problem .that the


mass passes through . at some instant if and
only if .and ,+ ,.have opposite signs.
26. (Critically damped) Deduce from Problem .that .has
a local maximum or minimum at some instant if and
only if and + ,.have the same sign.
27. (Overdamped) Show in this case that
where

-, ,,

uand )


1 48 Chapter 2 Li near Equati ons of Hi gher Order
28. (Overdamped) If . deduce from Problem 27 that
.

sinh y
Y
29. (Overdamped) Prove that in thi s case the mass can pass
through its equilibrium position . at most once.
30. (Underdamped) Show that in this case

+ ,.

.
O
COS W[ +
WI
smwl t .
31. (Underdamped) If the damping constant is small in com
parison with

. apply the binomial series to show that


WI W
O

I -

.
.
32. (Underdamped) Show that the local maxima and minima
of
occur where
tan(wl t - a)

WI
Conclude that

l 2n/wl if two consecutive maxima


occur at times tl and

33. (Underdamped) Let XI and .

be two consecutive local


maximum values of . Deduce from the result of Prob
lem 32 that
In
.I

2np
.

WI
The constant 2n P/WI is called the logarithmic
decrement of the oscillation. Note also that mWI /n
because ,
D0I6 ....,........
.....viscosity ........,.
,...............
..... ........,........
.......,..........
.,.... FR 6n .. ......,.
......,..........
..........,.........,.
.. .. ..,.WI ......
............
.............
.....
34. (Underdamped) A body weighing Ib (mass
3. 1 25 slugs in fps units) is oscillating attached to a spring
and a dashpot. Its frst two maximum displacements of
6. 73 in. and 1 . 46 in. are observed to occur at times 0. 34
s and 1 . 1 7 s, respectively. Compute the damping con
stant (in pound-seconds per foot) and spring constant (in
pounds per foot) .
Diferential Equations and Determinism
.... .....,......,...
. .. ,....,..
.+ .+ .. (26)
....unique ..,...,....
.. . . ..........
....,.....,...,...
., .,..........
..,......,......,..
....precisely. ......,
......,........
,.....
35. Suppose that 1 , 2, and . 1 in Eq. (26) . Show
that the solution with . and . 1 is
.

36. Suppose that 1 and 2 but . 1 1 -
2
n
. Show
that the solution of Eq. (26) with . and . 1
is
.

l
n
sinh l O-
n
t .
37. Suppose that 1 and 2 but that . 1 + 1 -
2n
.
Show that the solution of Eq. (26) with . and
. 1 is
38. Whereas the graphs of . and .

resemble those
shown in Figs. 2.4.7 and 2. 4. 8, the graph of .

exhibits
damped oscillations like those illustrated in Fig. 2. 4. 9, but
with a very long pseudoperiod. Nevertheless, show that
for each fxed it is true that
lim .

lim .

.
n
o
n
o
Conclude that ...the three solu
tions are in "practical" agreement if is sufciently large.
Nonhomogelous Eqtations and Undetermined Coeficients
We|earnedlnSectlon2. 3howtoso|vehomogeneous|lnearequatlonswlthconstant
coefnclents,butwesawlnSectlon2. 4thatanextema|forcelnaslmp|emechanlca|
systemcontrlbutesanonhomogeneoustermtoltsdlerentla|equatlon.Thegenera|
nonhomogeneousnth-order|lnearequatlonwlthconstantcoefnclentshastheform
( I )
cXump| e 1
2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 49
ByTheorem5 ofSectlon2. 2, agenera| so|utlonofEq. , I ) hastheform
y= y
+y

'`)
wherethecomp|ementaryfunctlony
,x;lsagenera|so|utlonoftheassoclatedac
mogeneousequatlon
|')
and,, ,x;lsapartlcu|arso|utlonofEq., I ) .Thusourremalnlngtasklstonnd) .
Themethod of undetermined coefcients ls astralghtforwardwayofdolag
thlswhentheglvenfunctlon),x; lnEq., I ) lssufnclent|yslmp|ethatwecaamake
an lnte||lgentguessas to the genera|form ofy, For examp|e, supposethat ](x)
ls a po|ynomla| ofdegree m. Then, because the derlvatlves ofa po|yaomlal are
themse|ves po|ynomla|s of |ower degree, lt ls reasonab|e to suspect a partlcular
so|utlon
,, ,x;= ~
-
x
-
+~
-
.

x
-

.
+
. . .
+~, x +~,
thatlsa|soapo|ynomla|ofdegreem, butwlthasyetundetermlnedcoefncleats We
may,therefore,substltutethlsexpresslonfory,lntoEq., I ),andthenbyequatlag
coefnclentsof|lkepowersofxonthetwo sldes oftheresu|tlngequatlonattempt
todetermlnethecoemclents~,, ~, ,~
-
sothaty

wl||,lndeed,beapartlcular
so|utlonof Eq., I ).
Slml|ar|y,supposethat
),x;=O cos/x+ /slnkx .
Thenltlsreasonab|etoexpectapartlcularso|utlonofthesameform.
,, ,x;= ~.. /x+ s|a /x,
a |lnearcomblnatlon wlth undetermlned coemclents ~ and s The reasoa ls that
anyderlvatlveofsucha|lnearcomblnatlonofcos/xandsln/xhasthesameform.
Wemay therefore substltutethlsform of
y,ln Eq. , I ), andthenbyequatlngco
efnclents ofcos/xand sln/x on both sldesoftheresu|tlngequatlonattemptto
determlnethecoemclents~andsso thaty

wl||,lndeed,beapartlcularsolutloa
lt turns out that thls approach does succeed whenever all the derlvatlves ot
),x; have the sameformas ),x; ltse|f. Beforedescrlblngthe methodlntu|lgea
era|lty,wel||ustrateltwlthsevera|pre|lmlnaryexamp|es.
. . . .
Flndapartlcu|arso|utlonofy
+,+ :,= x+2.
Sol ution Here),x; = x+2lsapo|ynomla|ofdegree I,soourguesslsthat
y, ,x; = ~x+ s
Then y= ~and,= 0, so,,wl||satlsfythedlfferentla|equatlonprovldedthat
,0) + ,~; +:,~x+ s; = x+2,
thatls,
,:~;x+ , ~+:s;= x+ 2
1 50 Chapter 2 Li near Equati ons of Hi gher Order
cXump| eZ
fora||x. Thlswl||betruelfthex-termsandconstanttermsonthetwosldesofthls
equatlonagree. Itthereforesufncesfor~and stosatl sfy thetwo|lnearequatloas
:~= and~+:s = 2thatwereadl|yso|vefor~= and s= _ Thuswe
havefoundthepartlcu|arso|utlon

Flndapartlcu|arso|utlonof, :,=
Sol ution Anyderlvatlveofe

lsaconstantmu|tlp|eofe

,soltlsreasonab|etotry
cXump| e
Then ,

= ~e

,sotheglvendlfferentla|equatlonwl||besatlsnedprovldedthat
thatls, ~= 2, sothat~= Thus ourpartlcu|arso|utlonls,, (x) = e


Flnd apartlcu|arso|utlonof,+ ,' 2, = 2cosx.
Sol ution Anrstguessmlghtbe,, (x) = ~cosx,butthepresenceof,'onthe|e-handslde
slgna|sthatweprobab|yneedatermlnvo|vlngslnx aswe|| . Sowetry
cXump| e4
,, (x) = A cos x + ssln x,
, (x) = -~sln x + scos x,
,

(x) = -~cos x ssln x.


Thensubstltutlonof,
o
andltsderlvatlveslntotheglvendlerentla|equatlonyle|ds
,-~cosx - sslnx) + (-~slnx + scosx) - z,~cosx + sslnx) = 2cos x ,
thatl s (co||ectlngcoefnclentsonthe|e),
, -~+ s;cos x + , -~- s; sln x = 2 cos x.
Thlswl||betruefora||x provldedthatthecoslneandslnetermsonthetwosldes
ofthls equatlon agree. It therefore sufncesfor~ and s to satlsfy the two |lnear
equatlons
-~+ s= 2,
-~- s= 0
wlthreadl|yfoundso|utlon~= , s = Henceapartlcu|arso|utlonls
( )

,
o
X = _cos x + _ smx.
The fo||owlngexamp|e, whlch superncla||y resemb|es Examp|e 2, lndlcates
that the method ofundetermlned coefnclents ls not a|ways qulte so slmp|e as we
havemadeltappear.
Flndapartlcu|arso|utlonof, - :, = ze
:

2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 51


Sol uti on Ifwetry,, ,x;= ~e

.wenndthat
Thus, nomatterhow ~ls chosen, ~e

cannot satlsfy theglvennonhomogeaeous


equatlon In fact, theprecedlngcomputatlon shows that ~e

satlsneslastead the
assoclated/o-o,eeosequatlon Therefore, we shou|dbeglnwlthatrlalfunctloa
,, ,x;whose derlvatlve lnvo|ves both e

.aso-e//,e/sethatcan caace| upoa


substltutlon lnto the dlfferentla| equatlon to |eave the e

term that we aeed ^


reasonab|eguessls
forwhlch
, ,x;= ~e

+
z~xe

and ,

,x;= :~e

+
:~
,e

Substltutlonlntotheorlglna|dlfferentla|equatlonyle|ds
,:~e

+
:~xe

;- :,~xe

;= ze

The terms lnvo|vlng xe

ob|lglng|y cance| , |eavlng on|y :~e

= ze

. so that
~= | Consequent|y,apartlcu|arso|utlonls

The General Approach


Ourlnltla|dlfncu|tylnExamp|e:resu|tedfromthefactthat),x;= ze

satlsnes
theassoclatedhomogeneousequatlon Ru|e I , glvenshort|y, te||swhattodowhea
wedonothavethlsdlmcu|ty,andRu|ezte||swhattodowhea wedohavelt
Themethodofundetermlnedcoemclentsapp|les wheneverthemnctlon](x)
lnEq , I ) ls a|lnearcomblnatlonof(nnlte)products offunctlonsofthefo||owlng
threetypes .
1. Apo|ynomla|lnx,
Z. Anexponentla|functlone
,
J. cos/xorsln/x
Anysuchfunctlonforexamp|e,
),x;= ,- :x

; e

- :x

cosi cx.
,:;
hasthe crucla| property that on|y]/e/,many |lnear|y lndependent functlonsap-
pearas terms (summands)ln ),x;andltsderlvatlves ofa||orders lnRules I aad
zweassumethat i, = ),x; ls anonhomogeneous |lnearequatlonwlthconstaat
coefnclentsandthat),x;lsafunctlonofthlsklnd
RULE 1 Method of Undetermi ned Coeffcients
Supposethatnotermappeingeitherin),x;orlnanyofltsderivatlvessatlsnes
theassociatedhomogeneousequatlon L, = 0. Thentakeasatrla|so|utionfor
) a|lnearcombination of a||| lnear|yindependentsuchterms and thelrderiva-
tlves. Then determlnethe coefnclents by substltutlon ofthls tria| so|utlon lnto
thenonhomogeneousequatlonL, = ),x;
1 52 Chapter 2 Li near Equati ons of Hi gher Order
Exampl e
Notethatthlsru|els notatheoremrequlrlngproof, ltls mere|yaprocedureto
befo||owedln searchlngfor apartlcu|arso|utlon), . Ifwesucceedlnnndlng), ,
thennothlngmoreneedbesald. (Itcanbeproved,however,thatthlsprocedurewll|
a|wayssucceedunderthecondltlons speclnedhere. )
lnpractlcewecheckthesupposltlonmadelnRu|e I bynrstuslngthecharac
terlstlcequatlonto nndthe comp|ementary functlon), and then wrltea|lstofall
theterms appearlngln )(x) andlts successlvederlvatlves. Ifnoneofthetermsla
thls|lstdup|lcatesatermln) , thenweproceedwlthRu|e I .
Flndapartlcu|arso|utlonof
,+ 4, =3x

. (5)
Sol uti on The(faml|lar)comp|ementaryso|utlonofEq. (5)ls
Exampl e
) (x) = cj cos2x+c_ sln2x.
Thefunctlon )(x) = 3x

and lts derlvatlves are constantmu|tlp|esofthe|lnear|y


lndependentfunctlonsx

,x
:
,x, and I . Becausenoneoftheseappearsln) , wetry
)
o
=Ax

+ Bx
:
+ Cx + D,
)=3Ax
:
+2Bx +C,
)=6Ax+ 2B.
SubstltutlonlnEq. (5)glves
)+4,, = (6Ax +2B) +4(Ax

+Bx
:
+Cx+ D)
=4Ax

+4Bx
:
+ (6A +4C)x+ (2B +D) =3x

Weequatecoemclentsof|lkepowersofx lnthe|astequatlontoget
4A = 3,
6A + 4C =0,
4B = 0,
2B + D =0
wlth so|utlonA = , B = 0, C = g and D =0. Henceapartlcu|arso|utlonof
Eq. (5)ls
So|vethelnltla|va|ueprob|em
( )

,
)
o
x = x - _x.
) - 3,'
+ 2,=3e

- l 0cos 3x,
,(0) = I , ,' (0) =2.

(6)
Sol uti on Thecharacterlstlc equatlon r
:
- 3r+ 2 = 0hasroots r = I and r = 2, so the
comp|ementaryfunctlonls
Exampl e
2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 53
1|e:e:ms|aveivea|a),x) =,
- l c .esxaaa|:sae:|va:|vesa::,
,.asx,
aaas|axse.aaseaeaeei:|eseaea:s|ay_ , we::y
y
,
= n,
+ s .es x+ c s|a x,
y=-n,
- s s|a x+c.esx
y
,
= n,
- s .es x- c s|a x
Aue:wesa|s:|:a:e:|esees:ess|eas|a:e:|ea|ne:ea:|aieaa:|ea|a,:)aaa.e|i:.:
.eeia.|ea:s,we,e:
y- y+zy
,
=:n,

+, -s- c) .esx+,s c)s|a x


=,

- l c .es x
weeaa:e:|e.eeia.|ea:sei:|e:e:ms|aveiv|a,,
,:|ese|aveiv|a.esx, aaa
:|ese|aveiv|a,s|ax. 1|e:esai:|s:|esys:em
:n = ,
-s- c=-l c,
s - c = c
w|:|seia:|ean= , s= _ aaac= _ 1||s,|ves:|ea::|.aia:seia:|ea
, )
.

9
y
,
x = ,
+_ .es x +
r
sm x
w||.|,|eweve:,aeesae:|ave:|e:ea|:ea|a|:|aivaiaes| a,:)
1esa:|siy:|ese|a|:|ai.eaa|:|eas,we|e,|aw|:|:|ey
,,m/seia:|aa
y,x) =y

,x)+y
,
,x)

9
= c, , + .

, + , +_ .es x +
r
sm x
w|:|ae:|va:|ve
1|e|a|:|ai.eaa|:|eas|a,:)ieaa:e:|eeaa:|eas
y,c)=c, +.

+,+= l
y,c)=c, +z.

- +=z
w|:|seia:|eac, =, .

= 1|eaes|:eaa::|.aia:seia:|ea|s:|e::ia::
, )
,

+
e

+
,

+
7
+
9
y x = - ,
r
, , _ .es x _ sm x
t|aa:|e,eae:aiie:meiaa::|.aia:seia:|eaei

,)
1 54 Chapter 2 Li near Equati ons of Hi gher Order
Sol ution 1|e.|a:a.:e:|s:|.eaa:|ea

+ =0|as:ee:s=c, = -/ , aaa=/ sa
:|e.emiemea:a:yiaa.:|ea|s
1|eae:|va:|vesei:|e:|,|:|aaas|ae|a,7)|aveive:|e:e:ms
.esx, s|ax, x.esx, xs|ax,
e
:

, xe
:

, aaa x
:
e
:

se.aase:|e:e|saeaai|.a:|eaw|:|:|e:e:msei:|e.emiemea:a:yiaa.:|ea,:a:
::|aiseia:|ea:a|es:|eie:m
y, ,x; =A.es x+ Bs|a x+ cx.es x+ bxs|a x+ ie
:

+ ixe
:

+ ox
:
e
:

Ueasa|s:|:a:|a,y

|a,7)aaaeaa:|a,.eeu.|ea:sei i|ie:e:ms, we,e:sev:a


eaa:|easae:e:m|a|a,:|esevea.eeia.|ea:sA, B, C, b, i, i,aaao
The Case of Duplication
Newwe:anea:a::ea:|ea:e:|es|:aa:|ea|aw||.akaielaeesae:aiy semeei
:|e:e:ms|aveivea|a ),x; aaa |:s ae:|va:|vessa:|siy:|easse.|a:ea|eme,eaeeas
eaa:|ea te:|as:aa.e, saese :|a:wewaa::e aaa aa::|.aia:seia:|ea ei:a:
a|iie:ea:|aieaa:|ea
(S)
r:e.eea|a,as|akaiel , ea:a:s:,aessweaia|e
,;
1||sie:meiy,,x; w|iiae:|eaaeaa:e|e.aase:|e.emiemea:a:yiaa.:|eaai
(S)|s
, l c)
se sa|s:|:a:|ea ei,; |a :|e iei:|aaa s|ae ei(S) weaia y|eia ze:e :a:|e: :aaa
,zx- ; e

1esee|ew:eameaaea:a:s:,aess, wee|se:ve:|a:
|y , i ; eise.:|eaz ii,,x; |s ., seia:|eaei (S) aaawe aiy :a:
ee:a:e:, b- ;
:
:e|e:|s|aes,wesee:|a:,,x; |saiseaseia:|eaei:|eeaa:|aa
,b- ; `,=c 1|e,eae:aiseia:|eaei:||s/o-o,eeoseaa:|ea.aa|ew:|:::a
as
1|ase.eseia:|eaeiea:e:|,|aaieaa:|ea|a(S) |s:|esameia.emiemea:a:y
iaa.:|eaaaaap././.so//oei:|eie:m
, l l )
2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 55
Ne:e:|a::|e:|,|:|aaas|ae|a, I I ) .aa|ee|:a|aea|ymai:||y|a:a.a:e:m
eiea:a:s:,aess|a(9)|y:|eieas:es|:|ve|a:e,:aiewe:eix ,|a:a|s.as:,x

)
:|a:saia.es:eei|m|aa:eaai|.a:|ea|e:weea:|e:e:msei:|e:esai:|a,::|aise|a:|aa
,, ,x;aaa:|e.emiemea:a:yiaa.:|ea)c ,x),|vea|a, I 0) 1|| s:e.eaa::sa..e:as
|a:|e,eae:ai.ase
1es|mi|iy:|e,eae:ais:a:emea:eikaie2, wee|se:v::aa::eaaaaa::|.aia:
seia:|eaei:|eaea|eme,eaeeasi|aea:a|ne:ea:|aieaa:|ea
i,=

,x;+,x; , , I 2)
|:saia.es:eaaa sep.me/,a::|.aia:seia:|easr,,x;aaa l

,x) ai:a::waeaa
:|eas
i,=

,x) aaa i,= ,x; .


:ese.:|veiyte:i|aea:|:y:|ea,|ves
i r,+ r
:
}= ir, +ir
:
= I, ,x;+,x; ,
, I J)
aaa:|e:eie:e)
]
= r, + r

|saa::|.aia:seia:|eaei , I 2) ,1a|s|sa:yeei
sae:es|:|ea:|a.|ieie:aea|eme,eaeeasi|aea:eaa:|eas )
Newea::e|iem|s: eaaaaa::|.aia:seia:|eaei:|eeaa:|eai,= ],x) ,
w|e:eI ,x;|sai|aea:.em||aa:|eaei:eaa.:sei:|eeiemea:a:yiaa.:|aasi|s::a|a
,4) 1|asI ,x;.aa|ew:|::eaasasamei:e:msea.|ei:|eie:m
, I 4)
w|e:e i
-
,x; |saeiyaem|ai |ax eiae,:eem. Ne:e:|a:aayae:|va:|v:aisa.a
a:e:m|sei:|esameie:m|a:w|:|/o/s|aesaaa.es|aesaea:|a, 1a::a.:
aa:e|yw||.|wea::|veaea:i|e:a::|ea::|.aia:seia:|ea|a, I I ) ei(S) .aa|:
,eae:ai|zea:es|ew:|a::|eieiiew|a,:e.eaa:e|saiwayssa..essiai
RULE Z Method of Undetermi ned Coeffcients
ii:|eiaae:|ea],x; |sete|:|e:te:m|a, I 4), :aieas:|e::iaiseia:|ea
,

,x;= x , ~,+n, x+~


:
x

+
. . .
+~
-
x
-
; e

.es /x
+ , s,+s, x+ s

x
:
+
.
+s
-
x
-
; e

|a /x} , , I 5)
w|e:es|||esmaiies:aeme,a:|ve|a:e,e:sa.|:|a:sa.|:|a:ae:e:m|a)
]
aa
i|ea:esa:e:mia:|e.emiemea:a:yiaa.:iea) . 1|eaae:e:miae:|e.eeu.|ea:s
|aL.( I 5)|ysa|s:|:a:|a,)

|a:e:|eaea|eme,eaeeaseaa:|ea
ia:a.:|.eweseiaemaeea:eaeaiw|:|aiaa.:|eaI ,x) :sa|||:|a,:a:iai|
,eae:ai|:y|a, I 4) 1|e:a|ie|at|, 2 5 I i|s:s:|eie:mei,,|ava:|eas.ammaa
.ases,.e::eseaa|a,:e:|eess|||i|:|esm = 0,= 0,aaa/= c
Oa:|ee:|e:|aaa, |:|s.emmea:eaave
I,x) =

,x;+,x; ,
w|e:e II ,x; aaa ,x; a:e a|iie:ea: iaa.:|eas ei:|e se:: i|s:ea| a:|: :a|i: |a
t|, 2 5 I ia:|| sevea:we:aieas,,:|esamei:|e::|aiseia:|easia:II ,x) aaa
,x; . .|ees|a,ssep.me/,ie:ea.|a:::eei|m|aa:eaai|.a:|eaw|:a:a:.amie
mea:a:yiaa.:|ea.1||s:e.eaa:e|s|iias::a:ea|asamiesS:a:ea,aI 0
1 56 Chapter 2 Li near Equati ons of Hi gher Order

cXump| eb
Pm = bo + b
t x + b
2
x
2
+
. . .
+ b
mx
m
a cos kx + b sin kx
e'X (a cos kx + b sin kx)
Pm (x) e'X
Pm (x) (a cos kx + b sin kx)
xS (Ao + At x + A
2
X
2
+
. . .
+ Amx
m
)
xS (A cos kx + B sin kx)
xS e'X (A cos kx + B sin kx)
xS (Ao + At x + A
2
X
2
+
. . .
+ Amx
m
)e
rx
xS [ (Ao + At x +
. . .
+ Amx
m
) cos kx
+ ( Bo + Bt x +
. . .
+ Bmx
m
) sin kx]
FIGURE 2.5. 1. Substitutions in the method of undeterined coefcients.
t|aaaa::|.aia:seia:|eaei
, l ,
Sol uti on 1|e.|a:a.:e:|s:|.eaa:|ea

=0|as:ee:s, =

=0aaa

= -l , se:a:
.emiemea:a:yiaa.:|ea|s
Asaa:s:s:e:ewa:aea:a::|.aia:seia:|ea,weie:m:|esam
1|ea::n, .e::eseaa|a,:e,
aeesae:aai|.a:eaaya::ei:ae.emiem:a
:a:yiaa.:|ea,|a::|ea:: s +cx+bx

mas:|emai:|i|ea|yx

:eei|m|aa::
aai|.a:|ea. uea.ewe:aie

=n,

+sx

+cx

+bx
-
,

=n,

+zsx+cx

+:bx

=n,

+zs+:cx+l zbx

aaa
y

= n,

+:c+z:bx
sa|s:|:a:|eaei:|eseae:|va:|ves|a., l :,y|eias
zn,

+,zs+:c)+,:c+z:b)x+l zbx

=,

+:x

1|esys:emeieaa:|eas
zn= ,
:c+ z:b=0,
zs+ :c=0,
l zb= :
|as:|eseia:|ean = s =:, c= - , aaab= uea.e:|eaes|:eaa::|.aia:
seia:|ea|s

Exampl e
2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 57
De:e:m|ae:|ea:e:|a:eie:mie:aa::|.aia:seia:|eaei
,

+:,+ I , e
-

.es zx
Sol ution 1|e.|a:a.:e:|s:|.eaa:|ea

+:+ l 0|as:ee:s-+z/ ,se:ae.emi:m:a


:a:yiaa.:|ea|s
Exampl e 1
, ,x) e

,.,.eszx+ .

s|azx)
1||s| s :|esameie:masaa:s:a::em:e
-

, n.eszx+ ss|azx) a:aa::|.aia:


seia:|ea,sewemas:mai:|iy|yx:eei|m|aa:eaai|.a:|ea uea.eweweaia:a|:
,_ ,x) e
-

,nx.eszx+ sxs|azx)
___ ,___ u , __ . ... .. . .. ____ __ _ . _ _ _ _ u _ _ _____ __ u ___ _. _ ___ v _ ..

De:e:m|ae:|ea:e:|a:eie:mie:aa::|.aia:seia:|eaei:|eau|e:ae:eaa:|aa
Sol uti on 1|e.|a:a.:e:|s:|.eaa:|ea,- z)

+9)= 0|as:ee:s z, z, z,/, aaa-/ ,


se:|e.emiemea:a:yiaa.:|ea|s
Asaa:s:s:e:ewa:a:|eie:meiaa::|.aia:seia:|ea,weesam|ae:|:sam
,n+ sx+Cx

) e

+ , b+ Ix) .esx+ ,F +ox)s|ax


1eei|m|aa:eaai|.a:|eaw|:|:e:mseiy ,x) , :|ea:s:a::-.e::eseaa|a,:e
,

_
mas:|emai:|i|ea|yx

,aaa:|ese.eaaa::-.e::eseaa|a,:exs|ax-mas:|:
mai:|i|ea|yx uea.eweweaia:aie
Variation of Parameters
t|aaiiy,ie:ase|a:ea::|ei|aaeis|:aa:|ea|aw||.|:|eme:|eaeiaaa::e|a:a
.eeia.|ea:s.aaae:|easea Ceas|ae:,ie:esamie,:|eeaa:|ea
,

+, :aax, , I 7)
w||.|a:a:s:,iaa.emay aea:s|m|ia::e:|ese.eas|ae:ea|a:|e:e.ea|a,es
amies Ne:se, :|eiaa.:|ea),x) :aa x|as/]/e/,-.,i|aea:iy|aae:aa:a:
ae:|va:|ves
se.

x, zse.

x:aax, :se.

x:aa

x+zse.
-
x
1|e:eie:e, weaeae:|ave ava|ia|iea]/e i|aea: .em||aa:|ea:eas:as a::|ai
seia:|ea
1 58 Chapter 2 Li near Equati ons of Hi gher Order
wea|s.ass|e:e:|eme:|eaeivariation of parameters, w||.a-|a:|a.||:
,:|a:|s,|i:|e|a:e,:ais:|a:aea:.aa|eevaiaa:ea)-.aaaiways|:asea:eaaad
a::|.aia:seia:|eaei:|eaea|eme,eaeeasi|aea:a|iie:ea:|aieaa:|ea
,

+;-
-
,,x; ,

+ + ;,,x; ,+;
, ,x; ,= ),x; , ( l )
:ev|aea:|a:weai:eaay|aew:|e,eae:aiseia:|ea
y
= ., y,+ .
:
y
:
+
. . .
+.
-
y-
( l 9)
ei:|easse.|a:ea|eme,eaeeaseaa:|ea
,

+ ;-
-
,,x; ,

-
-
,

+

+ ;,,x; ,+;
, ,x; ,= 0
(20)
ue:e,|a|:|ei,|s:|e|as|.|aeaei:|eme:|eaeiva:|a:|eaeia:am::e:s sa
ese:|a:we:eia.e:|e.eas:aa:s, e:p..-ees,., ,.
:
, ,.
-
|a:|e.emiem:a
:a:y iaa.:|ea|a. ( I 9) w|:|.././/es iaa.:|eas[
:
, ,
-
eix we as|
w|e:|e:|:|sess||ie:e.|eese:|eseiaa.:|eas|asa.|away:|a::|e.em||aa:|ea
|saa::|.aia:seia:|eaei:|eaea|eme,eaeeaseaa:|ea|a( I ). i::a:asea::aa:
:||s/saiwaysess||ie.
1|eme:|ea|sessea:|aiiy:|esameie:aiie:ae:s 2, |a:wew|iia:s.:||:
|:|aae:a|ieaiyie::|e.ase= 2 sewe|e,|aw|:|:|ese.eaae:ae:aea|eme:
aeeaseaa:|ea
i,}= ,

+ i,x; ,+ _,x; ,= ),x; (22)


w|:|.emiemea:a:yiaa.:|ea
(23)
easemeeea|a:e:vail w|e:e:|eiaa.:|easiaaa _a:e.ea:|aaeas. wewaa::a
aaaiaa.:|eas,aaa
:
sa.|:|a:
(24)
|saa::|.aia:seia:|eaei.(22)
Oae.eaa|:|eaea:|e:weiaa.:|eas , aaa
:
|s:|a:i,,} = ),x; s:.aas:
:we.eaa|:|easa:e:ea|:ea:eae:e:m|ae:weiaa.:|eas, wea:ei:ee:e|mes:aa
aaa|:|eaai.eaa|:|eaeiea:.|e|.e.wew|iiaese|aaway:|a:s|mi|aes:ae.am
a:a:|easasma.|asess||ie sa:a:s:,:e|mese:|e.eaa|:|eai,,}= ),x; , :
mas:.ema:e:|eae:|va:|ves,aaa,

1|e:eaa.::aie,|ves
1eave|a:|eaea:aa.eei:|ese.eaaae:|va:|vesaaa. :aeaaa|:|eaai.eaa|:|aa
:|a:weaew|mese|s:|a::|ese.eaasam|::emas:vaa|s|
(25)
2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 59
1|ea
=

+

aaa:|e:eaa.::aie,|ves
//
,

) ,

)

=

+

sa:|e:|
aaa

sa:|siy:|e|eme,eaeeaseaa:|ea
,

+_,= c
asse.|a:eaw|:|:|eaea|eme,eaeeaseaa:|ea|a,zz),se
ie:/ = l , 2i::|e:eie:eieiiewsi:em,z):|a:
iav|eweis. (24) aaa,z) ,:||smeaas:|a:
|ea.e
//
,

)
_
_

i }

,z)
,z)
,zs)
,z)
1|e:ea|:emea::|a:

sa:|siy:|eaea|eme,eaeeaseaa:|ea|a,zz)-:|a:|s,:|a:
i,,}= ],x)-:|e:eie:e|mi|es:|a:
,c)
t|aaiiy,s ,25) aaa ,c)ae:e:m|ae:|eiaa.:|eas
aaa

:|a:weaeea
Ceiie.:|a,:|eseeaa:|eas, wee|:a|aasys:em
, l )
ei:wei|aea:eaa:|eas| a:|e:weae/../.es aaa

Ne:e:|a::|eae:eo|aaa:
ei.eeu.|ea:s|a, l ) |ss|miy:|ew:eas||aa

Oa.ewe|aveseivea
:|eeaa:|eas|a, l ) ie::|eae:|va:|ves aaa we|a:e,:a:eea.|:ee|:a|a:|e
iaa.:|eas aaa

sa.|:|a:
,z)
| s:|eaes|:eaa::|.aia:seia:|eaei ,zz) iar:e|iemweas|yea:e:a::y
ea::||s:e.essesi|.|:iyaaa:|e:e|yve:|iy:|eienaiaie:

|a:|eieiiew|a,
:|ee:em
1 60 Chapter 2 Li near Equati ons of Hi gher Order
cXump| e 1 1
THEOREM 1 Variati on of Parameters
it:|eaea|eme,eaeeas eaa:|ea,

+ _,x) y = |aseemie
mea:a:ytaae:|ea

=
+

:|eaa:|eaia:seia:|ea|s,|vea
|y
.


a + ),
y, ,x) ),x)
a

x

x x

x,
w,x) w,x)
(33)
w|e:ew= w

|s:|ew:easi|aaet:|e:we|aaeeaaea:seia:|easy, aaa

et:|eassee|a:ea|eme,eaeeaseaa:|ea.
~. .. ~ ...... ... . . .
t|aaaa::|.aia:seia:|eaei:|eeaa:|ea

+
= :aax.
Sol uti on 1|e.emiemea:a:yiaa.:|ea| s

= .esx+

s|a x, aaawe.eaias|miy
sa|s:|:a:ea|:e.:iy|a, 33) . sa:|:|sme:e|as::a.:|ve:ese:a:|eeaa:|eas|a
(3 I ) aaaseiveie:eaaae_ ,sewe|e,|aw|:|
= .es x,
,

= - s|a x,

= s|a x,
,= .es x
uea.e:|eeaa:|eas|a,3 I ) a:e
,e ,.es x) + ,e,s|a x) = 0,
,e ,- s|a x) + ,e,.es x) = :aa x
weeas|iyseive:|eseeaa:|easie:
uea.ewe:a|e

s|a

x
e

= - smx:aax =
-
-= .esx- se. x,
.es x
e= .es x:aax = s|a x
= ,,.esx- se. x) ax= s|a x- iase.x+:aa x
aaa

= ,s|a xax= .es x.


,Deyea seew|y we.|eese:|e.eas:aa:s ei|a:e,:a:|ea :e|eze:e:) 1|asea:
a::|.aia:seia:|ea|s

=

+

= ,s|ax- iase.x+:aax ) .esx+ ,- .esx,s|ax) ,


:|a:| s,

= -,.es x) ia se. x+:aa x .


2. 5 Nonhomogeneous Equati ons and Undetermi ned Coeffi ci ents 1 61
..1 .......,......y
p
.
..,.......,.,.....
....,x.
1. y" + 1 6y = 3x
2. y" - y' - 2y = 3x + 4
3. y" - y' - 6y = 2 sin 3x 4. 4y" + 4y' + y = 3xex
5. y" + y' + y = sin
2
x 6. 2y" + 4y' + 7y = x2
7. y" - 4y = sinh x 8. y" - 4y = cosh 2x
9. y" + 2y' - 3y = I + .
10. y" + 9y = 2 cos 3x + 3 sin 3x
11. y(3) + 4y' = 3x - I 12. y(3) + y' = 2 - sin x
13. y" + 2y' + 5y = sin x 14. y(4) - 2y" + y = .
15. y
(5)
+ 5y(4) - y = 1 7 16. y" + 9y = 2x23x + 5
17. y" + y = sin x + x cos x
18. y(4) - 5y" + 4y = x2x
19. y
(5)
+ 2y( 3) + 2y" = 3x2 - I
20. y(3) - y = + 7
........,..,,,..
,......y
p
, ..........
,..
21. y" - 2y' + 2y = sin x
22. y
(5)
- y(3) = + 2x2 - 5
23. y" + 4 y = 3x cos 2x
24. y(3) - y" - 1 2y' = x - 2x3x
25. y" + 3y' + 2y = .

26. y" - 6y' + 1 3y = x3x sin 2x


27. y(4) + 5y" + 4y = sin x + cos 2x
28. y(4) + 9y" = (x2 + I ) sin 3x
29. - 1 )
3
( D
2
- 4) y = .+ 2x +

30. y(4) - 2y" + y = x


2
cos x
.....,.......
31. y" + 4y = 2x; y(O) = I , y' (O) = 2
32. y" + 3y' + 2y = y(O) = 0, y' (O) = 3
33. y" + 9y = sin 2x; y(O) = I , y' (O) = 0
34. y" + y = cos x; y(O) = I , y' (O) = -I
35. y" - 2y' + 2y = x + I ; y(O) = 3 , y' (O) = 0
36. y(4) -4y" = x2 ; y(O) = y' (O) = I , y"(O) = y
(3)
(O) = -I
37. y(3) - 2y" + y' = I + xex ; y(O) = y' (O) = 0, y"(O) = I
38. y" + 2y' + 2y = sin 3x; y(O) = 2, y' (O) = 0
39. y(3) + y" = x + -x ; y(O) = I , y' (O) = 0, y" (O) = I
40. y(4)
] = 5; y(O) = y' (O) = y" (O) = y
(3l
(O) = 0
41. Find a particular solution of the equation
y(4) y(3) y" y' 2y = 8x
5
.
42. Find the solution of the initial value problem consisting
of the differential equation of Problem 41 and the initial
conditions
y(O) = y' (O) = y"(O) = y
<
3
)
(0) = O.
43. (a) Write
cos 3x + sin 3x = e
3i
x = (cos x + sin x) 3
by Euler' s formula, expand, and equate real and imag
inary parts to derive the identities
cos
3
x = cos x + cos 3x,
sin
3
x = sin x sin 3x.
(b) Use the result of part (a) to fnd a general solution of
y" + 4 y = cos
3
x.
................
,..........
44. y" + y' + y = sin x sin 3x
45. y" + 9y = sin
4
x
46. y" + y = X cos
3
X
..............,.
.....,.........., ..
,...
47. y" + 3y' + 2y = 4eX
49. y" - 4y' + 4y = 2e
2
x
48. y" - 2y' - 8y = 3-

50. y" - 4y = sinh 2x


51. y" + 4y = cos 3x 52. y" + 9y = sin 3x
53. y" + 9y = 2 sec 3x 54. y" + y = csc
2
x
55. y" + 4y = sin
2
x 56. y" - 4y = xe
x
57. You can verify by substitution that ] = C1 X + C
2
X-
1
is a
complementary function for the nonhomogeneous second
order equation
But before applying the method of variation of parame
ters, you must frst divide this equation by its leading co
efcient x
2
to rewrite it in the standard form
"
I
,
I
3 ] + -y - -y = 72x .
x x2
Thus (x) = 72x
3
in Eq. (22). Now proceed to solve the
equations in (3 1 ) and thereby derive the particular solution
]
r
= 3x
5

..................
..,.......,.....] ..
.,,......,.....
.,...
58. x2y" - 4xy' + 6y = x3 ; ] = C1 X2 + C
2
x3
59. x
2
y" - 3xy' + 4y = X4 ; ] = x2 (C[ +
2
ln x)
60. 4x
2
y" - 4xy' + 3y = 8x4/3 ; ] = C1 X + C
2
X3/4
61. x2y" + xy' + y = In x; ] = Cl cos(ln x) + C
2
sin(ln x)
62. (x
2
- I ) y" - 2xy' + 2y = x
2
- I ; ] = C1 X + C
2
( l + x2)
63. Carry out the solution process indicated in the text to
derive the variation of parameters formula in (33) from
Eqs. (3 1 ) and (32).
64. Apply the variation of parameters formula in (33) to fnd
the particular solution ]
r
(x) = -x cos x of the nonhomo
geneous equation y" + y = 2 sin x.
1 62 Chapter 2 Li near Equati ons of Hi gher Order
Forced Oscillations and Resonance
La|||o;|am
FIGUR 2.6. 1. The cart
with-fywheel system.
iase.:|ea2. 4weae:|vea:|ea|iie:ea:|aieaa:|ea
-x+.x+/x=i,; ( I )
:|a:,eve:as:|eeaea|meas|eaaime:|eaeiamass- :|a:|sa::a.|ea:eas:|a,
,w|:|.eas:aa:/;aaaaaas|e:,w|:|.eas:aa:.;aaa|saisea.:eaea|yaaes:enai
ie:.e i, ; Ha.||aes w|:| :e:a:|a,.emeaea:s .emmeaiy |aveivemasss:|a,
sys:ems,e::|e|:ea|vaiea:s)|aw||.|:|ees:enaiie:.e|ss|mie|a:mea|.
i,; =i, .a . e: i,; =i, |a . , (2)
w|e:e :|e.eas:aa:i,|s :|eami|:aaeei:|ee:|ea|.ie:.eaaa. |s |:s .|:.aia:
i:eaea.y
te: aa esamieei|ew a:e:a:|a, ma.||ae.emeaea:.aa:ev|aea s|m
ie|a:mea|.ie:.e, .eas|ae::|e .a:: w|:| a:e:a:|a, ve::|.ai ayw|eei s|ewa |a
t|,2. 6. I . 1|e.a::|asmass-- -,, ae:|a.iaa|a,:|eayw|eeieimass-, 1|e
.ea::e|aei:|eayw|eei| seii.ea:e:a:aa|s:aa.e.i:em|:s.ea:e:,aaa|:saa,aia:
seea |s .:aa|aase:se.eaa 1|e.a::|s a::a.|ea:ea s:|a, ,w|:|.eas:aa:/;
ass|ewa Assame:|a::|e.ea::e|aei:|e.a::|:seii|s a|:e.:iy|eaea:|:|e.ea:e:
ei:|eayw|eei,aaaaeae:e|yx,; |:sa|sia.emea:i:em|:sea|i||:|ames|:|ea
,w|e:e:|es:|a,|saas::e:.|ea). t|,a:e2. 6. I |eisas:esee:|a::|ea|sia.emea:
xei:|e.ea::e|aei:|e.em||aea.a::iasayw|eei|s,|vea|y
~
,-- -,;x+-,,x+..es.; -,.
x~ x+ .es.
- -
Le:as |,ae:ei:|.:|eaaaaaiy New:ea sse.eaaiaw-x

= -/x, |e.aase:|e
ie:.eese::ea|y:|es:|a,|s-/xwesa|s:|:a:eie:x|a:|eias:eaa:|ea:ee|:a|a
-x

- -,..
:
.es.~ /x,
:|a:| s,
-x+/x=-,..
:
.es. ,;
1|as:|e.a::w|:||:s:e:a:|a,ayw|eeia.:si|ieamasseaas:|a,aaae::|e|a
aaea.eeias|mie|a:mea|.es:enaiie:.ew|:|ami|:aaei, -,..
:
sa.|a
sys:em|sa:easeaa|iemeaeieiai:ea:ieaa|a,was||a,ma.||aew|:|:|e.ie:|es
|e|a,was|eaieaaeaen.ea:e: 1||s|iias::a:es:|e:a.:|.ai|me::aa.eeiaaaiyz
|a,seia:|easei( I ) w|:|es:enaiie:.esas|a(2).
Undamped Forced Oscillations
1e s:aayaaaameaes.|iia:|easaaae::|e|aaaea.eei:|ees:enaiie:.e i,; =
i,.es., wese:.=0|a( I ),aaa:|e:e|y|e,|aw|:|:|eeaa:|ea
-x+/x=i, .a . (4)
w|ese.emiemea:a:yiaa.:|ea|sXc =.,.es.,+.
:
s|a., ue:e
., =
_
Exampl e 1
2. 6 Forced Osci l l ati ons and Resonance 1 63
,as|a.(9)eise.:|ea2 :)|s:|e,.|:.aia:)natnral frequency ei:|emass-s:|a,
sys:em.1|eia.::|a::|eaa,ie.,|smeasa:ea|a,a|meas|eaiess):aa|aas:em|aas
as:|a:|i|smeasa:ea|ase.eaas,s), :|eau |smeasa:ea|a:aa|aase:se:eaa-
:|a:| s, |a|ave:se se.eaas


Aise:e.aiii:em. , I :)|ase.:|eaz.::|a:a|
v|s|eaeia.|:.aia:i:eaea.y.|y:|eaam|e:z- ei:aa|aas|aa.y.ie,|ves:|e
.e::eseaa|a,,e:a|aa:y)frequency 1 =.}z- |auz,|e::z=:y.iese:se:eaa).
Le:asassame|a|:|aiiy:|a::|ees:enaiaaaaa:a:aii:eaea.|esa:ee,./
. = .,

wesa|s:|:a:ex, = A.es.|a. ,:):eaaaaa::|.aia:seia:|ea. ,Ne


s|ae:e:m|saeeaea|ax,|e.aase:|e:e|sae:e:m|aveiv|a,xea:|eiei:|aaas|ae
|a.,:). ) 1||s,|ves
se
aaa:|as
--.
:
A.es.+/A.es.=i,.es.,
i,
A =
/ - -.
:
i,}-
x, ,;=
: :
.a .
.
,
- .
1|e:eie:e,:|e,eae:aiseia:|eax=x,+x,|s,|vea|y

i,}-
x,; = ., .a ., + .
:
m., +
: :
.a . ,
.
,
- .
,;
,6)
(7)
w|e:e:|e.eas:aa:s.
, aaa.
:
a:eae:e:m|aea|y:|e|a|:|aivaiaesx,o;aaax ,o;
a|vaiea:iy,as|a.( I 2)eise.:|eaz. :, we.aa:ew:|:e.(7)as
i,}-
x,; =C.a,.,- o) +
: :
.a . ,
.
,
- .
,s;
sewesee:|a::|e:esai:|a,me:|ea| sasae:es|:|eaei:wees.|iia:|eas,eaew|:|
aa:a:ai.|:.aia:i:eaea.y.,,:|ee:|e:w|:|:|ei:eaea.y.ei:|ees:eoaiie::e.
...-. . . . .. _ u _ . n h h _ _n _ ___ n _ _ _nh_n n u _n ___ n h _ ___ h _ _ h h n _ _ . . . . . . . .
saese:|a:-= I , /=9, i,=s, aaa.=5, se:|ea|ne:ea:|aieaa:|ea|a(4)
| s
x

+x=s .a
t|aax,; |ix ,o;=x ,o;=o
Sol uti on ue:e:|eaa:a:aii:eaea.y., = aaa:|ei:eaea.y. = 5 ei:|ees:e:aaiie::e
a:e aaeaai, as |a :|e :e.ea|a,a|s.ass|ea. t|:s:we sa|s:|:a:ex, = A:es |a
:|ea|iie:ea:|aieaa:|eaaaaaaa:|a:-25A+9A = s, se:|a:A = -5 1|as d
a::|.aia:seia:|ea| s
x, ,;=- .a
1|e.emiemea:a:yiaa.:|ea|sx,=.,.es+.
:
s|a , se:|e,eae:aiseia:|eaei
:|e,|veaaea|eme,eaeeaseaa:|ea| s
x ,; =., .es+.
:
s| a- 5.es ,
w|:|ae:|va:|ve
x,; =-., s|a+.
:
.es+25s|a .
1 64 Chapter 2 Li near Equati ons of Hi gher Order
I a
I 0
a
-a
- I 0
crod `
| I




1|e|a|:|ai.eaa|:|easx ,o;= aaax ,o;= aewy|eia., = aaaC
2
= ,se:ae
aes|:eaa::|.aia:seia:|ea| s
x ,)= .es - .es
As|aa|.a:ea|at| ,. z : z, :|ee:|eaeix , ; |s:|eieas:.emmeamai:|iez-ei:a:
e:|easz- }aaaz-}ei:|e:we.es|ae:e:ms

- I a
Beats
0 K ZK JK 4K aK K
FIGURE 2.6.2. The response
x(t) 5 cos 3t - 5 cos 5t in
Example 1 .
cXump| eZ
x sn a/
I , 0
0, a
/
0.0 @
-0. a
- I . 0
` . \
x sn a/ sn a0/
- I. a
0, 0 0. a I . 0 I . a Z. 0 Z, a J 0
FIGURE 2.6.3. The
phenomenon of beats.
iiwe|mese:|e|a|:|ai.eaa|:|easx ,o;= x ,o;= ea:aeseia:|ea|a(7), weaaa
:|a:
I,
., = - aaa C
2
= ,
-,o- (
2
)
se:|ea::|.aia:seia:|ea| s
I,
x, ; =
2
2
,.es o - :es o, )
-,o
,
- o )
(9)
1|e::|,eaeme::|.|aea:|:yzs|ans|as= .es ,n- s;- .es,n+B) , ai|eaw|:a
n= ,o,+o)aaas= ,o,- o) , eaa|iesas:e:ew:|:e (9)|a:|eie:m
zI,

| |
x, ; =
2 2
SIR ,o,- o)SIR ,o,+o)
-,o
,
- o )
, i ;
saeseaew:|a:o o,,se:aa:o,+o|sve:yia:,e|a.ema:|seaw|:a o,- .
1|ea s|a ,o, +o) | sa mp/a/, va:y|a,iaa.:|ea, w|e:eas s|a ,o, - o)| s a
s/o/,va:y|a,iaa.:|ea wemay:|e:eie:e|a:e::e:, i ;asa:a|aes:|iia:|ea
w|:|.|:.aia:i:eaea.y ,o,+o) ,
x, ; = n, ) s|a ,o,+o) ,
|a:w|:|asiewiyva:y|a,ami|:aae
zI,

|
n, ) =
2 2
SIR ,o,- o)
-,o
,
- o )
w|:|-= i , I,= c, o,= , aaao= :,, i ;,|ves
x,;= s|as|ac
t|,a:ez : s|ews:|e.e::eseaa|a,es.|iia:|eaei i:eaea.y ,o,+.;= c:aa:
|smeaaia:ea|y:|eami|:aae iaa.:|ean, ) = s|a ei i:eaea.y| ,,.;= 5.

A:a|aes.|iia:|eaw|:|a,.ema:a:|veiy)siewiyva:y|a,e:|ea|.ami|:aae
es||||:s:|e|eaemeaeaei/e.s te:esamie,|i:we|e:asae:esa.:iya::aaea
:eeaeaae:|e:s|mai:aaeeasiyiay :|e|:m|aaieC, eaea:o,},z-) = zs Hz aaa
:|ee:|e:a:o},z-) = z:Hz, :aea eae|ea:sa|ea:-aaaaa||ieva:|a:|ea|a:a:
.-p//aeei:ae.em||aeaseaaa-w|:aai:eaea.yei
,o,- o)}z zs- z:
--
= = z(Hz) .
z- z
I , a

I . 0
x /s na0/
Resonance
2. 6 Forced Osci l l ati ons and Resonance 1 65
LooklngatEq. (6), weseethattheamp|ltudeA ofx
r
ls |argewhenthenatura|and
externa| frequencles Wo and W are approxlmate|y equa|. It ls sometlmes usefu| to
rewrlte Eq. (5)lntheform
A =
Fo
=
k - mo
_
Folk
= +
,i,
I - (wlWo)
2
k '
( I I )
whereFolk lsthestatic displacement ofasprlngwlthconstantk duetoa.os.
force Fo, andtheamplifcation factor ,ls dennedtobe
( I 2)
It ls c|ear that , - +> as o - Woo Thls ls the phenomenon ofresonance
the lncrease wlthout bound (as W wo) ln the amp|ltude of oscl||atlons ofan
undamped system wlth natura| frequencyWo lnresponseto an extema|forcewlth
frequencyW u.
We have been assumlng that W = Wo o What sortofcatastrophe shou|done
expectlfo andWo arepreclse|yequa| ! Then Eq. (4),upondlvlslonofeachtermby
m, becomes
l/ _
Fo
x +o
o
x = -cos Wot .
m
( I 3)
Because cosWot ls a term ofthe comp|ementary functlon, the method ofundeter-
mlnedcoemclentsca||sforus totry
x
r
( ) = (Acosoo+sslno,t ) .
We substltute thls lnEq. ( I 3) and thereby nndthat A = 0 and s = Fo
/
(2mwo) .
Hencethepartlcu|arso|utlonl s
Fo
.
x
r
() = -smo,t . ( I 4)
2moo
- I . a

0 00 0 Za 0 a0 0. a I . 00 I . Za I . a0
Thegraphofx
r
() lnFlg. 2. 6. 4(lnwhlchm = I, Fo = I 00,andu = 50)shows
vlvld|y how the amp|ltude ofthe oscl||atlon theoretlca||y wou|d lncrease wlthout
FIGURE 2.6.4. The boundlnthls caseofpeeso..e, W = Woo Wemay lnteqret thls phenomenon
phenomenon of resonance.
as relnforcement ofthe natura| vlbratlons ofthe system by externa||y lmpressed
vlbratlonsatthesamefrequency.
cXump| e
Supposethatm = 5kgandk =500NJmlnthecartwlththeHywhee|ofFlg. 2 6. I .
Then the natura| frequencyl s Wo = ,klm = I 0radJs, that ls, I 0J(2) I . 59
Hz. We wou|dthereforeexpectoscl||atlonsofvery|argeamp|ltudeto occurlfthe
ywhee|revo|vesatabout( I . 59) (60) 95 revo|utlonspermlnute(qm).
Inpractlce,amechanlca|systemwlthvery|ltt|edamplngcanbedestroyedby
resonancevlbratlons. A spectacu|arexamp|ecanoccurwhenaco|umnofso|dlers
marcheslnstepoverabrldge. Anycomp|lcatedstructuresuchasabrldgehasmany
natura|frequenclesofvlbratlon. Ifthefrequencyoftheso|dlers` cadencelsapprox-
lmate|yequa| tooneofthenatura|frequenclesofthestructure,thenustaslnour
slmp|eexamp|eofamassonasprlngresonancewl||occur. Indeed, theresu|tlng
1 66 Chapter 2 Li near Equati ons of Hi gher Order
La|||o;|am
es|t|ea

Exampl e 4
Sol uti on
resonancevlbratlonscan beofsuch|arge amp|ltudethat the brldgewl||co||apse.
Thls has actua||y happenedfor examp|e, theco||apse ofBroughton Brldgenear
Manchester, Eng|and, ln I S3 I andlt ls the reason forthenow-standard practlce
ofbreaklng cadence when crosslnga brldge. Resonancemay havebeen lnvo|ved
ln the I 9SI KansasClty dlsasterln whlch a hote|ba|cony (ca||ed as(.//;co|-
|apsedwlthdancersonlt. Theco||apseofabul|dlnglnanearthquakelssometlmes
dueto resonance vlbratlons caused by thegroundoscl||atlng at oneofthenatura|
frequencles ofthe structure, thls happened to many bul|dlngs ln the Mexlco Clty
earthquakeofSeptember I 9, I 9S5. Onoccaslonanalq|anehascrashedbecauseof
resonant wlng oscl||atlons causedby vlbratlons ofthe englnes. It ls reported that
for some ofthe nrst commercla| et alrcraft, the natura| frequency ofthe vertlca|
vlbratlons ofthe alq|ane durlng turbu|ence was a|most exact|y that ofthe mass
sprlng systemconslstlngofthepl|ot` shead (mass) andsplne(sprlng). Resonance
occurred,causlngpl|otstohavedlfncu|tylnreadlngthelnstruments. Largemodem
commercla|etshavedlerentnatura|frequencles, sothatthls resonanceprob|em
no|ongeroccurs.
Modeling Mechanical Systems
Theavoldanceofdestructlveresonancevlbratlonslsanever-presentconslderatlon
ln the deslgn ofmechanlca| structures and systems ofa|| types. Olten the most
lmportant steplndetermlnlngthenatura|frequencyofvlbratlon ofasystemls the
formu|atlonoflts dlerentla| equatlon. In addltlon to Newton' s|aw F = m., the
prlnclp|eofconservatlonofenergyls sometlmesusefu| for thls purpose(as lnthe
derlvatlon of the pendu|um equatlon ln Sectlon 2. 4). The fo||owlng klnetlc and
potentla|energyformu|asareoenusefu| .
1. x/e/.ee,, = tmv
2
fortrans|atlonofamassm wlthve|ocltyv;
Z. x/e/.ee,, = t I w
2
for rotatlon ofa bodyofa momentoflnertla I
wlthangu|arve|ocltyw;
J. ioe/./ee,, \ = tkx
2
for a sprlngwlth constant/ stretchedorcom-
pressedadlstancex;
4. ioe/./ee,, \ = -,/forthegravltatlona|potentla|energyofamassm
athelght/abovethereference|eve| (the |eve|atwhlch\ = 0), provldedthat
,mayberegardedasessentla||yconstant.
Flnd thenatura|frequency ofa massm ona sprlng wlth constant /lf, lnstead of
s|ldlng wlthoutfrlctlon, ltlsaunlformdlskofradlus.thatro||swlthouts|lpplng,
asshownlnFlg2. 6. 5.
Wlththeprecedlngnotatlon,theprlnclp|eofconservatlonofenergyglves
tmv
2
+t 1w
2
+tkx
2
= i
where i ls a constant (the tota| mechanlca| energy ofthe system). We note that
v = .w and reca|| that I = m.
2
J2 for a unlform clrcu|ar dlsk. Then we may
slmp|lfythe|astequatlonto
mv
2
+tkx
2
= i
Becausetherlght-handsldeofthlsequatlonl sconstant,dlfferentlatlonwlthrespect
x = o
to(wlth v =x' and v' =x") nowglves
F GU 2 6 5

/ //
/
/
0 I R The rolling disk. 'mx x + xx =
Exampl e
2. 6 Forced Osci l l ati ons and Resonance 1 67
wea|v|aeea.|:e:m|y-x :ee|:a|a

z/
x + -x = o

1|as:|eaa:a:aii:eaea.yei|e:|zea:ai|a.|aaaie::|es.|iia:|eaeiea::eii|a,
a|s| |s z/}, w||.||s z} s i : :|mes :|e iam|i|a:aa:a:ai i:eaea:y
/}eiamasseaas:|a,:|a:| ssi|a|a,w|:|ea:i:|.:|ea:a:|e::|aa:eii|a,w|:|ea:
si|a|a, i:|s|a:e:es:|a,,aaae:|assaq:|s|a,):|a::||saa:a:aii:eaea.yaeesae:
aeeaaea:|ema/sei:|ea|s| i:.eaia|ee|:|e:aa|mee:aia:,ea|s|w|:|a
:aa|aseieaeme:e:,|a:ei:|esamemass)
saese:|a:a.a:es.|iia:esve::|.aiiyas|i|:we:eamass = s|,eaas|a,ie
s:|a,,w|:|.eas:aa:/=:Z i
-
N}m), a::a.|ea:eas|a,ieaas|e:,w|:|:eas:aa:
. = N s}m) saese:|a::||s.a:w|:|:|eaas|e:a/s.oe.e1|sa:|vea
aiea,awas||ea:a:eaa sa:ia.e w|:|aa ami|:aaeei .maaaa waveiea,:| ei
i= i m,t|, z : :; A:w|a:.a:seeaw|ii:eseaaa.ev||:a:|ease:.a::
y= a ees
s = o
FIGURE 2.6.6. The washboard
road surface of Example 5.
es|t|ea
FIGURE 2.6.7. The "unicycle
model" of a car.
Sol uti on we :||a|ei:|e.a:asaaa|.y.ie, as|.:a:ea|at| , z : : Le:x,;aeae:e:|e
awa:aa|sia.emea:ei :|e mass i:em |:sea|i||:|ames|:|ea, we|,ae:e:|e
ie:.eei,:av|:y,|e.aase|:me:eiya|sia.es:|eea|i||:|ames|:|eaas|ar:e|iem
9eise.:|eaz : wew:|:e:|eeaa:|eaei:|e:eaasa:ia.eas
z-s
y=a.es

,a= m,i= i m) , i ;
w|ea:|e.a:| s|ame:|ea,:|es:|a,|ss::e:.|ea|y:|eameaa:x- y. seNew:ea s
se.eaaiaw,F =a, ,|ves
-x=-/,x- y; ,
:|a:| s,
-x+ /x=/y , i :;
ii:|eveie.|:yei:|e.a:|sc, :|eas=c|a, i ;, se, i :;:a|es:|eie:m
,
z-c
-x + /x=/a .es

, i :;
1 68 Chapter 2 Li near Equati ons of Hi gher Order
1||s| s:|ea|iie:ea:|aieaa:|ea:|a:,evens:|eve::|.aies.|iia:|easei:|e.a:. ia
.ema:|a, |: w|:| . ,:), we see:|a: we |aveie:.ea es.|iia:|easw|:|.|:.ai
i:eaea.yo= z-c}I. keseaaa.ew|iie..a:w|eao= u = /}. weaseea:
aame:|.aiaa:a:eaaa:|eseeaei:|e.a:a::eseaaa.e
I l c l c
-
c~ = l :. s,m}s),
z- z- scc
:|a:| s, a|ea:33. 3m|}|,as|a,:|e.eave:s|eaia.:e:eiz. zm|}|e:m}s).
Damped Forced Oscillations
ia:eai|ys|.ai sys:ems:|e:e| saiways semeaam|a,,i:emi:|.:|eaaiene.:s|i
ae:||a,eise. 1|e.emiemea:a:yiaa.:|eaxei:|eeaa:|ea
x+cx+/x= i,.eso , l )
|s,|vea|y. , l ), ,zc) ,e:,zl ) eise.:|eaz. :, aeeaa|a,eaw|e:|e:c ~ c
; *
:/, c = c; , e:c < c; . 1|ese.|a.ie:m |s ae: |me::aa: |e:e. w|a:|s
|me::aa:|s :|a:, |a aay.ase, :|eseie:maiass|ew :|a:x ,) cas +.
1|asx |satransient solution ei. , l )-ae:|a:a|esea:w|:|:|eassa,eei
:|me,ieav|a,eaiy:|ea::|.aia:seia:|eax
,
.
1|eme:|eaeiaaae:e:m|aea.eeu.|ea:s|aa|.a:es:|a:wes|eaiasa|s:|:a:e
x , ) = n.eso+ss|ao
|a. , l ). w|eaweae:|| s, .eiie.::e:ms, aaaeaa:e.eeia.|ea:sei.esoaaa
s|ao ,wee|:a|a:|e:weeaa:|eas
,/- o

) n+cos= i, , -con+,/- o

) s=c , l s)
:|a:weseivew|:|ea:a|u.ai:yie:
, l )
iiwew:|:e
n.eso+ss|ao= c,.eso.esa+s|aos|aa) = c.es,o- a)
asasaai,wesee:|a::|e:esai:|a,s:eaaye:|ea|.es.|iia:|ea
x
,
, ) = c.es ,o- a) ,zc)
|asami|:aae
,zl )
New, l )|mi|es:|a:s|a a = s}c ~ c, s e| : ieiiews:|a::|e|aseaa,ieai|es|a
:|ea:s:e:se.eaaaaa:aa:.1|as
s co
:aa a= = w|:| c< a < -,
n /- o

,zz)
Exampl e
se
2. 6 Forced Osci l l ati ons and Resonance 1 69

:aa
,
..
|i/ > o

,
/ - o

o =
..
-+:aa
,
|i/ < o

/ - o

,w|e:easo =-n|i/=o

) .
Ne:e:|a:|i. > c, :|ea:|eie:.eaami|:aae-aeaaeaasaiaa.:|eac,.
|y,zl )-aiways:ema|asaa|:e, |a.ea::as:w|:|:|e.aseei:eseaaa.e|a:aeaa
aamea .ase w|ea :|e ie:.|a, i:eaea.y o eaais :|e .:|:|.ai i:eaea.y W
o

/}. ua::|eie:.eaami|:aaemay a::a|a amas|mamie: semevaiaeei.,|a


w||.| .ase we seai eip../../ eso..e 1e see|iaaaw|ea :a.:|.ai:es
eaaa.ee..a:s, we aeea eaiy,:a| C as aiaa.:|eaeio aaaiee|ie:aie|ai
mas|mam. i:.aa|es|ewa,r:e|iemz):|a:C |sas:eaa|iyae.:eas|aiaa.:|ea
eio |i.
z/. ua:|i. <
z/, :|ea:|eami|:aaeeic a::a|asamas|
mamvaiae-aaase:a.:|.ai:eseaaa.ee..a:s-a:semevaiaeeioiess:aaao,,
aaa:|eaa:ea.|esze:easo +> i:ieiiews:|a:aaaaae:aameasys:em
:y|.aiiyw|iiaaae:eie:.eaes.|iia:|easw|eseami|:aae|s
La:,e|io|s.iese:e:|e.:|:|.ai:eseaaa.ei:eaea.y,
Ciese:eI,}/|io|sve:ysmaii ,
Ve:ysmaii|io|sve:yia:,e.
... .
t|aa:|e::aas|ea:me:|eaaaas:eaaye:|ea|.es.|iia:|easeiaaameamassaaa
s:|a,sys:emw|:| = l , . = z, aaa/ = z:aaae::|e|aaaea.eeiaaes:eoai
ie:.ei,; = sz.es:w|:|x,c)= :aaa x ,c, = c.Aise|aves:|,a:e:|eess|||i|:y
ei:a.:|.ai:eseaaa.eie::||ssys:em.
Sol uti on 1|e:esai:|a,me:|eax,; = x,, ,;+xg , ei:|emasssa:|saes:ae|a|:|aivaiae
:e|iem
x+zx+z:x= sz.es: , x,c)= :, x,c) = c. ,z,
ias:eaaeiaiy|a,:|e,eae:aiie:maiasae:|veaea:i|e:|a:||sse.:|ea,|:|s|e::e:|a
a.ea.:e:e:e|iem:ewe:i|:a|:e.:iy.1|e:ee:sei:|e.|a:a.:e:|s:|.eaa:|ea

+z+z:= ,+l )

+z=c
a:e= -l +/ ,se:|e.emiemea:a:yiaa.:|ea|s
w|eawesa|s:|:a:e:|e::|aiseia:|ea
x,; = n.es :+ss|a :
|a:|e,|veaeaa:|ea,.eiie.:i|ie:e:ms, aaaeaa:e.eeu.|ea:sei.es:aaas|a: .
we,e::|eeaa:|eas
l cn+ ss= sz,
-sn+ t cs= c
1 70 Chapter 2 Li near Equati ons of Hi gher Order
wlthso|utlonA =, s=:. Hencethegenera|so|utlonoftheequatlonl n,z)l s
x, ) =e
-
(c
|
cos+c
:
sl n ) +cos:+:sl n: .
Atthlspolntwe lmpose the lnltla| condltlonsx,c) = :, x ,c) = cand nnd that
c
|
= I and c
:
= -. Therefore, the translent motlon and the steady perlodlc
oscl||atlonofthemass areglvenby
and
x,, , ) =,

,.es - s|a )
x, ,)=.es :+: s|a :=
_
.es :+ sln :,
=cos ,:- :;
where:=tan
|
, , c. ::.
Flgurez. :. sshowsgraphsoftheso|utlonx,)=x,, ,,+x, ,)ofthelnltlal
va|ueprob|em
x' + zx + z:x = sz .es : x,c) = x,, x ,c) = c ,z:,
forthedlfferentva|uesx, = -zc, -l c, c, I 0, andzcofthelnltla| posltlon. Here
we see c|ear|y what lt means for the translent so|utlon x,, ,)to dle out wlth the
passageoftlme, |eavlngon|y the steadyperlodlc motlonx,, ) . Indeed, because
x,, , ) - cexponentla||y, wlthln a very few cyc|es the fu|| so|utlon x, , and the
steady perlodlc so|utlonx, ,)are vlrtua||y lndlstlngulshab|e (whateverthe lnltla|
posltlonx,) .

, :a
:
a
| a

| a
-:a
FIGURE 2.6.8. Solutions of the initial value problem in (24)
with Xq = -20, -1 0, 0, 1 0, and 20.
2. 6 Forced Osci l l ati ons and Resonance 1 71
I 0

o
lractcaI rcsonancc
Tolnvestlgate theposslbl|lty ofpractlca| resonanceln theglvensystem, we
substltute the va|ues = i , . = z, and/ = z:ln ,zl ) and nndthattheforced
amp|ltudeatfrequencyols
1

\ a
4
sz
c,o)=
::- :so

+o
-
J
Z
Thegraphofc,o)ls shownlnFlg. z. :. Themaxlmumamp|ltudeoccurswhen
I a Z0
/
-:l,:o

- :o)
c,o)= -
,:

-
)

-l ::o,o

- z:)
-- -= - 0
,::- :so

+
-

FIGURE 2.6.9. Plot of


amplitude versus external
frequency .
Thuspractlca|resonanceoccurswhentheexterna|frequencylso= .(ablt|ess
thanthemass-and-sprlng` sundampedcrltlca|frequencyofo,= /} = .) .
Q Problems
..1 .... .,..........
...,.............,
......,.........,
...,..............
.........,........ .,
.
1. .. l O cos . . 0
2. ... 5 sin . . 0
3. .1 00x 225 cos 5300 sin .(0) 375,
. 0
4. .25x 90 cos . ; . 0, . 90
5. ... q cos .with ..uq , .(0) .q, .(0) 0
6. ... qcos wt with . uq , . 0, . 0q
..............,..
..., cos .- a) ...,....
... ..,......
,........,.., ....,...
........... j .
7. ..... I cos 3
8. .. . -. cos 5
9. 2X" .. 3 sin l Ot
10. .. . 8 cos l 6 sin
....11 .........,..
...,....., cos .- a) ...
., ..,............... ....
...........
11. .... l O cos . . 0
12. ..1 3x 1 0 sin . . 0
13. ..26x 600 cos l Ot ; . 1 0, . 0
14. ..25x 200 cos 520 sin . -30,
. - 1 0

..........,...
.....,......,....,.....
.. q cos ......,..,..
........,.......,...
...,........,.....
...,........,.....,..
...
15. I , 2, . 2, q 2
16. m = l , c = 4, k = 5, Fo = 1 0
17. l , 6, . .5, q 50
18. I, 1 0, . 650, q 1 00
19. A mass weighing 1 00 lb (mass 3. 1 25 slugs in fps
units) is attached to the end of a spring that is stretched
I in. by a force of 1 00 lb. A force Fo cos .acts on the
mass. At what frequency (in hertz) will resonance oscilla
tions occur? Neglect damping.
20. A front-loading washing machine is mounted on a thick
rubber pad that acts like a spring; the weight = ,
(with . 9. 8 mjs
2
) of the machine depresses the pad ex
actly 0. 5 cm. When its rotor spins at .radians per second,
the rotor exerts a vertical force Fo cos .newtons on the
machine. At what speed (in revolutions per minute) will
resonance vibrations occur? Neglect friction.
21. Figure 2. 6. 1 0 shows a mass on the end of a pendulum
(of length .also attached to a horizontal spring (with
constant . Assume small oscillations of so that the
spring remains essentially horizontal and neglect damp
ing. Find the natural circular frequency of motion of
the mass in terms of ..and the gravitational con
stant .
1 72 Chapter 2 Li near Equati ons of Hi gher Order
|
FIGURE 2.6. 10. The pendulum
and-spring system of Problem 2 1 .
22. A mass m hangs on the end of a cord around a pulley of
radius .and moment of inertia as shown in Fig. 2. 6. 1 1 .
The rim of the pulley is attached to a spring (with constant
k). Assume small oscillations so that the spring remains
essentially horizontal and neglect friction. Find the natu
ral circular frequency of the system in terms of m, .k,
and g.

FIGURE 2.6. 11. The mass-spring


pulley system of Problem 22.
23. A building consists of two foors. The frst foor is at
tached rigidly to the ground, and the second foor is of
mass m 1 000 slugs (fps units) and weighs 1 6 tons
(32,000 Ib) . The elastic frame of the building behaves as a
spring that resists horizontal displacements of the second
foor; it requires a horizontal force of 5 tons to displace the
second foor a di stance of 1 ft. Assume that in an earth
quake the ground oscillates horizontally with amplitude
Ao and circular frequency w, resulting in an exteral hor
izontal force F(t) mAow
2
sin wt on the second foor.
(a) What is the natural frequency (in hertz) of oscillations
of the second foor? (b) If the ground undergoes one
oscillation every 2. 25 s with an amplitude of 3 in., what
is the amplitude of the resulting forced oscillations of the
second foor?
24. A mass on a spring without damping is acted on by the
exteral force F (t) Fo cos
3
wt. Show that there are .
values of w for which resonance occurs, and fnd both.
25. Derive the steady periodic solution of
mx" + CX' + kx Fo sin wt.
In particular, show that i t i s what one would expect-the
same as the formula in (20) with the same values of and
w, except with sin(wt - a) in place of cos (wt - a) .
26. Given the diferential equation
mx" + cx' + kx Eo cos wt + Fo sin wt
-with both cosine and sine forcing terms--erive the
steady periodic solution

E
2
+ F
2
xg_ (t)
0 0
cos (wt - a - p) ,
.(k - m(2) 2 + (cw)
2
where a is defned in Eq. (22) and p tan-
I
( Fo/Eo) .
.....Add the steady periodic solutions separately
corresponding to Eo cos wt and Fo sin wt (see Problem
25). )
27. According to Eq. (21 ), the amplitude of forced steady
periodic oscillations for the system mx" + cx' + kx
Fo cos wt is given by
28.
(a) If c ccr/h, where Ccr ,4km, show that
steadily decreases as w increases. (b) If c ccp/.,
show that attains a maximum value (practical reso
nance) when
As indicated by the cart-with-fywheel example discussed
in this section, an unbalanced rotating machine part typ
ically results in a force having amplitude proportional to
the .,..of the frequency w. (a) Show that the am
plitude of the steady periodic solution of the diferential
equation
mx" + cx' + kx mAw2 cos wt
(with a forcing term similar to that in Eq. ( 1 7 is given by
(b) Suppose that c
2
2mk. Show that the maximum
amplitude occurs at the frequency uq given by
uq
k 2mk
;; 2mk - c
2
Thus the resonance frequency in this case is ..(in
contrast with the result of Problem 27) than the natural fre
quency uq ,k/m. .,,..Maximize the .,..
of
2. 7 El ectri cal Ci rcui ts 1 73
Automobile Vibrations
.29 .............,
..,.....,..........,...
.+ .+ .. . + .............
..... 0). .. .sin .......
..., ..,....
amplitude slightly over 5 cm. Maximum resonance vibra
tions with amplitude about 14 cm occur around 32 mi/h,
but then subside to more tolerable levels at high speeds.
Verify these graphically based conclusions by analyzing
the function . In particular, fnd the practical reso
nance frequency and the corresponding amplitude.
.+ .+ .. cos .+ sin .
I a
. ..... ..
I Z
29. Apply the result of Problem 26 to show that the amplitude
of the resulting steady periodic oscillation for the car is
given by
5
c
O
7

9
7

5
<
Because . 2U/.when the car is moving with velocity
U, this gives as a function of U.
3
30. Figure 2. 6. 1 2 shows the graph of the amplitude function
.using the numerical data given in Example 5 (in
cluding 3000 N s/m). It indicates that, as the car
accelerates gradually from rest, it initially oscillates with
Z0 40 0 o0 I 00
VcIocty (mm)
FIGURE 2.6. 12. Amplitude of vibrations of
the car on a washboard surface.
Electrical Circuits
L
L
K
FIGURE 2.7. 1. The series ..
circuit.

Inductor
Resistor
Capacitor
.
.
.
.
1
c
FIGURE 2.7.2. Table of voltage
drops.
Hereweexamlnethericclrcultthatlsabaslcbul|dlngb|ocklnmorecomp|lcated
e|ectrlca|clrcultsandnetworks. As shownln Flg. 2. 7 I , ltconslstsof
Aresistor wlthareslstanceofro/-s,
Aninductor wlthanlnductanceofi/e/es,and
Acapacitor wlthacapacltanceofc).mas
ln serles wlth a source ofe|ectromotlve force (such as a battery or a generator)
that supp|les a vo|tage of i,;.o/sat tlme Ifthe swltch shown ln the clrcult
ofFlg. 2. 7. I ls c|osed, thls resu|ts ln a currentofi,;.-peesln the clrcultand
a charge of _, ; .o/o-/son the capacltor at tlme The re|atlon between the
functlons _andils
= i ,; ( I )
We wl||a|waysusemkse|ectrlcunlts,l nwhlchtlmel s measuredl nseconds.
Accordlng to e|ementary prlnclp|es ofe|ectrlclty, thevoltage drops across
thethreeclrculte|ementsarethoseshown lnthetab|elnFlg.2 7 2. Wecanana|yze
thebehavlorofthe serles clrcultofFlg. 2. 7. I wlth the aldofthls tab|eandoneof
Klrchho` s|aws.
The(a|gebralc)sumofthevo|tagedropsacrossthee|ementslna
slmp|e|oopofane|ectrlca|clrcultls equa|totheapp|ledvo|tage.
As a consequence, the current and charge lnthe slmp|e ricclrcultofFlg. 2 7 I
satlsfythebaslcclrcultequatlon
ai I
i- + ri + -_ = i,;
a c
(2)
1 74 Chapter 2 Li near Equati ons of Hi gher Order
Ifwesubstltute( 1 ) lnEq. (2), wegetthesecond-order|lneardlerentla|equatlon
1
i _

+r_+-_ = i,;
c
forthecharge _, ; , undertheassumptlonthatthevo|tagei,; lsknown.
(3)
Inmostpractlca|prob|emsltlsthecurrent iratherthanthecharge _thatls
ofprlmarylnterest,sowedlfferentlatebothsldesofEq. (3) andsubstltuteifor Q'
toobtaln
I
ii

+ri+-i = i ,;
c
,:,
Wedooassumehereaprlorfaml|larltywlthe|ectrlca|clrcults. Itsufncesto
regardthereslstor, lnductor, andcapacltorlnane|ectrlca|clrcultas b|ackboxes
thatareca|lbratedbytheconstants r,i,andcAbatteryorgeneratorlsdescrlbed
bythevo|tagei,;thatltsupp|les. Whentheswltchlsopen,nocurrentHowslnthe
clrcult,whenthe swltchls c|osed, there ls acurrenti,;ln the clrcultandacharge
_, ; onthe capacltor. A||we needtoknowabouttheseconstants andfunctlonsls
thattheysatlsfyEqs. ( 1 ) through ,:),ourmathematlca|mode|forthericclrcult.
Wecanthen|earnagooddea|aboute|ectrlcltybystudylngthlsmathematlca|mode|.
The Mechanical-Electrical Analogy
Itls strlklngthatEqs. (3)and,:)havepreclse|ythesameformastheequation
x+.x+/x= i,; (5)
ofamass-sprlng-dashpotsystemwlthextema|force i,; Thetab|elnFlg. 2. 7. 3
detal|sthls lmportantmechanical-electrical analogy. As aconsequence, mostof
theresu|tsderlvedln Sectlon2. 6formechanlca| systemscanbe app|ledatonceto
e|ectrlca|clrcults. Thefactthatthe samedlerentla|equatlonservesasamathemat-
lca|mode|forsuchdlerentphyslca|systemslsapowerfu|l||ustratlonoftheunlfy-
lngro|eofmathematlcslnthelnvestlgatlonofnatura|phenomena. Moreconcrete|y,
thecorrespondenceslnFlg. 2. 7. 3canbeusedtoconstructan e|ectrlca| mode|ofa
glvenmechanlca| system,uslnglnexpenslveandreadl|yaval|ab|eclrculte|ements.
Theperformanceofthemechanlca| systemcanthenbepredlctedbymeansofac-
curatebut slmp|emeasurements lnthee|ectrlca| mode|. Thls ls especla||y usefu|
whenthe actua|mechanlca| systemwou|dbeexpenslvetoconstructorwhenmea-
surements ofdlsp|acements and ve|ocltles wou|d be lnconvenlent, lnaccurate, or
evendangerous. Thls ldea ls thebaslsof../o,.oues|ectrlca|mode|sof
mechanlca| systems. Ana|ogcomputersmode|edthe nrstnuc|earreactorsforcom-
mercla|powerandsubmarlnepropu|slonbeforethereactorsthemse|veswerebul|t.
Mass m
Damping constant L
Spring constant .
Position
Force F
Inductance .
Resistance .
Reciprocal capacitance 1 /
Charge Q (using (3) (or current using (4)))
Electromotive force (or its derivative .
FIGURE 2.7.3. Mechanical-electrical analogies.
2. 7 El ectri cal Circuits 1 75
Inthetyplca|caseofana|tematlngcurrentvo|tageE(t ) E,slnot, Eq. (4)
takestheform
I
LI

+RI
'
+
C
I oE, cos ot . (6)
As ln a mass-sprlng-dashpot system wlth a slmp|e harmonlc extema| force, the
so|utlon of Eq. (6) ls the sum ofa transient cnrrent I,, that approaches zero as
t - +>(under the assumptlon that the coemclents ln Eq. (6) are a|| posltlve,
so the roots ofthe characterlstlc equatlon have negatlve rea| parts), and asteady
periodic current I,
-
,thus
I+I,
,
. (7)
Reca||fromSectlon2. 6(Eqs ( I 9)through(22)there)thatthesteadyperlodlcso|u-
tlonofEq. (5)wlthF(t ) F,cosotls
where
F,cos (ot- :;



J(k - -.
:
;
:
+(.o)
:
.|
..
: = :+a
:

o : -
k - mw
Ifwemakethe substltutlons L for -, R for .. I J Cfor k, andoE,forF,, weget
thesteadyperlodlccurrent
wlththephaseang|e
.|
oRC
: tan , o : -
I - LCo
:
Reactance and Impedance
Thequantltylnthedenomlnatorln(S),
Z R
:
+ uL ~
u
, (ohms) ,
ls ca||edtheimpedance oftheclrcult. Thenthe steadyperlodlccurrent
has amp|ltude
E,
I,
-
(t)
.
cos (ot- :;
E,

remlnlscentofOhm` s|aw, I EJR.


(S)
(9)
( I 0)
( I I )
( I 2)
1 76 Chapter 2 Li near Equati ons of Hi gher Order
-S
K
FIGURE 2.7.4. Reactance and
delay angle.
FIGURE 2.7.5. Time lag of
curent behind imposed voltage.
cXump| e 1
Equatlon, l l ) glvesthe steady perlodlc current asacoslnefunctlon,whereas
the lnput vo|tage i,; = i,slnot was a slne functlon. To convert i,to a slne
functlon,wenrstlntroducethereactance
l
= .i -
.c
, l ,
Then Z =r
:
+
:
,andweseefromEq. (9)thato lsaslnFlg. z :, wlthde|ay
ang|e =o - | - Equatlon, l l ) nowyle|ds
Therefore,
where
i, ,;= (cosL cos.+slnosln. ;
=
i,

cos .+
r
sln .
,
Z Z Z
i,

. .

= (cosU sm.- smU cos. ;


Z
i,
.

i, ,;= m,. U
Z
ic.
:
- l
=tan
|
=tan
|

r .rc
, l :,
, l ,
Thls nna||y glves the time lag oJo (l nseconds) ofthe steady perlodlc current i,
behlndthelnputvo|tage(Flg. z )
Initial Value Problems
Whenwewantto nndthetranslentcurrent, weare usua||yglventhelnltla| va|ues
i,c)and Q(0) . Sowemustnrstnnd i,c) Todoso, wesubstltute=clnEq. ,z,
toobtalntheequatlon
ii,c)+ri ,c)+Q(0) =i,o; , l ,
todetermlne i,c, lnterms ofthelnltla|va|uesofcurrent, charge, andvo|tage.
Conslderan ricclrcult wlth r = cohms (L), i = c l henry (H), and c
Z l c

-
farad(F) . Attlme c,when both i,c)and Q,c,are zero, theclrcultls
connectedto a l l cV,cHz a|ternatlng current generator. Flnd the current lnthe
clrcultand thetlme|agofthesteadyperlodlccurrentbehlndthevo|tage.
Sol uti on
Afrequencyof:cHzmeansthat. ,z-) ,c)radJs, approxlmate|yradJs. So
wetakei, = l l cslnanduseequa|ltylnp|aceof thesymbo|forapproxlmate
equa|ltylnthlsdlscusslon. Thedlerentla|equatlonln,)takestheform
,c l ) t+c/+zccc/=,) , l l c,cos
cXump| eZ
2. 7 El ectrical Ci rcuits 1 77
Wesubstltutetheglven va|uesofR, L, C, and.=377lnEq. ( I 0)tonndthatthe
lmpedancels Z =59. 5SL, sothesteadyperlodlcamp|ltudels
I I 0(vo|ts)
I
a
= = I . S46amperes (A) .
59. 5S(ohms)
Wlththesamedata,Eq. ( I 5)glvestheslnephaseang|e.
=tan

,
(0. 64S) =0. 575.
Thusthetlme|agofcurrentbehlndvo|tagels
=
0. 575
=0. 00I 5s
. 377

andthesteadyperlodlccurrentls
I,
,
=( I . S46)sln(377t- 0. 575) .
Thecharacterlstlcequatlon(0. I ) r
:
+ 50r +2000=0hasthetworootsr]
-44andr
:
456. Wlththeseapproxlmatlons, thegenera| so|utlonls
I (t ) =c, e
++
' +c
:
e
+-
' + ( I . S46)sln(377t- 0. 575) ,
wlthderlvatlve
I
'
(t) =-44c, e
++
' 456c
:
e
+
-
' +696 cos (377t- 0. 575)
Because I (0) = _,o; = 0, Eq. ( I 6) glves I' (0) = 0 as we||. Wlth these lnltla|
va|uessubstltuted,weobtalntheequatlons
I (0) =c, +c
:
- I . 004 =0,
I
'
(0) =-44c, - 456c
:
+5S4=0,
thelrso|utlonlsc, =-0. 307, c
:
I . 3 I I . Thusthetranslentso|utlonls
I,, (t ) = (-0. 307) e
++
'
+ ( I . 3 I I ) e
+-
' .
The observatlon that after one-nfth ofa secondwehave I,, (0. 2) < 0. 000047 ^
(comparab|etothecurrentlnaslng|ehumannervenber) lndlcatesthatthetranslent
so|utlondlesoutveryrapld|y,lndeed

Supposethatthe RLC clrcultofExamp|e I , stl|| wlth I (0) = _,o; = 0, ls con-


nectedattlme t = 0toabattery supp|ylngaconstant I I 0V Now nndthecurrent
lntheclrcult
1 78 Chapter 2 Li near Equati ons of Hi gher Order
Sol uti on Wenowhavei, ; = I I 0,soEq.( I 6)glves
FIGURE 2.7.6. The efect of
frequency on 10
,
i,o; I I 0
i (0) = -= -= I I 00(As),
I 0. I
andthedlerentla|equatlonls
(0. I ) l

+50I'+2000I = i ,;= 0.
Itsgenera|so|utlonlsthecomp|ementaryfunctlonwefoundlnExamp|e I .
i ,; = c, e
++
' +c
:
e
+
-
'
Weso|vetheequatlons
I (0) = c
(
+c
:
= 0,
i(0) = -44c, - 456c
:
= I I 00
forc, = -c
:
= 2. 670. Therefore,
i,;= (2. 670) (e
++
'
a
e
+
-
' ) .
Notethati ( ) 0as +ceventhoughthevo|tagelsconstant.
Electrical Resonance

Conslderagalnthecurrentdlfferentla|equatlonln(6)correspondlngtoaslnusolda|
lnput vo|tage i,; = i,sln. We have seen that the amp|ltude oflts steady
perlodlccurrentls
i, i,
I
a
=
-
=

k
:
+ oI~
u

( I 7)
Fortyplca|va|uesoftheconstantsk, I, c, andi,,thegraphofI
a
asafunctlonof
resemb|estheoneshownlnFlg. 2. 7. 6. Itreachesamaxlmumva|ueatu_ = t}Ic
andthenapproacheszeroas. +c, thecrltlca|frequencyu_ lstheresonance
frequency oftheclrcult.
In Sectlon 2.6 weemphaslzedthelmportanceofavoldlngresonancelnmost
mechanlca|systems(thece||olsanexamp|eofamechanlca|systemlnwhlchreso-
nancelsso,/; By contrast, many commone|ectrlca|devlces cou|dnotfunctlon
proper|y wlthouttaklngadvantageofthephenomenonofresonance. Theradlolsa
faml|larexamp|e. Ahlgh|yslmp|lned mode| ofitstunlngclrcultls thekIcclrcult
wehave dlscussed. Its lnductance Iandreslstance kareconstant,butlts capacl-
tanceclsvarledasoneoperatesthetunlngdla| .
Suppose that we wanted to plck up a partlcu|ar radlo statlon that ls broad-
castlng at frequency ., and thereby (ln effect) provldes an lnput vo|tage i() =
i,sln.tothetunlngclrcultoftheradlo. Theresu|tlngsteadyperlodlccurrenti,
lnthetunlngclrcultdrlvesltsamp|lner,andlntumlts|oudspeaker,wlththevo|ume
ofsound wehearrough|yproportlona|totheamp|ltude I
a
ofi, Tohearourpre-
ferred statlon (offrequency .;the |oudestand slmu|taneous|y tune out statlons
broadcastlngatotherfrequencleswethereforewantto choose cto maxlmlze I
a
.
2. 7 El ectrical Ci rcuits 1 79
ButexamlneEq. ( I 7), thlnklngofw asaconstantwlthCtheon|y varlab|e. Wesee
atag|ancenoca|cu|usrequlredthatI
a
ls maxlma|when
thatls, when
I
wL - = 0,
oC
I
C = .
Lw
2
( I S)
Sowemere|y tumthedla| to setthecapacltancetothlsva|ue.
Thlslsthewaythato|dcrysta|radlosworked,butmodemAMradloshavea
moresophlstlcateddeslgn. Ap./ofvarlab|ecapacltorsareused. Thenrstcontro|s
thefrequency se|ected asdescrlbedear|ler, the secondcontro|sthefrequencyofa
slgna| that the radlo ltse|fgenerates, kept c|oseto 455 kl|ohertz (kHz) above the
deslredfrequency. Theresu|tlng/e.frequencyof455 kHz, knownasthe/ee
a/.e)e,e.,, lsthenamp|lnedlnsevera|stages. Thlstechnlquehastheadvantage
thatthesevera|RLCclrcultsusedlntheamp|lncatlonstageseasl|ycanbedeslgned
toresonateat455 kHzandreectotherfrequencles,resu|tlnglnfarmorese|ectlvlty
oftherecelveraswe|| asbetteramp|lncatlonofthedeslredslgna|.
Problems "
.1 ............. .
.......................
......................
...,......,
........,...
..
|
R
FIGURE 2.7.7. The circuit for
Problems through
1. In the circuit of Fi g. 2. 7. 7, suppose that . 5 . . 25
, and the source of emf is a battery supplying V
to the circuit. Suppose also that the switch has been in po
sition for a long time, so that a steady current of .A is
fowing in the circuit. At time the switch is thrown
to position 2, so that .and for Find

2. Given the same circuit as i n Problem I , suppose that the
switch is initially in position 2, but is thrown to position
at time so that and for
Find and show that .as
3. Suppose that the battery in Problem 2 is replaced with
an alterating-current generator that supplies a voltage of
cos 60t volts. With everything else the same,
now fnd
4. I n the circuit of Fig. 2. 7. 7, with the switch i n position I ,
suppose that . 2 , . .

and
Find the maximum current in the circuit for

5. In the circuit of Fig. 2. 7. 7, with the switch in position I ,
suppose that

.. . 2. 2,
and Find
6. I n the circuit of Fig. 2. 7. 7, with the switch i n position I ,
take . I , . and cos 4O sin 60t .
(a) Substitute , A cos B sin 60t and then
determine A and B to fnd the steady-state current ,in
the circuit. (b) Write the solution in the form ,
cos(wt - a) .
..............
.............,......
................
.,........, ..,...
.,
.,
.
...., ,...,.. ..
,
1 80 Chapter 2 Li near Equati ons of Hi gher Order
R
FIGURE 2.7.8. The circuit for
Problems through
7. (a) Find the charge , and current in the .circuit
if = (a constant voltage supplied by a battery) and
the switch is closed at time so that , (b)
Show that
lim , and that lim

-
-

-
-
8. Suppose that in the circuit of Fig. we have .
, and

(volts). (a)
Find ,and (b) What is the maximum charge on
the capacitor for and when does it occur?
9. Suppose that in the circuit of Fig. .
Z

, and cos
(a) Find , and (b) What is the amplitude of the
steady-state current?
10. An emf of voltage cos .is applied to the .
circuit of Fig. at time (with the switch closed),
and , Substitute ,, A cos .B sin .in
the diferential equation to show that the steady periodic
charge on the capacitor is

,, cos .- f)
.I .

where f tan-
I
..
.......,........
..,... ......
, A cos .B sin .
.,......,,,...........
,.....,

sin.- ) .
11. . . . . F; si n V
12. . .. . F;
sin t V
13. . .. .* J F;
cos V
14. . .. . F;
cos .sin V
15. . .. . ?
F;
sin V
16. . .. H, ?

F;
cos V
.. ............,.
..................
......,..........,..
..
17. . ..- . F;
V; ,
18. . .. . F;

V; ,*
19. . .. . F;

V; * , I
........,......,
..., .......- Isp (t) Itr (t).
20. The circuit and input voltage of Problem I I with
and ,
21. The circuit and input voltage of Problem .with
and ,
22. The circuit and input voltage of Problem with
and ,
23. Consider an .circuit-that is, an ..circuit with .
O-with input voltage q sin . Show that un
bounded oscillations of current occur for a certain reso
nance frequency; express this frequency in ters of .and

24. I t was stated in the text that, i f . . and are positive,
then any solution of . . is a transient
solution-it approaches zero as Prove this.
25. Prove that the amplitude ,of the steady periodic solution
of Eq. is maximal at frequency . I/...
Enp(int Problem and Egnvalues
Youarenowfaml|larwlththefactthataso|utlonofasecond-order|lneardlerentla|
equatlon ls unlque|y determlned by two lnltla| condltlons. In partlcu|ar, the on|y
so|utlonofthelnltla|va|ueprob|em
,

+p,x; ,+, ,x; ,= 0, ,,.; = 0, , ,.;~ 0 ( I )


l s thetrlvla|so|utlon,,x; = 0. MostofChapter2hasbeenbased,dlrect|yorlndl-
rect|y,ontheunlquenessofso|utlonsof|lnearlnltla|va|ueprob|ems(asguaranteed
byTheorem2ofSectlon2. 2).
Exampl e 1
FIGUR 2.8. 1. Various possible
solutions y (x) B sin x. of the
endpoint value problem in
Example I . For no B / does the
solution hit the target value y
for x .
Exampl e Z
X
FIGUR 2.8.2. Various possible
solutions y (x) B sin 2x of the
endoint value problem in Example
2. No matter what the coefcient
B is, the solution automatically
hits the target value y for
A .
2. 8 Endpoi nt Probl ems and Ei genval ues 1 81
Inthlssectlonwewl||seethatthesltuatlonlsradlca||ydlerentforaprob|em
suchas
,

+(x) ,
'
+q (x) ,= 0,
,(.) = 0, ,(b)= 0.
(2)
Thedlfferencebetweentheprob|emslnEqs. ( I ) and (2)lsthatln(2)thetwocon-
dltlonsarelmposedattwoa_ eepolnts.andbwlth(say). < b. In(2)weare
tonndaso|utlonofthedlfferentla|equatlonon thelnterva| ,.,b) thatsatlsnesthe
condltlons,(.) = 0and ,(b) = 0atthe endpolntsofthe lnterva|. Suchaprob|em
lsca||edanendpoint orboundary value problem. Examp|es I and2l||ustratethe
sortsofcomp|lcatlonsthat can arlselnendpolntprob|ems.
Consldertheendpolntprob|em
,

+3, = 0, ,(0) = 0, ,() = 0 (3)


Thegenera|so|utlonofthedlfferentla|equatlonls
,(x) = Acos x.+Bsln x..
Now,(0) = A, sothecondltlon,(0) = 0lmp|lesthatA = 0. Thereforetheon|y
posslb|eso|utlonsareoftheform,(x) = Bsln x.. Butthen
,() = Bsln . -0. 745SB,
s otheothercondltlon,() = 0requlresthatB = 0 a|so. Graphlca||y, Flg. 2. S. I
l||ustratesthefactthatoposslb|eso|utlon,(x) = Bsln x.wlth B = 0hltsthe
deslred target va|ue , = 0 when x = Thus the o/,so|utlon ofthe endpolnt
va|ueprob|emln(3)lsthetrlvla|so|utlon,(x) = 0(whlchprobab|ylsnosurprlse).
.. .... ............. . ... ........
Consldertheendpolntprob|em
,

+4, = 0, ,(0) = 0, ,() = 0


Thegenera| so|utlonofthedlerentla|equatlonls
,(x) = A cos 2x+Bsln 2x.

(4)
Agaln,,(0) = A, sothecondltlon,(0) = 0lmp|lesthatA = 0. Thereforetheon|y
posslb|eso|utlonsareoftheform,(x) = Bsln2x. Butnow,() = Bsln2 * 0
no matterwhatthe va|ue ofthecoefnclent B l s. Hence, as l||ustrated graphlca||y
lnFlg.2. S. 2, e.eposslb|eso|utlon,(x) = Bsln2x hlts automatlca||ythedeslred
targetva|ue , = 0 whenx = (whatevertheva|ueofB) . Thustheendpolntva|ue
prob|emln (4) has /]/e/,-.,dlerent nontrlvla| so|utlons. Perhaps thlsdoes
seemabltsurprlslng.
Remark 1 : Notethattheblg dlerencelntheresu|tsofExamp|es I and2
stemsfromtheseemlng|ysma||dlfferencebetweenthedlerentla|equatlonsln(3)
and(4), wlththecoefnclent3 lnonerep|acedbythecoemclent4lntheother In
mathematlcsase|sewhere,sometlmesblgdoors tumonsma||hlnges.
Remark 2: Theshootlngtermlno|ogy used lnExamp|es I and2lsof-
ten usefu| ln dlscusslngendpolntva|ueprob|ems. We conslderaposslb|eso|utlon
whlchstartsatthe|eftendpolntva|ueandaskwhetherlthltsthetargetspeclned
bytherlghtendpolntva|ue.
1 82 Chapter 2 Li near Equati ons of Hi gher Order
Exampl e
Eigenvalue Problems
Rather than belng the exceptlona| cases, Examp|es I and 2 l||ustrate thetyplca|
sltuatlonforanendpolntprob|emasln(2) . Itmayhaveno nontrlvla| so|utlons,or
ltmayhavelnnnlte|ymany nontrlvla| so|utlons. Notethattheprob|emsln(3)and
(4)bothcanbewrlttenlntheform
,+p(x) ,'
+q (x) ,=0, ,(.) =0, ,(b) =0, (5)
wlth p(x) = 0, q (x) = I , . = 0, andb = B Thenumber ls a parameterln
theprob|em(nothlngtodowlththeparametersthatwerevarledlnSectlon2. 5) It
we take =3, we get the equatlonsln (3), wlth =4, we obtalnthe equatlonsln
(4). Examp|es I and2 show thatthesltuatlonlnanendpolntprob|emcontalnlnga
parametercan (and genera||ywl||)dependstrong|yonthe speclncnumerlca|va|ue
oftheparameter.
Anendpolntva|ueprob|em such asthe prob|em ln(5)onethatcontalns
unspeclnedparameterlsca||edaneigenvalue problem. Thequestlonweaskln
anelgenva|ueprob|emls thls. Forwhatva|ues oftheparameter does thereexlst
a nontrlvla| (l. e. , nonzero) so|utlon ofthe endpolnt va|ue prob|em! Such a va|ue
of ls ca||edan eigenvalue oftheprob|em. One mlght thlnk ofsuch ava|ueasa
properva|ueofforwhlchthereexlstproper(nonzero)so|utlonsoftheprob|em.
Indeed,theprenxe/,elsaGermanwordthat(lnsomecontexts)maybetrans|ated
astheEng|lshwordpope,soelgenva|uesaresometlmes ca||edpape../es(or
./...e/s/.../es;
ThuswesawlnExamp|e2that =4lsanelgenva|ueoftheendpolntprob|em
,+) =0, ,(0) =0, ,() =0,
whereasExamp|e I showsthat = 3 lsnotanelgenva|ueofthlsprob|em.
(6)
Suppose that, l s an elgenva|ue ofthe prob|emln (5) and that,, (x) ls a
nontrlvla|so|utlonoftheendpolntprob|emthatresu|tswhentheparameterln(5)
l srep|acedbythespeclncnumerlca|va|ue,, so
Thenweca||,,aneigenfunction assoclatedwlththeelgenva|ue,. Thuswesaw
lnExamp|e2that,, (x) =sln2x lsanelgenfunctlonassoclatedwlththeelgenva|ue
, =4, aslsanyconstantmu|tlp|eofsln2x.
Moregenera||y,notethattheprob|emln(5)ls /o-o,eeoslnthesensethat
any constantmu|tlp|eofan elgenfunctlon ls agaln an elgenfunctlonlndeed, one
assoclatedwlththe same elgenva|ue. Thatl s, lf, =,,(x) satlsnestheprob|emln
(5)wlth =,, then sodoesanyconstantmu|tlp|ec,, (x) . Itcan beproved(under
ml|d restrlctlonsonthecoefnclentfunctlonspandq) thatanytwoelgenfunctlons
assoclatedwlththesameelgenva|uemustbe|lnear|ydependent.
Determlnetheelgenva|uesandassoclatedelgenfunctlonsfortheendpolntprob|em
,+) =0, ,(0) =0,
,(L) =0 (L > 0) .
(7)
Sol uti on We mustconsldera||posslb|e(rea|)va|uesofposltlve,zero,andnegatlve.
If =0, thentheequatlonlsslmp|y,=0andltsgenera|so|utlonls
,(x) =Ax+B.
FIGUR 2.8.3. The hyperbolic
sine and cosine graphs.
y
FIGUR 2.8.4. The
.
f
'
(
.
nnx
elgen unctIOns ] x) Sll -
.
for n .

2. 8 Endpoi nt Probl ems and Ei genval ues 1 83


Thentheendpolntcondltlons,(0) =0 =,(L) lmmedlate|ylmp|ythatA =s=0,
sotheon|yso|utlonlnthlscaselsthetrlvla|functlon,(x) = 0. Therefore,=0ls
oanelgenva|ueoftheprob|emln(7) .
If < 0, |et us then wrlte = -o
:
(wltho > 0)to be speclnc. Thenthe
dlerentla|equatlontakestheform
andltsgenera|so|utlonls

:
0 y - o y= .
,(x) =., e

+.
:
e

=Acoshox + sslnhox,
where A = ., +.
:
and s = ., - .
:
(Reca||thatcoshox = ,e

+e

;}zand
thatslnhox =,e

- e

;}z ;Thecondltlon,(0) =0thenglves


,(0) =Acosh0+ sslnh0=A =0,
sothat,(x) =sslnhox. Butnowthesecondendpolntcondltlon,,(L)=0,glves
,(L) = sslnh oL = 0. Thls lmp|lesthat s = 0,becauseo = 0,and slnh x = 0
on|yforx = 0 (examlnethe graphs of, = slnh x and , = coshx ln Flg. z s
Thusthe on|y so|utlonoftheprob|emln(7)lnthecase < 0lsthetrlvla|so|utlon
, = 0,andwemaythereforeconc|udethattheprob|emhasonegatlveelgenva|ues.
Theon|yremalnlngposslbl|ltylsthat=o
:
> 0wltho > 0. Inthlscasethe
dlerentla|equatlonl s
wlthgenera|so|utlon

:
0 y + o , = .
,(x) =Acosox + ssln ox.
Thecondltlon ,(0) = 0 lmp|les that A = 0, so ,(x) = sslnox. The condltlon
,(L) =0thenglves
,(L) =ssln oL=0.
Can thls occurlfs = 0! Yes, buton|yprovldedthatoL ls a (posltlve) lntegra|
mu|tlp|eof-
oL=-, z-, -, -
thatls, l f
. . . ,
Thuswehavedlscoveredthattheprob|emln(7)hasan/]/ese,e.eofposltlve
elgenva|ues
,s
Wlths= I, theelgenfunctlonassoclatedwlththeelgenva|ue
-
ls

-x
y-
,x; = m
_
= i , z, ,. . . . (9)
Flgurez s :showsgraphsofthenrstsevera|oftheseelgenfunctlons. We seevls-
lb|yhow theendpolntcondltlons ,(0) = ,(L) = 0 serveto se|ectustthoseslne
functlonsthatstartaperlodatx =0andwlndupatx =Lpreclse|yattheendofa
ha|f-perlod.
1 84 Chapter 2 Li near Equati ons of Hi gher Order
Examp|e 3 l||ustrates thegenera| sltuatlon. Accordlng toa theorem whose
preclsestatementwewl||deferuntl| Sectlon i , undertheassumptlonthat, (x;~
0 onthelnterva|. /} , anyelgenva|ueoftheformln(5)has adlvergentlncreaslng
sequence
, <
:
< ,< <
-
<
. . .
- +c
ofelgenva|ues, eachwlth an assoclatedelgenfunctlon. Thlslsa|sotrueofthefol
|owlngmoregenera|typeofelgenva|ueprob|em,lnwhlchtheendpolntcondltlons
lnvo|veva|uesofthederlvatlve,aswe||asva|uesof,
,

+p,x; ,+, ,x; y= 0,


., ,,.; +.
:
, ,.;= 0, /, ,,/;+/
:
, ,/;=0,
, i ;
where., , .
:
,/, ,and/
:
areglvenconstants. Wlth., = I =/
:
and.
:
= 0=/, ,we
gettheprob|emofExamp|e4 (ln whlch(x) = 0and, ,x; = I , aslntheprevlocs
examp|e).
Exampl e 4
Determlnetheelgenva|uesandelgenfunctlonsoftheprob|em
Y
FIGUR 2.8.5. The
eigenfunctions
.
(2n - 1 ) nx
Y
n
(x) sm
2L
for n 1 , 2, 3, 4.
,+y= 0, ,,o;= 0, , ,i;= 0. , i i ;
Sol ution
Vlrtua||ythesameargumentasthatusedlnExamp|e3showsthattheon|yposslb|e
elgenva|uesareposltlve, sowetake = o
:
> 0 (o > 0)tobespeclnc. Thenthe
dlfferentla|equatlonls
wlthgenera|so|utlon
,,x; = Acosox+ Bsln ox.
Thecondltlon,,o;= 0lmmedlate|yglvesA = 0, so
,,x; = B slnox and , ,x;= Bocosox.
Thesecondendpolntcondltlon, , i; = 0nowglves
, , i; = Bo cos oL= 0.
Thlswl||ho|dwlth B = 0provldedthatoL lsanoddposltlvelntegra|mu|tlp|eot
-J2.
-
oL =
2

thatl s, lf
-
:
=
-
4L

3-
`
o
-
:
4L
'
(2n - i ;-
2
(2n - i ;
:
-
:
4L
. . . ,
Thusthenthelgenva|ue
-
andassoclatedelgenfunctlonoftheprob|emln, i i ; are
glvenby
(2n - i ;-x
and y-
,x;= sln
2L
( i 2;
for n = i , 2, 3, . . . . Flgure 2. . 5showsgraphs ofthe nrst severa| oftheseelgen
functlons. Wesee vlslb|yhowtheendpolntcondltlons,,o;= , ,i;= 0 serveto
se|ectustthose slne functlons thatstartaperlodatx = 0 but wlnd upatx = L
preclse|ylnthemldd|eofaha|f-perlod.
2. 8 Endpoi nt Probl ems and Ei genval ues 1 85
A genera| procedurefordetermlnlngthe elgenva|ues ofthe problem ln ( l 0)
can be out|lned as fo||ows. We nrst wrlte the genera| so|utlon otthe dlerentla|
equatlonlntheform
, = A,,(x,; + B)
:
(x,;
We wrlte ) (x, ; because,, and )
:
wl|| depend on , as lnExamples 3 and4, in
whlch
,,(x) = cos:x= cos x
_
and )
:
(x) = sln:x= sln x
_
.
Then welmposethetwoendpolntcondltlons,notlngthateach ls|lnearln)and,' ,
andhencea|so|lnearlnA and B. When weco||ectcoemclentsofA and B lnthe
resu|tlngpalrofequatlons,wethereforegetasystemoftheform
:,() A+f () B= 0,
:
:
,; ~+f () B= 0.
( l 3)
Nowlsanelgenva|uelfandon|ylf thesystemln( I 3)hasanontrlvla|solutlon(one
wlth A and B notboth zero). But such ahomogeneoussystem ofllnearequatlons
has anontrlvla| so|utlonlfand on|y lfthedetermlnantofltscoemclentsvanlshes.
We therefore conc|ude that theelgenval ues oftheprob|em ln ( I 0) are the (rea|)
so|utlonsoftheequatlon
n,;= :,,;/:
,;- :
:
,;
/
,;= 0. ( I 4)
Tol||ustrateEq. ( I 4) lnaconcreteprob|em,|et' srevlslttheelgenva|ueprob
|em of Examp|e 3. If > 0, then the dlfferentlal equatlon ," + ) = 0 has
the genera| so|utlon,(x) = Acos(

x) + Bsln(

x) . The endpolnt condltlons


,(0) = 0and,(L) = 0thenyle|dtheequatlons
,(0) = A

I + B 0 =0,
,(L) = A cos(

L) + B sln,

L) = 0
(ln the unknowns A and B) whlch correspond to the equatlons ln ( l 3). The de-
termlnant equatlon n,; = 0 correspondlng to ( I 4) ls then slmp|y the equatlon
sln(

L) = 0, whlchlmp|lesthat

L =n,or=
:
-
:
,i
:
for= l , 2, 3, . . .
(aswesawlnExamp|e3) .
Formoregeneralprob|ems, theso|utlonof theequatlonn,;=0l n( I 4)may
presentformldab|edlfncu|tlesandrequlreanumerlcalapproxlmatlonmethod(scch
asNewton' smethod)orrecoursetoacomputera|gebrasystem.
Mostofthelnterestlnelgenva|ueprob|emslsduetothelrverydlversephysl-
ca|app|lcatlons. Theremalnderofthlssectlonlsdevotedtothreesuchapp|lcatlons.
Numerous addltlona| app|lcatlonsarelnc|udedlnChapters and9 (onpartlaldlf-
ferentla|equatlonsandboundaryva|ueprob|ems).
The Wirling String
Whoofushasnotwonderedaboutthe shapeofaqulck|ysplnnlngumprope`Let
usconsldertheshapeassumedbyatlght|ystretchedHexlb|estrlngof|engthL and
constant|lneardensltyp (massperunlt|ength)lfltlsrotatedorwhlr|ed(l lkeaump
rope) wlthconstantangu|ar speed .(ln radlans persecond)aroundltsequl|lbrlum
posltlona|ongthex-axls. Weassumethatthe portlonofthe strlngto one sldeof
anypolntexertsaconstanttenslonforce ontheportlon ofthe strlngto theother
1 86 Chapter 2 Li near Equati ons of Hi gher Order
:
x = O
st;|a,
La|||o;|ames|t|ea

(x
. y (x))
0
wa|;||a,st;|a,
( o)
'-J
x = L
x
FIGURE 2.8.6. The whirling
string.
:
Z
Z
I
x x x Lx
x
FIGURE 2.8.7. Forces on a short
segment of the whirling string.
sldeofthepolnt,wlththedlrectlonofI tangentla|tothestrlngat thatpolnt. We
further assumethat, as the strlng whlr| s around thex-axls, eachpolntmoves ln a
clrc|ecenteredatthatpolnt' sequl|lbrlumposltlononthex-axls. Thus thestrlngls
e|astlc, so that as ltwhlr|s lta|sostretches toassume a curved shape. Denoteby
,(x)thedlsp|acementofthestrlngfromthepolntxontheaxlsofrotatlon. Flna||y,
weassumethatthedeectlonofthestrlnglssos|lghtthatsln0 tan0 = ,'(x) ln
Flg. 2. S. 6(c) .
Wep|an to derlve adlfferentla|equatlonfor,(x) byapp|lcatlonofNewton' s
|aw F = mO tothepleceof strlng ofmass jx correspondlng tothe lnterva|
x,x + x] . The on|y forces actlng on thls pleceare thetenslonforces at itstwo
ends. FromFlg.2. . 7weseethatthenetvertlca|forcelntheposltlve,-dlrectlonls
F = I sl n(0 + 0) - Isln 0 Itan(0 + 0) Itan 0,
s othat
F I,
'
(x+ x)- I,
'
(x) . ( I 5)
Next wereca|| from e|ementary ca|cu|us orphyslcs the formu|a O = .forthe
(lnward)centrlpeta|acce|eratlonofabodylnunlformclrcu|armotlon,istheradlus
ofthe clrc|e and o ls the angu|ar ve|oclty ofthe body). Here wehave = ,, so
thevertlca|acce|eratlonofourpleceofstrlnglsO =
.
o
:
,, themlnusslgnbecause
thelnwarddlrectlonlsthenegatlve,-dlrectlon. Becausem = jx,substltutlonof
thlsand( I 5)lnF = mO yle|ds
sothat
I,' (x+ x)- I,' (x)
.
jo
:
,x,
,' (x+ x) ,' (x)
:
I -jo ,.
x
We nowtakethe|lmltas x -- 0togetthedlfferentla|equatlonofmotlonofthe
strlng.
Ifwewrlte
I
:
0 ) + jo ) =

( I 6)
( I 7)
and lmpose the condltlon that the ends ofthe strlng are nxed, we nna||y get the
elgenva|ueprob|em
,+, = 0, ,(0) = 0, ,(L) = 0 (7)
thatweconslderedlnExamp|e3. Wefoundtherethattheelgenva|uesoftheprob|em
ln(7) are
()
wlththeelgenfunctlon), (x) = sln(nxJL) assoclatedwlth, .
Butwhat does a||thls meanlnterms ofthewhlr|lngstrlng!Itmeans thatun-
|ess ln( I 7)lsoneoftheelgenva|uesln(), thentheon|yso|utlonoftheprob|em
FIGUR 2.8.8. Distortion of a
horizontal beam.


= = = " "
"

es|t|ve
:
-va|aes
L
FIGUR 2.8.9. The defection
curve.
X
2
.
8
Endpoi nt Probl ems and Ei genval ues 1 87
ln(7)ls thetrlvla|so|utlon,(x) = 0. Inthls casethe strlng remalnslnltsequl|lb-
rlumposltlonwlth zerodeectlon. Butlfweequate( I 7) and()andso|veforthe
va|ue
-, correspondlngto, ,
( I )
for= I , 2, 3, . . . , wegetasequenceofcritical speeds ofangu|arrotatlon. On|y
atthesecrltlca|angu|arspeedscanthestrlngwhlr|upoutofltsequl|lbrlumposltlon.
Atangu|arspeed6 ltassumesashapeoftheform ), = c,sln(nxJL) l||ustrated
ln Flg. 2. S. 4 (wherec
, = I ) . Our mathematlca|mode| ls not sufnclent|ycomp|ete
(orrea|lstlc)todetermlnethecoefnclentc
, ,butltassumesmuch sma||erdeHectlons
thanthoseobservedlnFlg. 2. S. 4,sothenumerlca|va|ueofc, wou|dnecessarl|ybe
slgnlncant|ysma||erthan I .
Supposethatwestartthestrlngrotatlngatspeed
thengradua||ylncreaseltsspeedofrotatlon. So|ongas6 < 6j , thestrlngremalns
ln lts undeectedposltlon, = 0. But when6 =6j , thestrlngpopslntoawhlr|lng
posltlon , = c, sln(xJL) . Andwhen 6 lslncreased stl|| further, the strlngpops
backlnto ltsundeHectedposltlona|ongtheaxlsofrotatlon'
The Defection of a Uniform Beam
Welnc|udenowanexamp|eoftheuseofare|atlve|yslmp|eendpolntva|ueprob|em
toexp|alnacomp|lcatedphyslca|phenomenontheshapeofahorlzonta|beamon
whlchavertlca|forcels actlng.
Conslderthehorlzonta|beamshownlnFlg. 2. S. S, unlformbothlncrosssec-
tlon and ln materla|. Ifltls supported on|y at lts ends, then the force ofltsown
welghtdlstortslts|ongltudlna| axls ofsymmetry lntothecurve shownasadashed
|lne ln the ngure. We want to lnvestlgate the shape , = ,(x) ofthls curve, the
defection curve of the beam. We wl|| use the coordlnate system lndlcated ln
Flg.2. . 9, wlththeposltlve,-axlsdlrecteddownward.
Aconsequenceofthetheoryofe|astlcltylsthatforre|atlve|ysma||deHectlons
ofsuch a beam (so sma|| that )' (x) ]
:
ls neg|lglb|elncomparlson wlthunlty), an
adequatemathematlca|mode|ofthedeHectloncurvelsthefourth-orderdlerentlal
equatlon
iiy
-
= F(x) , ( I 9)
where

ils aconstantknown asthero, s-oa/softhematerla|ofthebeam,

i denotes the moment oflnertla ofthe cross sectlon ofthe beam around a
horlzonta||lnethroughthecentroldofthecrosssectlon,and

F(x) denotesthedensltyofao.aforceactlngvertlca||yonthebeamat
thepolnt x.
1 88 Chapter 2 Li near Equati ons of Hi gher Order
x = o x = L
s| m|ysae;teee;a|a,ee
x = O x = L

Da||t |a
FIGUR 2.8.10. Two ways of
supporng a beam.
nes/-offorce! Yes, thls meansthattheforce actlngdownward onavery
short segment x,x + x] of the beam l sapproxlmate|y F(x) x. The unlts ot
F(x) arethoseofforceperunlt|ength, suchaspounds per foot. We wl||conslder
herethecaseln whlchtheon|yforcedlstrlbuteda|ongthebeamlsltsownwelght,
U poundsperfoot,sothatF(x) = U. ThenEq.( I 9)takestheform
iiy
-
= U (20)
wherei, i , andU area||constant.
Note: Weassumenoprevlousfaml|larltywlththetheoryofe|astlcltyorwlth
Eq. ( I 9)or(20)here. Itls lmportanttobeab|etobegln wlthadlerentla| equatlon
that arlses ln a speclnc app|led dlsclp|lne and then ana|yze lts lmp|lcatlons, thus
we deve|op anunderstandlngoftheequatlonby examlnlnglts so|utlons. Observe
that, ln essence, Eq. (20) lmp|les that thefourthderlvatlve y
-
ls proportlona| to
the welght denslty U. Thls proportlona|lty lnvo|ves, however, moconstants. i,
whlch depends on|yonthematerla|ofthebeam, and i , whlchdependson|yonthe
shapeofthecrosssectlonofthebeam. Va|uesoftheYoung' smodu|usiofvarlous
materla|scanbefoundlnhandbooksofphyslca|constants, i= -.
-
foraclrcu|ar
crosssectlonofradlus.
A|thoughEq. (20)ls afourth-orderdlerentla|equatlon,lts so|utlonlnvo|ves
on|y theso|utlonofslmp|enrst-orderequatlonsby successlveslmp|elntegratlons
OnelntegratlonofEq. (20)yle|ds
asecondyle|ds
anotheryle|ds
iiy= Ux
3
+ t C, x
2
+C
2
X +C3 ;
anna|lntegratlonglves
where C" C
2
, C3 , and C
+
arearbltrary constants. Thus weobtaln a so|utlon ot
Eq. (20)ofthe form
(2I )
where~, s, C, and bareconstantsresu|tlngfromthefourlntegratlons.
These|astfourconstantsaredetermlnedbythewaylnwhlchthebeamlssup
portedatltsends,wherex = 0andx = L. Flgure2. . I 0showstwocommontypes
ofsupport. Abeammlghta|sobesupportedonewayatoneendbutanotherwayat
the otherend. Forlnstance,Flg.2. . I I showsacantilever-a beamnrm|yfastened
atx = 0but)ee(no supportwhatsoever)atx = L. Thefo||owlngtab|e showsthe
boundary orendpoint conditions correspondlngtothethreemostcommoncases.
Wewl||seethatthesecondltlonsareapp|ledreadl|ylnbeamprob|ems,a|thougha
dlscusslonhereofthelrorlglnwou|dtakeustoofarane|d.
2. 8 Endpoi nt Probl ems and Ei genval ues 1 89
Endpoint Condition
Simply supported y = y
"
=
Built-in or fxed end y = y
'
=
Free end y
"
= y
(3)
=
For examp|e, the deHectlon curve ofthe cantl|ever lnFlg. 2. S. I I wou|d be
glvenbyEq. (2I ),wlththecoemclentsA, B, C, and ndetermlnedbythecondltlons
,(0) =,
'
(0) = and ,"( L) =,

(L) =, (22)
correspondlngtothenxedendatx =andtheneeendatx = L. Thecondltlons
FIGUR 2.S. 11. The cantilever. ln(22)togetherwlththedlerentla| equatlonln(2I ) constltutean endpoint value
problem.
Exampl e DetermlnetheshapeofthedeHectloncurveofaunlformhorlzonta|beamof|ength
L andwelghtuperunlt|engthandslmp|ysupportedateachend.
Sol ution Wehavetheendpolntcondltlons
,(0) =," (0) ==,(L) =,"( L) .
Rather than lmposlng then dlrect|y onEq. (2I ), |et us begln wlth the dlerentlal
equatlon EI,
+
= u and determlne the constants as we proceed wlth the four
successlvelntegratlons. The nrsttwo lntegratlonsyle|d
EI,

=ux+ A, EI,"= 1 ux
:
+Ax+B.
Hence," (0) =lmp|lesthatB =, andthen,"(L) =glves
= 1 wL
2
+AL.
Itfo||owsthatA =-uLJ2 andthusthat
EI,"= 1x
2
- 1 wLx.
Thentwomorelntegratlonsglve
andnna||y,
EI,(x) = _ux
+
- uLx

+Cx+n
Now,(0) =lmp|lesthatn=, then,because,(L) =,
I t fo||owsthatC =uL

J4. HencefromEq.(23)weobtaln
,(x) =

(x
+
- 2Lx

+L

x)
24EI
(23)
(24)
1 90 Chapter 2 Li near Equati ons of Hi gher Order
cXump| e
asthe shapeoftheslmp|ysupportedbeam. Itls apparentfromsymmetry(seea|so
Prob|em I 7) thatthe-.x/--ae)e./o)max
ofthebeam occurs at ltsmldpolnt
x=LJ2,andthushastheva|ue
thatl s,
L

W
+
2
+
I
+

)max
= )

=
24Ei
L _L + _L ,
5uL
+
)max
=
3S4Ei

(25)

Forlnstance,supposethatwewanttoca|cu|atethemaxlmumdeectlonofaslmp|y
supportedstee|rod20longwlthaclrcu|arcrosssectlon I ln. lndlameter. Froma
handbookwenndthattyplca|stee|hasdenslty=7. 75gJcm

andthatltsYoung's
modu|us l s E = 2 Z I 0
| :
gJcm S
:
, so ltwl||be more convenlenttowork ln cgs
unlts. Thusourrodhas
and
|ength. L = (20) 30 4S
c
,=609. 60cm
radlus. .=

ln.

2. 54

= I 27cm.
Its//e.massdenslty(thatls, ltsmassperunlt|ength)ls
so
W = p = 39 27,9S0
c
, 3S4S4 6
dyn
.
cm s cm
The areamomentoflnertlaofaclrcu|ardlskofradlus. around adlameterls i =
.
-
,so
ThereforeEq. (25) yle|ds

(5) (3S4S4. 6) (609. 60)


+

I 6. 96cm )max
(3S4) (2 I 0
| :
) (2. 04)
7
about6. 6Sln. , asthemaxlmumdeectlonoftherodatltsmldpolnt. Itlslnterestlng
tonotethat
)max
ls proportlona| to L
+
, solftherodwere only I 0ft|ong,ltsmaxl-
mumdeectlonwou|dbeon|yone-slxteenthasmuchon|yabout0 42ln. Because
i = .
-
, weseefrom Eq. (25) thatthe samereductlonln maxlmum deectlon
cou|dbeachlevedbydoub|lngtheradlus.oftherod.
y
y - y,x,
The Bucked Rod
2.8 Endpoi nt Probl ems and Ei genval ues 1 91
Flgure z s i z shows aunlform rodof|ength L, hlnged ateach end, thathas been
'buck|ed by an axla| force ofcompresslon i app|led at one end We assume
" "
thls buck|lng to be so s|lght that the deectlon curve y = y(x) ofthe rod may
x
beregardedasdennedonthelnterva|cx L
x = c

x - L
FIGURE 2.8. 12. The buckled
rod.
lnthetheoryofe|astlcltythe|lnearendpolntboundaryva|ueprob|em
iiy

+iy=c, y,c;=y,i; =c ,z:;


lsusedtomode|theactua|(non|lnear)behavloroftherod Aslnourdlscusslonot
thedeectlonofa unlform beam, idenotes the Young' smodu|us ofthematerlal
ofthe beam and idenotesthemomentoflnertlaoteachcrosssectlonotthebeam
around ahorlzonta||lnethroughltscentrold
If wewrlte
i
= -
ii

then theprob|emln,z:;becomestheelgenva|ueprob|em
y

+ y=c,
y,c;=y,i;=c
,z:;
,:;
thatweconslderedlnExamp|eWefoundthatltselgenva|ues ,
-
|areglvenby
,s;
wlththeelgenfunctlon
y-
=sln(nxJL) assoclatedwlth
-
(Thuswhlr|lngstrlngs
andbuck|edrods|eadtothesameelgenva|uesandelgenfunctlons )
To lnterpretthlsresu|tln terms ofthebuck|edrod, reca|| nom Eq ,z:;that
i=ii Theforces
= i , z, , ,zs;
arethe.//..//.///,)o.esofthe rod On|y whenthecompresslveforce ils
oneofthesecrltlca|forcesshou|dtherod'buck|eoutofltsstralght(undeected)
shape Thesma||estcompresslveforceforwhlchthlsoccursls
,z;
Thls sma||estcrltlca|force i, lsca||edthei/e/.///,)o.efortherod,ltlsthe
upperboundforthosecompresslveforces towhlchtherodcan safe|y besubected
wlthoutbuck|lng (Inpractlcearodmayfal|ataslgnlncant|ysma||erforceduetoa
contrlbutlonoffactorsnottaken lnto accountbythemathematlca|mode|dlscussed
here )
1 92 Chapter 2 Li near Equati ons of Hi gher Order
Exampl e Forlnstance, supposethatwe wanttocomputetheEu|erbuck|lngforceforastee|
rod ft |ong havlngaclrcu|arcrosssectlon ln. lndlameter. Incgs unltswehave
L = )

= cm, and
i= ln. )

cm
+
.
Upon substltutlngtheseva|ueslnEq. we nndthatthecrltlca|forceforthlsoa
ls
i, Z dyn lb,
uslngtheconverslonfactor Z

dynJ|b.
Problems
....................
..................
,....................
1.
2.
3.
4.
5.
6. Consider the eigenvalue problem



All the eigenvalues are nonnegative, so write a
2
where a (a) Show that = is not an eigen
value. (b) Show that = A cos ax B sin ax satisfes
the endpoint conditions if and only if B and a is
a positive root of the equation tan , l/z . These roots
{a
n
|are the abscissas of the points of intersection of the
curves tan , and ,, as indicated in Fig.
Thus the eigenvalues and eigenfunctions of this problem
are the numbers {a l \ and the functions { cos a
n
xl j, re
spectively.
`
FIGURE 2.8.13. The eigenvalues are determined by
the intersections of the graphs of = tan , and ,
(Problem
Consider the eigenvalue problem

=
all its eigenvalues are nonnegative. (a) Show that U
is not an eigenvalue. (b) Show that the eigenfunctions
are the functions { sin a
n
xl j, where a
n
is the nth positive
root of the equation tan , , (c) Draw a sketch indi
cating the roots {a
n
|as the points of intersection of the
curves tan , and ,. Deduce from this sketch
that a
n . when .is large.
8. Consider the eigenvalue problem

all its eigenvalues are nonnegative. (a) Show that U
is an eigenvalue with associated eigenfunction . A.
(b) Show that the remaining eigenfunctions are given by

.sin fnx, where fn is the nth positive root of the


equation tan , ,. Draw a sketch showing these roots.
Deduce from this sketch that fn
. when .is
large.
9. Prove that the eigenvalue problem of Example .has no
negative eigenvalues.
10. Prove that the eigenvalue problem


has no negative eigenvalues. ..... Show graph
ically that the only root of the equation tanh , , is
,
11. Use a method similar to that suggested in Problem to
show that the eigenvalue problem in Problem has no neg
ative eigenvalues.
12. Consider the eigenvalue problem

which i s not of the type i n because the two endpoint
conditions are not "separated" between the two endpoints.
(a) Show that

= is an eigenvalue with associated


eigenfunction ;,.= (b) Show that there are no neg
ative eigenvalues. (c) Show that the nth positive eigen
value is .

and that it has two linearly independent associ


ated eigenfunctions, cos ..and sin ..
13. Consider the eigenvalue problem
y" 2y' ; = 0; y) = . =
(a) Show that = is not an eigenvalue. (b) Show
that there is no eigenvalue such that (c) Show
that the nth positive eigenvalue is , = .

with
associated eigenfunction ;,.= c
-\
sin ..
14. Consider the eigenvalue problem
y" 2y' ; = 0; y) = 0, y' ( I ) =
Show that the eigenvalues are al l positive and that the nth
positive eigenvalue is , = awith associated eigen
function ;,.= c
-\
sin a
n
x, where a,is the nth positive
root of tan , = ,
15. (a) A uniform cantilever beam is fxed at .= 0 and free
at its other end, where .= . Show that its shape is given
by
.
.. = .

...


241
(b) Show that y' .= 0 only at .= 0, and thus that i t fol
lows (why?) that the maximum defection of the cantilever
is ;-.
= ..= ..

/( 8 I ) .
2. 8 Endpoi nt Probl ems and Eigenval ues 1 93
16. (a) Suppose that a beam is fxed at its ends .= 0 and
.= . Show that its shape is given by
.
.. = .

..


241
(b) Show that the roots of . . = 0 are .= U, .= .
and .= . so it follows (why?) that the maximum
defection of the beam is
. ..

;-.
= ;
_
=
3841
'
one-ffh that of a beam with simply supported ends.
17. For the simply supported beam whose defection curve is
given by Eq. (24), show that the only root of . . = U in
[0, .,is .= .so it follows (why?) that the maximum
defection is indeed that given in Eq. (25).
18. (a) A beam is fxed at i ts lef end .= 0 but is simply sup
ported at the other end .= .Show that its defection
curve i s
(b) Show that its maximum defection occurs where .=
5
_
,
. and is about . of the maximum de
fection that would occur if the beam were simply sup
ported at each end.
1 94
Po"er Series
Methods

n Sectlon 2. 3 we saw that so|vlng a homogeneous |lnear dlerentla| equatlon


wlth constantcoemclentscanbereducedtothe a|gebralc prob|emofnndlngthe
rootsoflts characterlstlcequatlon. Therels noslml|arprocedureforso|vlng|lnear
dlerentla|equatlonswlth .././/ecoemclents,at|eastnotroutlne|yandlnnnlte|y
manysteps. Wlththeexceptlonofspecla|types,suchastheoccaslona|equatlonthat
can be so|ved by lnspectlon, |lnearequatlons wlth varlab|e coemclents genera||y
requlrethepowerserlestechnlquesofthlschapter.
Thesetechnlquessumceformanyofthenone|ementarydlerentla|equatlons
thatappearmostnequent|ylnapp|lcatlons. Perhapsthemostlmportant(becauseot
ltsapp|lcatlonslnsuchareasasacoustlcs, heatow,ande|ectromagnetlcradlatlon)
lsBessel 's equation oforder
Legendre's equation oforderl s lmportantlnmanyapp|lcatlons. Ithastheform
( I - x
:
) , - 2x,'+ ,+ I ) ,= 0.
Inthlssectlonwelntroducethepower series method lnltsslmp|estformand,
a|ongtheway,state(wlthoutproof)severa|theoremsthatconstltutearevlewofthe
baslcfactsaboutpowerserles. Reca|| nrstthatapower series ln(powersof)x
ls anlnnnlteserlesoftheform

- . =c,+c, (x - .;+c
:
(x - a)
:
+
. . .
+c, (x - .+
. . .
. ( I )

If.= 0, thlsls apowerserleslnx.

= c,+c, x +c
:
x
:
+
. . .
+

(2)
3. 1 I ntroducti on and Revi ew of Power Seri es 1 95
Wewl||connneourrevlew maln|yto power serlesln buteverygenera|property
ofpowerserlesln canbeconvertedtoagenera| propertyofpowerserles ln .
byrep|acementof wlth .

Thepowerserlesln(2)converges onthelnterva| iprovldedthatthe|lmlt
(3)
exlstsfora|| lni Inthlscasethesum


(4)

l sdennedon i,andweca||theserles_ -


apower series representation ofthe
functlon oni Thefo||owlngpowerserlesrepresentatlonsofe|ementaryfunctlons
shou|dbefaml|lartoyounomlntroductoryca|cu|us.
and

, ,


cos x I -
.
(2n) ' 2' 4'

, ,


.-


(2n I ) ' 3 ' 5'




cosh x I


(2n) ' 2' 4'


slnh


(2n I ) ' 3 ' 5'




ln( I


n



I




.

(5)
(6)
(7)
(S)
(9)
( I 0)
( I I )
D
-- ,

-- , . - :


( I

I -
2'

3 '
+
. .
. ( I 2)
Incompactsummatlonnotatlon, weobservetheusua|conventlonsthat0' = I and
that

I for a|| lnc|udlng 0. The serles ln (5)through (9) converge to
thelndlcatedfunctlonsfora||

Incontrast, the serlesln ( I 0) and ( I I ) convergelf
< Ibutdlvergelf > I . (Whatlf I !) Theserlesln( I I ) ls thegeometric
series. The serlesln ( I 2) , wlth- an arbltrary rea|number, ls thebinomial series.
If- ls anonnegatlvelntegern, then theserlesln ( I 2) termlnates andtheblnomla|
serlesreducesto apo|ynomla| ofdegreen whlchconvergesfora||

Otherwlse,
theserleslsactua||ylnnnlteandltconvergeslf < I anddlvergeslf > I , lts
behavlorfor I dependsontheva|ueof-

1 96 Chapter 3 Power Seri es Methods
Remark: Power serles such as those |lsted ln (5) through ( I 2) are ottea
derlvedasTay|orserles. The Taylor series wlthcenter x =a otthe functlon]ls
thepowerserles

(a) ](a)

(x - a) =](a) +]'
(a) (x - a) + (x - a)
:
+

( I J)
,=a
n ' 2'
lnpowersofx - a, underthehypotheslsthat] lslnnnlte|ydlerentlab|eatx =
(sothatthecoemclentslnEq. ( I 3) area||denned). lfa =0, thentheserlesln( I J)
ls theMaclaurin series
( I J')
Forexamp|e, supposethat ](x) =e

Then ]

(x) =e

,andhence]

(0) = .
fora||n 0. InthlscaseEq. ( I 3')reducestotheexponential series ln(5).
Power Series Operations
lfthe Tay|or serles ofthe tunctlon ] converges to ](x) tora|| x ln some open
lnterva| contalnlng a, then we say that the functlon ] ls analytic atx = . For
examp|e,
everypo|ynomla|functlonls ana|ytlceverywhere,
everyratlona|functlonls ana|ytlcwhereverltsdenomlnatorls nonzero,

moregenera||y,lfthetwofunctlons] andg areboth ana|ytlcatx = then


so are thelr sum ] + g and thelr product ]
.
g, as ls thelr quotlent ]Jg lt
g(a) 0.
Forlnstance,thefunctlonh (x) =tanx = (sln x) J(cos x) ls ana|ytlc atx =
becausecos0 = . 0 and the slneand coslnefunctlons are ana|ytlc (by vlrtue
ofthelr convergent power serles representatlons ln Eqs. (6) and (7)). It ls rather
awkward to compute the Tay|or serles ofthe tangent functlon uslng Eq. ( I 3) be-
cause ofthe way ln whlch lts successlve derlvatlves grow ln comp|exlty (try lt ' ) .
Fortunate|y,powerserlesmaybemanlpu|ateda|gebralca||yl nmuchthesameway
aspo|ynomla|s. Forexamp|e,lf
then
and

](x) =_a,x
,
,
=a

and g(x) =b,x


,
,
,
=a

](x) +g(x) =_(a, +b, )x


,
,=a

](x) g(x) =c,x


,
,a
( I 4)
( I 5)
3. 1 I ntroducti on and Revi ew of Power Series 1 97
where.
-
= .,/
-
+., /
-
,+

+.
-
/,Theserlesln( I 5) lstheresultof termwise
addition andtheserlesln( I 6) lstheresultotformal multiplication-multiplying
each term otthe nrstserlesbyeachterm otthesecondandthencol|ectlng coet-
nclents otllke powers otx. (Thus the processes strongly resemble addltlon and
mu|tlpllcatlon ot ordlnary polynomla| s. ) The serles ln ( I 5)and ( I 6) converge to
](x) + g(x) and ](x) g(x) , respectlvely, on any open lnterval on whlch both the
serlesln ( I 4)converge. Forexamp|e,
.
I

I
:
I
+

smx cos x = x -

x +
I 20
x
. . .
I
_
x +
24
x -
. .
torallx.
=x +

+
I

+

6 2 24 I 2 I 20
4

I 6

= x - x + x
6 I 20
I (2x)

(2x)

I
.
=

(2x) + _ -
. . .
=

sm 2x
Slmllar|y,the quotlentottwopowerserlescan becomputedbylongdlvlslon,
as l|lustrated bythe computatlon shown lnFlg. 3. l . I . ThlsdlvlslonottheTaylor
serlestorcosx lntothattorslnx yleldsthenrsttewtermsottheserles
I

i: _
tan x =x+ x + x +x +

.
3 I 5 3 I 5
, i :;
Dlvlslonotpower serles ls more treacherous than multlpllcatlon, the serles thus
obtalnedtor ] Jg maytal|to converge atsomepolntswhere the serles tor] andg
bothconverge. Forexamp|e, the slneandcoslneserles converge tor allx, butthe
tangentserlesln( I 7)convergeson|ylt x < J2.
The Power Series Method
Thepower series method torso|vlngadlerentla|equatlonconslstsotsubstltutlng
thepowerserles
( I S)
lnthedltterentlal equatlon and then attemptlngto determlne what the coetncleats
.,, ., , .
:
, o must be ln order that the power serles wlll satlsty the dltterentlal
equatlon. Thls ls remlnlscentotthe method otundetermlnedcoemclents,butnow
we have lnnnlte|y many coetnclents somehow to determlne. Thls method ls not
always successtul, but when lt ls we obtaln an lnnnlte serles representatlon ota
so|utlon,lncontrasttothe'closedtorm solutlonsthatourprevlous methods have
ylelded.
Betorewecansubstltutethepowerserlesln( I S)lnadlerentlalequatlon,we
mustnrstknowwhattosubstltutetorthederlvatlves

,
o
, . . . . Thetollowlngthe-
orem(statedwlthoutproof)tel|susthatthederlvatlve

ot = _ .
-
x
-
lsobtalned
bytheslmp|eprocedureotwrltlngthesumotthederlvatlvesotthelndlvidualterms
lntheserles tor.
1 98 Chapter 3 Power Seri es Methods
x

2x

I 7x

x +
3
+
I 5
+
3 I 5
x
:
x
+
x
-
x

+ I -

+
24
-
720
+
.
x
6 I 20 5040
x

x - +
2 24 720
x

3 30
+
S40
x

3 6
+
72
2x

4x

I 5 3 I 5
2x

I 5 I 5
I 7x

3 I 5
FIGURE 3. 1. 1. Obtaining the series for tan A by division of series.
THEOREM 1 Termwlse Differentiati on of Power Series
Ifthepowerserlesrepresentatlon

](x) = c,x
,
= c,+c, x +c
:
x
:
+c

+
. . .
,=a
+

+

+

+

+

+

+

( I 9)
ofthefunctlon] convergesontheopenlnterva|I , then] lsdlerentlab|eoni,
and

(x) = nc,x
,

,
= c, +2c
:
x +3c

x
:
+
. . .
,=
,
ateachpolntofi
Forexample,dlerentlatlonofthegeometrlcserles
I

:
= x = l + x + x + x +

I - x
,=a
(20)
( I I )
Exampl e 1
glves
3. 1 I ntroducti on and Revi ew of Power Seri es 1 99

= ++

+
. . .
.

The process of determlnlng the coefnclents ln the serles = _

so
that ltwl|| satlsfy a glven dlerentla| equatlon depends a|soonTheorem

Thls
theorema|sostated wlthoutproofte||sus thatlftwopowerserlesrepresentthe
samefunctlon,thentheyarethe same serles. Inpartlcu|ar, theTay|orserles ln
l stheon|ypowerserles(lnpowersof thatrepresentsthefunctlon].
THEOREM Z I dentity Pri nci pl e
If
O O


foreverypolntx lnsomeopenlnterva|i , then

fora||
Inpartlcu|ar, lf_

= 0fora||lnsomeopenlnterva|, ltfo||owsfrom
Theoremthat

=0fora||0,
So|vetheequatlon

+ =0.
Sol uti on We substltutetheserles
andobtaln
O
and

O O

=0.


Tocomparecoemclents here, weneed thegenera|termlneach sum tobetheterm
contalnlng

Toaccomp|lshthl s, we shlftthelndexofsummatlonlnthenrstsum.
Toseehowtodothls, notethat
O O

= +

+
= +


Thus wecan rep|acewlth+ lf, at the same tlme, westartcountlngonestep
|ower, thatls, at= 0 rather than at=

Thls ls a shlftof+ln thelndex of


summatlon. Theresu|tofmaklngthlsshlftln Eq. ls theldentlty
thatls,
O O
+

=0,

O
+

=0.

200 Chapter 3 Power Seri es Methods


Ifthls equatlonho|ds on some lnterva|, then ltfo||owsfrom the ldentlty prlnclple
that+

= 0fora||0, consequent|y,

= -
+ I

fora|| < 0 Equatlon ls a recurrence relation from whlchwecansucces-


slve|ycomputec, ,c
;
,c

,
. . lntermsof the|atterwl||tumouttobethearbltrary
constantthatweexpecttonnd lnagenera|so|utlonofanrst-orderdlerentla|equa
tlon.
Wlth= 0,Eq. glves
Wlth= I , Eq. glves
Wlth= Eq. glves
c
;

c, = "
I
c

= =
3
By nowltshou|dbec|earthataftersuchsteps, wewll|have

c = ( -I ) -

(Thlslseasytoprovebylnductlonon

Consequently,ourso|utlontakestheforn
O O

,(x) =

= ( -I ) -x = =


Inthenna|stepwehaveusedthefaml|larexponentla|serleslnEq. (5)toldentlfyour
powerserles so|utlonasthe sameso|utlon,(x) =

wecou|dhaveobtalned
bythemethodofseparatlonofvarlab|es.
Shift of Index of Summation
Intheso|utlonofExample I wewrote
O O

= +

by shlftlng the lndex ofsummatlon by +I ln theserles onthe |eft. Thatls, we


slmu|taneous|y thelndex of summatlon by I (rep|aclng wlth + I ,
+ I ) andthe startlngpolntbyI , from = I to= 0, thereby
obtalnlngtheserlesontherlght. Thlsprocedurels va|ldbecauseeachlnnnlteserles
ln(23)lsslmp|yacompactnotatlonfortheslng|eserles

3. 1 I ntroducti on and Revi ew of Power Seri es 201


Moregenera||y,wecanshlthelndex ofsummatlonby/lnanlnnnlteserles
byslmu|taneous|y/.e.s/,thesummatlonlndexby/,-- +/;and1e.e.s/,
thestartlngpolntby/Forlnstance,ashlby+z,-- +z;yle|ds
If/l snegatlve,welnterpretadecreaseby/asanlncreaseby -/= / Thusa
shlftby -z,-- - z;lnthelndexofsummatlonyle|ds
O O
.
-
x
-

=,- z;.
-:
x
-

,
-,
-
wehave1e.e.se1thelndexofsummatlonbyzbut/.e.se1thestartlngpolntby
z,from= to=3. Youshou|dcheckthatthesummatlonontherlghtlsmerely
anotherrepresentatlonoftheserlesln,z:;
We kow that the power serles obtalned ln Examp|e converges for all x
because lt ls an exponentlal serles. More common|y, a power serles so|utlon ls
not recognlzab|e ln terms ofthe faml|lar e|ementary functlons. When we get aa
unfaml|lar power serles so|utlon, we need a way of nndlng where lt converges.
Aftera||, , = _ .
-
x
-
lsmere|yan.ss-e1formoftheso|utlon. Theprocedure
l||ustratedlnExamp|e fornndlngthecoemclents , c
-
,ls mere|y aforma|process
and may or may not be va|ld. Its va|ldltyln app|ylng Theorem to compute
, and app|ylng Theorem z to obtaln a recurrence relatlon for the coemclents
depends on the convergence ofthe lnltla||y /oserles , = _ .
-
x
-
Hence
thlsforma|processlsustlnedon|ylflntheendwecanshowthatthepowerserles
weobtaln converges on some openlnterva|. Ifso, ltthenrepresents aso|utlonof
the dlerentla| equatlon on that lnterva|. The fo||owlngtheorem (whlch we state
wlthoutproof maybeusedforthlspurpose.
THEOREM 3 Radi us of Convergence
Glventhepowerserles_ .
-
x
-
,supposethatthe|lmlt

.
-

, = hm
--
.
-
,
,z;
exlsts,,l snnlte)or l slnnnlte(lnthlscasewewl||wrlte,=). Then
(a) If,=0,thentheserlesdlvergesfora||x = 0.
(b) If0 < , < 0, then_ .
-
x
-
convergeslf x < ,anddlvergeslf x > ,
(c) If,=0, thentheserlesconvergesfora||x
Thenumber,ln,z;lsca||edtheradius of convergence ofthepowerserles
_ .
-
x
-
Forlnstance,forthepowerserles obtalnedlnExample i , wehave
_.

, -i ;
-
z
-
.,}

|
+ i
,= rm = rm =0
--
, -i ;
-
z
-
,
.,},+i ;
--
z

and consequent|y the serles weobtalned lnExamp|e i converges forall x

Even
lfthe|lmltln ,z;fal|stoexlst,therea|wayswl||beanumber,such thatexact|y
oneofthethree a|ternatlves lnTheorem 3 ho|ds. Thls number may be dlmcu|tto
nnd,butforthepowerserleswewl||conslderlnthlschapter,Eq.,z;wl|lbequlte
sumclentforcomputlngtheradlusofconvergence.
202 Chapter 3 Power Seri es Methods
Exampl e Z So|vetheequatlon,x- 3) ,'+z, =0.
Sol uti on Asbefore,wesubstltute
toobtaln
sothat
O
and ,
'
= .
-
x
-

,
-
,
O O
,x- 3) .
-
x
-
-

+z.
-
x
-
= 0,
-
, -
O O O
.
-
x
-
- 3.
-
x
-
-

+z.
-
x
-
=0.
-, -
,
-
Inthe nrst sum we canrep|ace = iwlth = 0wlthnoeectonthesum. lnthe
secondsum we shlftthelndexofsummatlonby+i Thlsyle|ds
thatls,
O O O
.
-
x
-
- 3,+i ; .
-
x
-
+z.
-
x
-
=0,
- - -
O
.
-
- ,+i ; .
-
,+z.
- x
-
= 0.
-
Theldentltyprlnclp|ethenglves
.
-
- ,+i ; .
-
,+z.
-
= 0,
fromwhlchweobtalntherecurrencere|atlon
+ z
.

.
-
,

,+i ;
-
for _0.
Weapp|ythlsformu|awlth= 0,= i, and=z, lntum,andnndthat
3 3
.
:
=
-
., = ., ,
3 z 3
:
Thls ls a|most enough to make the pattem evldent, lt ls not dlmcult to show by
lnductlononthat
+ i
.
-
= ., lf _ i

-
Henceourproposedpowerserlesso|utlonl s
Itsradlusofconvergencel s

.
-

.
3n+3
p = hm = hm = 3 .
--
.
-
,
--
n+z
(26)
cXump| e
3. 1 I ntroducti on and Revi ew of Power Series 203
Thus the serles ln (26) converges lf-3 < x < 3 and dlverges lf x > 3. lnthis
partlcu|arexamp|ewecanexp|alnwhy. Ane|ementaryso|utlon(obtalnedbysepa
ratlonofvarlab|es)ofourdlerentla|equatlonls = I J(3-x;
:
ltwediereatlate
termwlsethegeometrlcserles
wegetaconstantmu|tlp|eoftheserlesln(26) . Thusthlsserles(wlththearbltrary
constant.,approprlate|ychosen)representstheso|ution
I
y,x; =
(3
.
x;
:
on the lnterva| -3 < x < 3, and the slngu|arlty at x = 3 ls the reason why the
radlusofconvergenceofthepowerserlesso|utlontumed outtobep = 3

So|vetheequatl

- x I .
.
Sol ution
Wemaketheusua|substltutlons = _ .
-
x
-
and

= _ ..x
-
-
,whlchyle|d
sothat
O O
x
:
.
-
x
-

= -I - x+.
-
x
-
-, .
O O
.
-
x
-

= -I x+..x
-

-, .
Becauseof thepresenceofthetwoterms I andxontherlght-handside,weneed
tosp|ltothenrsttwoterms, .,+.
,
x, of theserlesontherlghttorcomparison lf
wea|soshlftthelndexofsummatlononthe|eftby I (rep|acen = iwlthn 2
andwlth I ),weget
O O
,- i ; .
-
, x
-
= -I - x+.,+.
,
x+.
-
x
-

- -
Becausethe |eft-hand sldecontalns nelther aconstant term noraterm containlag
x to the nrst power, the ldentlty prlnclp|e now yie|ds ., = i , ., = i , and .. =
,- I;.
-
-
,for 2. Itfo||owsthat
.
:
= I

., = I , .

= 2
.
:
= 2 , .
-
= 3

= 3 ,
and, lngenera|,that
.
-
= ,- i ; for 2.
Thusweobtalnthepowerserles
O
y,x;= I +x+,
.
| ; x
-

-:
204 Chapter 3 Power Seri es Methods
Exampl e 4
Buttheradlusofconvergenceofthlsserlesls

,- i I
p hm |lm =0,
--

--

so the serles converges on|y for x c What does thls mean! Slmp|y that the
glvendlfferentla|equatlondoesnothavea(convergent)powerserlesso|utlonofthe
assumedform _ .
-
x
-
Thlsexamp|eservesasawarnlngthattheslmp|eactof
wrltlng _ .
-
x
-
lnvo|vesanassumptlonthatmaybefa|se.
. . .
So|vetheequatlon
o
+ c
Sol uti on Ifweassumeaso|utlonoftheform
we nndthat
O

.
-
x
-
-
,
-,
and
O

o
n(- i ; .
-
x
-
-
:

-:
Substltutlonfor and
o
lnthedlerentla|equatlonthenyle|ds
O O
,- i ;.
-
x
-
-
:
+.
-
x
-
=c
-: -
We shlftthelndexofsummatlonlnthenrst sum by+2(rep|ace zwlthn = 0
andwlth+z; Thlsglves
O O
,+z; ,+ i ; .
-:
x
-
+.
-
x
-
c
- -
Theldentlty ,+z; (n + i ; .
-:
+c
-
0 nowfo||owsfromtheldentltyprlnclp|e,
andthusweobtalntherecurrencere|atlon
.
-:

,+ i ; ,+z;
,z:;
for cItl sevldentthatthlsformu|awl||determlnethecoemclentsc,
wltheven
subscrlptlntermsofc

and those ofoddsubscrlptlntermsofc, ,c

andc, arenot
predetermlned and thus wl||bethe twoarbltrary constants we expect to nnd ln a
genera|so|utlonofasecond-orderequatlon.
When we app|ytherecurrencere|atlon ln ,z:;wlth 0, z, and:lnturn,
weget
c

-
z
.
c

TaklngI , 3, andlntum, we nnd that


c,
c

= -
3 '

c,
c

and c
s
-

and
c,
c

:
3. 1 I ntroducti on and Revi ew of Power Seri es 205
Agaln,thepatternl sc|ear,we|eaveltforyoutoshow(bylnductlon)thatfork I ,
Thuswegetthepowerserlesso|utlon
x
:
X
4
x
-
x

X 7
,(x) = c

I - + - +
. . .
+c,
X - + - +
. . .
.
z : :

thatls, ,(x) = c

cos x +c,sln x. Notethatwehavenoprob|em wlththeradiusof


convergencehere, the Tay|orserlesfor the slneand coslnefunctlonsconvergefor
a|| x.
The so|utlonofExamp|e:can bearfurthercomment. Supposethatwehad
neverheardoftheslneandcoslnefunctlons, |eta|onethelrTay|orserles. Wewou|d
thenhavedlscoveredthetwopowerserlesso|utlons
and
O
(
.
I )
,
x
:
,
x
:
X
4
C(x) = = I - + -
. . .
,a
,z z :
O
(
.
I )
,
x
:
,+,
x

S(x) =
,z+ i
=x -

+
5
. .

,zs
,z
ofthe dlerentla| equatlon ,

+ , = 0. Both ofthesepowerserles convergefor


a||x. For lnstance,theratlotestlnTheoremlmp|lesconvergencefora||z ofthe
serles_ (
.
I )
,
z
,
},z obtalnedfrom ,zsby wrltlngz =x
:
. Henceltfo||owsthat
,zsltse|fconvergesfora||x, asdoes (by aslml|arp|oy)theserlesln,z
Itl sc|earthatC,= I andS,=, andtermwlsedlerentlatlonofthetwo
serlesln ,zsand,zyle|ds
C
'
(x) ~ -S(x) and S
'
(x) = C(x) . ,
Consequent|y, C', = and S', = I . Thus wlththeald ofthepower serles
method (a|| the whl|e kowlng nothlng about the slne and coslne functlons), we
havedlscoveredthat, = C(x) lstheunlqueso|utlonof
,

+, =
that satlsnes the lnltla| condltlons ,(0) = I and ,' (0) = , and that , = S(x)
l sthe unlqueso|utlonthat satlsnesthe lnltla| condltlons ,(0) = and ,' (0) = I .
It fo||ows that C(x) and S(x) are |lnear|y lndependent, andrecognlzlngthe lm-
portanceofthedlerentla|equatlon,
+, =0wecanagreetoca||Cthe.os/e
functlonandSthes/efunctlon. Indeed,a||theusua|propertlesofthesetwofunc-
tlonscanbeestab|lshed,uslngon|ythelrlnltla|va|ues(atx =andthederlvatlves
ln ,, there ls no need to refer to trlang|es or even to ang|es. (Can you use the
serlesln ,zsand ,ztoshowthat C(x) ]
:
+ S(x) ]
:
= I fora|| x!) Thlsdemon-
stratesthat
The cosine and sine functions are fully determined by the diferen
tial equation y
o
+ y = 0 of which they are the two natural linearly
independent solutions.
206 Chapter 3 Power Seri es Methods
Flgures i zand i showhowthegeometrlccharacterofthegraphsofcosxand
slnx ls revea|edby thegraphsoftheTay|orpo|ynomla|approxlmatlonsthatweget
by truncatlngthelnnnlteserlesln ,zsand ,z
Thls lsbynomeansanuncommon sltuatlon. Many lmportant specla| func
tlonsofmathematlcsoccur lnthe nrst lnstanceaspower serles solutlonsofdler
entla| equatlons and thus are ln practlce ae]eaby means ofthesepower serles.
Intheremalnlngsectlonsofthls chapterwewl|| seenumerous examp|es ofsuca
functlons.
y
. s . . s . ;
. s . . . ;;
FIGURE 3. 1.2. Taylor polynomial approximations to
cos x.
.........,.......
..., ..,...........
..................,.
......,.........
.............,...
...............,......
......,.
1. .=
3. . .
5. .= .

.
7. . . .=
9. . ..=
2. . ..
4. . ..=
6. . . .=
8. . . =
10. . .= .
.. ...........,..
......,...,..........
., ..,.............
.........,..........
.......
11. .=
13. . .=
...= ..
....= .
.......,...,.......
...,........= _

..
., ..,..........
15. .. .=
17. .

. .=
.... = .
...

. = .
y
. . .
. ;
,
. . . . ;

FIGURE 3. 1.3. Taylor polynomial approximations to


sin x.
............
...

. .......,
,...................
...

..............
..,.,............
....
19. ...= .= . =
20. . ..= .= . =
21. . . .= .= . =
22. .. .= .= . =
23. Show that the equation
has no power series solution of the form .= _.
24. Establish the binomial series in ( ) by means of the fol
lowing steps. (a) Show that .= .satisfes the
initial value problem .. = .. .= ..Show
that the power series method gives the binomial series in
( ) as the solution of the initial value problem in part (a),
and that thi s series converges if . (c) Explain why
the validity of the binomial series given in ( ) follows
from parts (a) and (b) .
25. For the initial value problem
.= .. .= . =
derive the power series solution
- _
y(x) = _x
n
n=1
.
where { Fn } o i s the sequence 1 , 1 , 2, 3, 5, 8 , 1 3 ,
. . . of .....defned by o = = 1 ,
n =

2
+ n [ for . 1 .
26. (a) Show that the solution of the initial value problem
y' = 1 + y
2
, yeO) =
is y(x) = tan x. (b) Because y(x) = tan x is an odd
function with y' = 1 , its Taylor series is of the form
y = x + C
3
X
3
+ csx
s
+ C7X
7
+
. . .
.
Substitute this series in y' = 1 + y
2
and equate like powers
of x to derive the following relations:
3C
3
= 1 , 5s = 2C
3
,
7C7 = 2s + (C
3
)
2
, 9C
_
= 2C7 + 2C
3
CS ,
= 2C
9
+ 2C
3
C7 + (CS )
2
.
(c) Conclude that
1 2 1 7
tan x = x + x
3
+ _x
s
+ x
7
3 1 5 3 1 5
62 _ 1 382
I I
+
2835
x +
1 55925
x +
. . .
.
(d) Would you prefer to use the Maclaurin series formula
in ( 1 3) to derive the tangent series in part (c)? Think about
it !
3. 2 Seri es Sol uti ons Near Ordi nary Poi nts 207
27. Thi s section introduces the use of infnite series to solve
diferential equations. Conversely, diferential equations
can sometimes be used to sum infnite series. For exam
ple, consider the infnite series
1 1 1 1 1
1 + - - - + - + - - - +

I ! 2! 3! 4! 5 !
'
note the + + - + + -
. . .
pattern of signs superimposed
on the terms of the series for the number We could
evaluate this series if we could obtain a formula for the
function
1
2
1
3
1
4
1
5
J(x) = 1 + x - -x + -x + -x - -x +
. . .
2! 3! 4! 5!
'
because the sum of the numerical series in question is sim
ply J( I ) . (a) It' s possible to show that the power series
given here converges for all x and that termwise differen
tiation is valid. Given these facts, show that J(x) satisfes
the initial value problem
/
3
)
= y; yeO) = y' (O) = 1 , y" (O) = -1 .
(b) Solve this initial value problem to show that
For a suggestion, see Problem 48 of Section 2. 3. (c) Eval
uate J ( 1 ) to fnd the sum of the numerical series given
here.
Series Solutions Near Ordinary Points
. ....
Thepowerserles method lntroduced lnSectlon i can be applledto llnear equa
tlonsofanyorder(aswe||asto certalnnonllnearequatlons),butlts mostlmportant
appllcatlons are to homogeneous second-order |lnear dlerentlal equatlons otthe
form
A(x) ,

+ B(x) ,
'
+C(x) ,=, , i ;
where the coefnclents A, B, and C are ana|ytlc functlons ofx. Indeed, l nmost
app|lcatlonsthesecoefnclentfunctlonsareslmp|epo|ynomlals.
We sawlnExamp|eofSectlon i thatthe serles methoddoes notalways
yle|daserlesso|utlon. Todlscoverwhen ltdoes succeed,werewrlteEq. , i ; lnthe
form
,

+ P(x) ,
'+ Q(x) ,= (2)
wlth|eadlngcoemclent i , andwlth P = BJA and Q = CJA. NotethatP(x)and
Q(x) wl||genera|lyfal| tobeana|ytlcatpolntswhereA(x) vanlshes. Forlnstance,
consldertheequatlon
x,

+,'+x,=c ,;
208 Chapter 3 Power Seri es Methods
Exampl e Z
Exampl e
Thecoefnclentfunctlonsln(3)arecontlnuouseverywhere. Butlntheformof(2)lt
lstheequatlon
I
y

+ -y+y=
x
wlth P(x) = I Jx notana|ytlcat x = c
,:;
Thepolntx = .lsca||edanordinary point ofEq. (2)andoftheequlva|ent
Eq. ( I )provlded that the functlons P(x) and Q(x) areboth ana|ytlc at x = .
Otherwlse, x = .ls asingular point. Thus theon|yslngu|arpolntofEqs. (3)and
,:; ls x = c Reca||thata quotlentofana|ytlc functlons ls ana|ytlc whereverthe
denomlnatorls nonzero. Itfo||ows that, lfA(.) ln Eq. ( I ) wlth ana|ytlccoef-
nclents, thenx = .ls anordlnarypolnt. IfA(x) , B(x) , and C(x) arepo/yo-/./s
wlthnocommonfactors,thenx =.ls an ordlnarypolntlfandon|y lfA(.) c
xy

+ (sln x) ,
'
+x
:
y= ,
despltethefactthatA(x) = x vanlshesatx = c Thereasonlsthat
lsneverthe|essana|ytlcatx = becausethedlvlslonbyxyle|dsaconvergentpower
serles.
Thepolntx =ls oanordlnarypolntoftheequatlon
For whl|e P(x) = x
:
lsana|ytlc at the orlgln, Q(x) = x
,

:
lsnot. Thereasonls
that Q(x) lsnotdlfferentlab|eatx = andhencelsnotana|ytlcthere. (TheoremI
ofSectlon3 . I lmp|lesthatanana|ytlcfunctlonmustbedlerentlab|e. )

Thepolntx = lsanordlnarypolntoftheequatlon
because the coefnclent functlons A(x) , B(x) , and C(x) are po|ynomla|s wlth
A(0) c
Theorem 2 ofSectlon 2. I lmp|les that Eq. (2) has two |lnear|y lndependent
so|utlons onanyopenlnterva|wherethe coefnclentfunctlons P(x) and Q(x) are
contlnuous. Thebaslcfactfor ourpresentpurposel sthatnearanoa/.polnt..
theseso|utlonswl||bepower serlesln powers ofx - . A proofofthefo||owlng
theorem can be found ln Chapter 3 ofCoddlngton, ~ioa./o o oa/.
n_ ee/./i,./os(Eng|ewoodC|ls, N. J. . PrentlceHa||, I 96I ) .
Exampl e 4
3. 2 Series Sol uti ons Near Ordi nary Poi nts 209
TH EOREM 1 Sol utions Near an Ordi nary Point
Supposethatcl sanordlnarypolntoftheequatlon
A(x) ,

+B(x) ,
'
+C(x) ,=, , i ;
thatl s , themnctlonsP = BJAand Q = CJAareana|ytlcatx =c ThenEq., i ;
hastwo|lnear|ylndependentso|utlons,eachof theform

, (x) =

,;

The radlusofconvergenceofany suchserles so|utlon ls at|eastas|geasthe


dlstancefromcto thenearest (rea| orcomp|ex) slngu|arpolntofEq. , i ; The
coemclentsln theserlesln ,canbedetermlnedbyltssubstltutlonln Eq. , i ;
DetermlnetheradlusofconvergenceguaranteedbyTheorem iofaserlesso|utlon
of
(6)
lnpowersofx. Repeatforaserleslnpowersofx

:
Sol ution Thls examp|el||ushatesthefact that wemusttake lnto accountcomp|ex slngu|ar
polntsaswe||asrea|ones. Because
x
P(x) =
x
:
+ 9
and
x
:
Q(x) =
x
:
+ 9

the on|y slngu|arpolnts ofEq. (6) are +/ The dlstance (ln the comp|ex p|ane)
ofeach of these from ls , so a serles so|utlon oftheform _

has radlus
ofconvergence at |east The dlstance ofeach slngu|ar polntfrom :ls , so d
serlesso|utlonoftheform _
(x - :; has radlusofconvergence at |east (see
Flg. z i ;
y
X
-3;
FIGURE 3.2. 1. Radius of convergence as distance to nearest singularity.
21 0 Chapter 3 Power Seri es Methods
Exampl e Flndthegenera|so|utlonlnpowersofxof
,x
:
- :;y

+xy+y=

Then nndthepartlcu|arso|utlonwlthy,c;= :,y, i


Sol uti on Theon|yslngu|ar polnts ofEq. (7) are +z, so the serles wegetwl||have radlus
ofconvergence at |eastz (SeeProb|em for the exactradlus ofconvergence.)
Substltutlonof
O

-
y

.
-
x ,
-
ln Eq. (7)yle|ds
O

-
-
,
y =

.
-
x ,
-
,
O
and y

= ,- i .
-
x
-

:
-:
O O O O
,- i ; .
-
x
-
- :,- i .
-
x
-

:
+.
-
x
-
+.
-
x
-
=
-: -: -
,
-
Wecan beglnthe nrst and thlrd summatlonsat = aswe||,becausenononzero
terms are thereby lntroduced. We shlftthelndexofsummatlon ln thesecondsun
by+z, rep|aclngwlth+zanduslngthelnltla| va|ue= cThlsglves
O O O O
,- i ; .
-
x
-
- :,+z; ,+i ; .
-:
x
-
+.
-
x
-
+.
-
x
-
=
- - -
-
Afterco||ectlngcoefnclentsofc, and.
-:
,weobtaln
O
,
:
+z+i ; .
-
- :,+z; ,+i ; .
-.
:
x
-
=
-
Theldentltyprlnclp|eyle|ds
,+i ;
:
.
-
- :,+z; ,+i ; .
-:
= ,
whlch|eadstotherecurrencere|atlon
,+i ; .
-
.
-:
=
:,+z;
for _ cWlth= ,z,and:ln tum, weget
.
-
.,
and .

=
:

: :

z
.
:
.
:
Contlnulngln thl sfashlon,weevldent|ywou|dnndthat
i


,z- i ;
.
:-
=
.,

:
-
z
.
:
. . .
,z;
Wlththecommonnotatlon
,z+i ;
,z+i ; = i

,z+i ; =
z
-


(S)
3. 2 Series Sol uti ons Near Ordi nary Poi nts 21 1
andtheobservatlonthatz

:

,z;=z
-

n' , wenna||yobtaln
,z- i ;
.
:-
=
z
-

.,

(Wea|sousedthefactthat:
-

z
-
=z
-
. ;
Wth= I , 3, andln Eq. ,s;, weget
:.

z :.,
.,=
-
= .
: :
:

Itls apparentthatthepattemls
and
:., z : :.,
.

= -=
:
.
:

.
7
z

:

,z;
.
:-.
= ., = .,

:
-

i

,z+i ; z
-

,z+i ;
,;
, i ;
Theformu|aln ,glvesthecoemclentsofeven subscrlptlntermsofc,, the
formu|a ln , i ; glves the coefnclents of odd subscrlpt ln terms of., After we
separate|yco||ectthetermsoftheserlesofevenandodddegree, wegetthegenera|
so|utlon
A|tematlve|y,
y,x; =., i+-x +-x +-x +

i
:

s i zs i z:

i
,
i

+ ., x + -x + -x + -x +

: i :
, i i;
Becausey,c; = c, andy ,c; = ., ,theglvenlnltla|condltlonslmp|ythat., :
and ., = i Uslng theseva|ues ln Eq. , i i ;, the nrst few terms ofthepartlcu|ar
so|utlonsatlsfylngy,c;=:andy ,c;= iare
i
:
i _
-
i
,
y,x; =:+x+
_
x +

x +
z
x +

x +

, i z;

Remark: As lnExamp|e, substltutlonofy=_.


-
x
-
lna|lnearsecond-
orderequatlonwlthx=anordlnarypolnt typlca||y|eadstoarecurrencere|atlon
that can be used to express each ofthe successlve coemclents .
:
, .

, .
-
, ln
termsofthenrsttwo,.,and., Inthlseventtwo|lnear|ylndependentso|utlonsare
obtalnedasfo||ows. Lety,,x;betheso|utlonobtalnedwlth.,= iand ., =,and
|ety,,x;bethe so|utlonobtalnedwlth.,=and., = i Then
y, ,c;= i . y ,c;= and y,,; =. y,;= i .
soltl s c|earthaty,andy,are|lnear|ylndependent. InExamp|e , y, ,x;andy,,x
aredennedbythetwo serlesthat appearontherlght-handsldelnEq. , i i ;,whlch
expressesthegenera|so|utlonlntheformy=.,y,+., y,
21 2 Chapter 3 Power Seri es Methods
0"l'B
Translated Series Solutions
IflnExamp|ewehad soughtapartlcu|arso|utlonwlthglvenlnltla| va|ues y,.;
andy ,.; , wewou|dhaveneededthegenera|so|utlonlntheform
O
y,x; =

- .;
-
, , i );

thatl s, ln powers ofx- .ratherthan ln powersofx For on|y wlthaso|utlonot


theformln, i ;ls lttruethatthelnltla|condltlons
y,.; =and y ,.;=
determlnethearbltraryconstantsandlntermsofthelnltla|va|uesofyandy
Consequent|y,toso|veanlnltla|va|ueprob|em, weneedaserlesexpanslonofthe
genera|so|utloncenteredatthepolntwherethe lnltla|condltlonsarespeclned.
So|vethelnltla|va|ueprob|em
:
a
:
y ay
, - z- ;
a
:
+,- i ;
a
+y= , y, i ; =:. y , i ; = -i , i :;
Sol uti on Weneedagenera|so|utlonoftheform_

Butlnsteadofsubstltutlngthls
serlesln, i :;todetermlnethecoefnclents, ltslmp|lnesthecomputatlonslfwenrst
makethesubstltutlonx= i . sothatwewlndup|ooklngforaserlesoftheforn
_

aer a||. TotransformEq., i :;lntoonewlth the new lndependentvarlab|e


x, we notethat
and

:
- z- =,x+ i ;
:
- z,x+ i ; - =x
:
- :.
ay ayax ay ,
a

axa
=
ax
=y.
a
:
y ay ax

a
:
ax ax a

ax
,y;

y .
whereprlmesdenotedlerentlatlonwlthrespecttox HencewetransformEq. , i :;
lnto
,x
:
- :;y

+xy+y=
wlthlnltla|condltlonsy= :andy= iatx= (correspondlngto= i ; Thlsls
thelnltla|va|ueprob|emwe so|vedlnExamp|e. sothepartlcu|arso|utlonln, i z;
ls aval|ab|e. We substltute - i forx ln Eq. , i z;and thereby obtaln the deslred
partlcu|arso|utlon
i i
y, ; =:+ ,- i ; +

,- i ;
:
+

,- i ;

)
-
i

+
z
,- i ; +

,- i ; +

.
cXump| e
3. 2 Series Sol uti ons Near Ordi nary Poi nts 21 3
Thlsserlesconvergeslf-i < < (Why!) A serles suchas thls canbeusedto
estlmatenumerlca|va|uesofthe so|utlon.Forlnstance,
i i
y, s; :+,- z;+_,- z;
:
+_,- z;

sothaty( s) s i ss
_ i

+-,- z; +-, z; +

z

The |ast computatlonlnExamp|e:l||ustratesthefactthatserlesso|utlonsot


dlerentla| equatlons are usefu| not on|y for estab|lshlng genera| propertles ota
so|utlon,buta|sofornumerlca|computatlonswhenanexpresslonoftheso|utlonln
termsofe|ementaryfunctlonslsunaval|ab|e.
Tyes of Recurrence Relations
Theformu|alnEq. ,s;lsanexamp|eofatwo-term recurrencere|atlon,ltexpresses
eachcoemclentlntheserleslntermsofoftheprecedlngcoemclents. Amany
term recurrencere|atlonexpresseseach coemclentlntheserleslntermsoftwoor
moreprecedlng coefnclents. In the case ofa many-term recurrence re|atlon, lt ls
genera||y lnconvenlent oreven lmposslb|eto nnd aformu|athat glves thetyplca|
coefnclent
lnterms of

The next examp|e shows whatwe sometlmes cando


wlthathree-termrecurrencere|atlon.
Flndtwo|lnear|ylndependentso|utlonsof

c , i ;
Sol uti on We maketheusua|substltutlonofthepowerserles _

Thlsresu|tslnthe
equatlon
O O O

c

We can start the second sum at wlthoutchanglng anythlnge|se. Tomake
eachtermlnc|ude

lnlts genera| term, weshlftthelndexofsummatlonlnthenrst


sumby+z(rep|acewlth+z;. andweshlftltby-zlnthethlrdsum(rep|ace
wlth- z;Theseshlftsyle|d
O O O
+z (+

c

The common range ofthesethree summatlons ls _ z. sowe must separate the
terms correspondlng to and i ln the nrst two sums before co||ectlng
coefnclentsof

Thls glves
O

+
++z (+

The ldentltyprlnclp|enow lmp|lesthatz

,that andthethree-term
recurrencere|atlon

+z (+i ;

21 4 Chapter 3 Power Seri es Methods
forn z Inpartlcu|ar,
.,+.

=
:z
.

+.,
.,=
z
:.

+.
-
.,=
:
, i :;
Thusa|| va|uesof.
-
forn :areglvenlnterms ofthearbltraryconstants.,and
., because.
:
=and.

= ., .
Togetournrstso|utlony, ofEq. , i ;, wechoose.,= i and ., =,sothat
.
:
=.

=cThentheformu|asln , i :;yle|d
thus
.,=.

i i z

i
-
i

,
y,,x;= i+ -x + -x + x +
. . .
.
i z i i z
, i s;
Because., =.

=,ltlsc|earfromEq. , i :;thatthlsserlescontalnson|y temsot


evendegree.
Toobtalnasecond|lnear|ylndependentso|utlonY
2
ofEq. , i ; , wetake.,-
and., = i , sothat.
:
=and.

= Thentheformu|asln , i :;yle|d
sothat
.
-
=.

.,=
:

= .
i

,
i

Y
2
(X) =x+ -x + -x + x +
. . .
.
: : i s
, i ;
Because., = .
:
= ,ltls c|earfrom Eq. , i :;thatthlsserlescontalns on|y tems
ofodd degree. The so|utlons
y,,x; and Y
2
,x; are |lnear|y lndependent because
y,,; = iandY

,; =,whereasY
2
,;=andY
,;= i Agenera|so|utlonof
Eq. , i ;lsa|lnearcomblnatlonofthepowerserlesln, i s;and, i ; Equatlon, i ;
has no slngu|arpolnts, sothepowerserlesrepresentlng y,,x;and Y
2
,x;conveqe
fora||x
The Legendre Equation
TheLegendre equation oforder- l sthesecond-order|lneardlerentla|equatlon
, i - x
2
; y - zxy+-- + i ; y=. ,z;
wheretherea|number- satlsnesthelnequa|lty- > -i Thlsdlerentla|equatlon
has extenslve app|lcatlons, ranglngfrom numerlca| lntegratlon formu|as (such as
Gausslan quadrature) to the prob|em ofdetermlnlng the steady-state temperature
wlthln a so|ld spherlca| ba|| when the temperatures at polnts oflts boundary are
known. The on|y slngu|arpolnts of the Legendre equatlon are at +i and - i , so
lthas two |lnear|y lndependent so|utlons thatcanbeexpressed as powerserles ln
3. 2 Series Sol uti ons Near Ordi nary Poi nts 21 5
powersofwlthradlusofconvergenceat|easti Thesubstltutlony _

ln
Eq. ,z|eads (seeProb|em totherecurrencere|atlon
,: - :+-+


,-+ -

z
for- c Weareuslng-asthelndexofsummatlonbecausewehaveanotherro|e
fortop|ay.
Intermsofthearbltraryconstantsand Eq. z yle|ds
:,:+ i ;

=
z
,:- i ; ,:+z


:,: z ,:+ i ; ,:+


,:- i ; ,:- ); ,:+z ,:+



Wecanshowwlthoutmuchtroub|ethatfor- > ,
:,: z ,: :- z-+z ,:+ i ; ,:+) |o+z-

,zz
and

,:- i ; ,:

,:- z-+ i ; ,:+z ,:+

,:+z-;

,-i ;
,z-+

z)
A|tematlve|y,
where

and

denotethefractlonslnEqs. ,zzand,z, respectlve|y. Wlth


thlsnotatlon,we get two |lnear|ylndependentpowerserlesso|utlons
O

ofLegendre' sequatlonoforder:
O
and

,z
Now suppose that: , a nonnegatlve /e,e If: = ls even, we see
from Eq. ,zzthat

when z- > In thls case,


lsapo/yo-/./of
degreen and

ls a (nontermlnatlng) lnnnlte serles. If: ls an oddposltlve


lnteger, we seefromEq. z that

= when z-+ > In thls case,

lsapo/yo-/./ofdegreen and
lsa(nontermlnatlng)lnnnlteserles. Thus
ln elther case, oneofthe two so|utlons ln zls a po|ynomla| and the other lsa
nontermlnatlngserles.
21 6 Chapter 3 Power Seri es Methods
Problems
Wlthanapproprlatecholce(madeseparate|yforeachn) ofthearbltrarycon-
stantsc

(n even) or., (n odd), thenth-degreepo|ynomla| so|utlonofLegendre's


equatlonofordern,
( I - x
:
) , - 2x,
'
+n(n +I ) , =0, (25)
l sdenotedby P,(x) andlsca||edtheLegendre polynomial ofdegreen. Itlscus-
tomary(forareason lndlcatedlnProb|em 32) to choosethe arbltrary constantso
thatthecoefnclentofx
,
lnP,(x) l s(2n) ' J2
,
(n ' )
:
. Itthentumsoutthat

( -I )
,
(2n - 2k) '
,

:
,
P, (x) = x ,
,
=

2
,
k' (n- k) '(n - 2k) '
(26)
wherex=]nJ2] ,thelntegra|partofnJ2. ThenrstslxLegendrepo|ynomla|sare
P

(x) = I ,
I
:
P
:
(x) =
j
(3x - I ) ,
I

(x) =
j
(5x - 3x) ,
I
P
+
(x) =

(35x
+
- 30x
:
+3) ,
I
P

(x) =

(63x
:
- 70x

+I 5x) ,
andthelrgraphsareshownl nFlg. 3. 2. 2.
)
FIGURE 3.2.2. Graphs .=

.of the Legendre polynomials for


.= .and The graphs are distinguished by the fact that all .
zeros of

.lie in the interval .


.........,....., ..,..
..............
.....................
7. .

. .. .= 0
8. .

. .. . 0
9. .

... .= 0
1. .

.... .= 0
10. ... ..= 0
2. .

..... 0
3. ... .= 0
11. . .. .= 0
12. . .

. ..= 0
4. .

... ..= 0
13. ..

. ..= 0
5. .

... = 0 14. ... 0 (an .,...


6. .

. .. . 0 15. . .

.= 0
.,.........,..
....
16. ( 1 .

... .0; .0,


. 1
17. ... .0; .1 , . 0
.....,.......
............ . .....
..._

......., ..,....
................
........
18. .. . .0; . 1 2, . 0
19. . .

. . . ..0; . 1 0, . 1
20. .

. . .. . .0;
.2
,
. 0
21. ..

. .. . 1 , . 0
22. .

. ... .0; . 0, . 2
............
........_

....
..............,...
....
23. .( 1 ..0
24. .

... ..0
25. . .

. .

.0
26. .

..

.0
27. Solve the initial value problem
....

.0;
.1 , . -1 .
Determine sufciently many terms to compute . ac
curate to four decimal places.
..............
.........,........
Y _

.................
..............,......
,..
28. ..0
29. cos . ..0
30. ... . . ..0
31. Derive the recurrence relation in (21 ) for the Legendre
equation.
32. Follow the steps outlined in this problem to establish Ro
drigues's formula
for the nth-degree Legendre polynomial . (a) Show that
.

satisfes the diferential equation


( 1 .

..
Diferentiate each side of this equation to obtain
.

. ..
3. 2 Series Sol uti ons Near Ordi nary Poi nts 21 7
(b) Diferentiate each side of the last equation .times in
succession to obtain
( 1 .

..

Thus .

satisfes Legendre' s equa


tion of order . (c) Show that the coeffcient of .

in .is . . then state why this proves Rodrigues'


formula. (Note that the coeffcient of .

in

.is
2.

.'


33. The Hermite equation of order a is
. ....=
(a) Derive the two power series solutions
2ma(a - 2) . . .
(a 2)
][ 1

( -I )" .

,
_
and
]_ .
2m (a .

(a 1 )

( -I )" .

.
I ) '
.
Show that ][ is a polynomial if a is an even integer,
whereas ]_ is a polynomial if a is an odd integer. (b) The
Hermite polynomial of degree .is denoted by H

. It
is the nth-degree polynomial solution of Hermite' s equa
tion, multiplied by a suitable constant so that the coef
cient of .

is 2
n
. Show that the frst six Hermite polyno
mials are
Hq . 1 ,

...

2,
H4 (x) .

..

12,
H
s
(x) 32x
s
- 1 60x
3
1 20x
.
HI .2x,
H
3
(x) .

- 1 2x,
A general formula for the Hermite polynomials is
Verify that this formula does in fact give an nth-degree
polynomial . It is interesting to use a computer alge
bra system to investigate the conjecture that (for each
.the zeros of the Hermite polynomials H
n
and H
n
+l
are "interlaced"-that is, the .zeros of H
n
lie in the .
bounded open intervals whose endpoints are successive
pairs of zeros of


34. The discussion following Example .in Section 3. 1 sug
gests that the differential equation ..0 could be
used to introduce and defne the familiar sine and cosine
functions. In a similar fashion, the .,...
. ..
21 8 Chapter 3 Power Seri es Methods
serves to introduce two new special functions that appear
in applications ranging from radio waves to molecular vi
brations. Derive the frst three or four terms of two dif
ferent power series solutions of the Airy equation. Then
verify that your results agree with the formulas
and

.
.
. .

.
.- 2)
Y
l
(x)

.
X
3k
Y
2
(X) x

_
2

5
. . . . .
.
X
3k+
l
k=l
.

I ) !
for the solutions that satisfy the initial conditions Y
l

Y and Y
2
Y respectively. The
special combinations
and
B.
Y
l
(x)

Y
2
(X)
(x) -

defne the standard .....that appear in math


ematical tables and computer algebra systems. Their
graphs shown in Fig. exhibit trigonometric-like os
cillatory behavior for x whereas Ai (x) decreases
exponentially and Bi (x) increases exponentially as
x It is interesting to use a computer algebra
system to investigate how many terms must be retained in
the Y
l
- and Y
2
-series above to produce a fgure that is visu
ally indistinguishable from Fig. (which i s based on
high-precision approximations to the Airy functions).
Regular Singular Points
. .
)
FIGURE 3.2.3. The Airy function graphs
Y Ai(x) and Y Bi (x) .
35. (a) To determine the radius of convergence of the series
solution in Example 5, write the series of terms of even
degree in Eq. in the form
- -
Yo (x)

_C
2
n
X
2
n

_a
n
z
n
n=l n=
l
where a
n
C
2
n
and z x
2
Then apply the recurrence
relation in Eq. and Theorem in Section to show
that the radius of convergence of the series in z is .Hence
the radius of convergence of the series in x is 2. How does
this corroborate Theorem in this section? (b) Write the
series of terms of odd degree in Eq. in the form
to show similarly that its radius of convergence (as d
power series in x) is also 2.
Wenowlnvestlgatetheso|utlonofthehomogeneoussecond-order|lnearequatlon
~,x; y

+ s,x; y+c,x; y= , i
near a slngu|ar polnt. Reca|| that l fthe functlons ~, s, and c are po|ynomla|s
havlng no common factors, then the slngu|ar polnts ofEq. , i are slmp|y those
polntswhere~,x;=cForlnstance,x=lstheon|yslngu|arpolntoftheBessel
equatlonoforder,
x
:
y

+xy+ ,x
:
-
:
; y= .
whereastheLegendreequatlonoforder,
, i - x
:
; y

- zxy+,+ i ; y=.
hasthetwoslngu|arpolntsx=-iandx= i lttumsoutthatsomeofthefeatures
oftheso|utlonsofsuchequatlonsofthemostlmportanceforapp|lcatlonsare|arge|y
determlnedbythelrbehavlornearthelrslngu|arpolnts .
3. 3 Regul ar Si ngul ar Poi nts 21 9
We wl||restrlctourattentlon tothecase ln whlchx = ls a slngu|arpolnt
ofEq. , i ; Adlerentla|equatlonhavlngx =O asaslngu|arpolntls easl|ytrans-
formedbythesubstltutlon=x- O lntoonehavlngacorrespondlngslngu|arpolnt
atcForexamp|e,|etus substltute=x - i lntotheLegendreequatlonoforder
Because
,
ay aya ay
y = - = -- = -,
ax aax a
and i- x
:
= i- ,+ i )
:
=-z-
:
,wegettheequatlon
a
:
y ay
- ,+z)
:
- z(+ i ; - +,+ i ) y=c
a a
Thls new equatlon has the slngu|ar polnt = correspondlng to x = i ln the
orlglna|equatlon, lthas a|sotheslngu|arpolnt= -zcorrespondlngtox=-i
Tyes of Singular Points
A dlfferentla|equatlonhavlng aslngu|arpolntatordlnarl|ywl||ohavepower
serlesso|utlonsof theformy,x; = _.
-
x
-
,so thestralghtforwardmethodofSec-
tlon zfal|slnthlscase. Tolnvestlgatetheformthataso|utlonofsuchanequatlon
mlghttake, weassumethatEq. , i ; hasana|ytlccoefnclentfunctlonsandrewrltelt
lnthestandardform
y

+ i,x; y+ _,x; y=. ,z;


where i= s,~and Q =c,~Reca||thatx=lsanordlnarypolnt(asopposed
to a slngu|ar polnt) ofEq. ,z;lfthefunctlons i,x; and _,x; are ana|ytlc at x
, that l s, lf i,x; and Q,x; have convergent power serles expanslons ln powers
ofx on some open lnterva| contalnlng x = c Now lt can be proved that each
ofthe functlons i,x; and _,x; e//els ana|ytlc oapproaches +> as x - c
Consequent|y, x = ls a slngu|ar polnt of Eq. ,z; provlded that elther i,x;or
_,x; (or both) approaches +>asx - c Forlnstance, lfwe rewrlte theBesse|
equatlonoforderlntheform

i
,

:

y + y + i- y=.
X x
:
weseethati,x;= i,xand Q,x;= i- ,,x;
:
both approach lnnnltyasx- c
Wewl||seepresent|ythatthepowerserlesmethodcanbegenera|lzedtoapp|y
neartheslngu|arpolntx =ofEq. ,z;, provldedthati,x;approacheslnnnltyno
more rapld|y than i }x, and _,x; nomore rapld|ythan i }x
:
,asx - c Thls lsa
wayofsaylngthati,x;and Q,x;haveon|y 'weakslngu|arltlesatx=cTostate
thlsmorepreclse|y,werewrlte Eq. ,z;lntheform
where

+
p,x;
' +
, ,x;
y

y y = .
X x
:
p,x; =xi,x; and , ,x; =x
:
_,x;
,;
,:;
220 Chapter 3 Power Seri es Methods
Exampl e 1
DEFI NITI ON Regul ar Si ngul ar Point
Theslngu|arpolntx =ofEq. ,)l saregular singular point lfthefunctlons
p,x) and, ,x)arebothana|ytlcatx =c Otherwlseltls anirrgular singular
point.
Inpartlcu|ar, theslngu|arpolntx =l sae/.slngu|arpolntlfp,x;and
, ,x; are both po|ynomla|s. For lnstance, we seethatx = ls aregu|ar slngu|m
polntofBesse| ' sequatlonoforderbywrltlngthatequatlonlntheform
i x
:
-
:
y

+y+
:
y=.
x x
notlngthatp,x; = iand, ,x; =x
:
-
:
arebothpo|ynomla|slnx
Bycontrast,consldertheequatlon
zx

+, i +x; y+xy=.
whlchhastheslngu|arpolntx=cIfwewrltethlsequatlonlntheformof,;, we
get

, i +x; },zx
:
;
,
]
y + y +

y=
x x
Because
i+x i i
p,x; = = -+
zx
:
zx
:
zx
asx (a|though, ,x; = ] ls apo|ynomla|), weseethatx = ls an lrregu|ar
slngu|arpolnt. We wl||notdlscusstheso|utlonofdlerentla|equatlonsnearlrreg-
u|arslngu|arpolnts ,thls ls aconslderab|ymoreadvancedtoplcthantheso|utlonot
dlerentla|equatlonsnearregu|arslngu|arpolnts.
.. . .. ......... ... . . ......
Conslderthedlerentla|equatlon
x
:
, i +x; y
+x,:- x
:
; y+,z+x; y=c
Inthestandardformy+iy+_y=ltls
Because

:
.
x
:
,
z+x
y +
x, i+ x;
y +
x
:
, i + x;
y=c
: - x
:
i,x;

x,
-
i
-
+
-
x

;
and
z + x
_,x; =
x
:
, i +x;
bothapproachasx, weseethatx=lsaslngu|arpolnt. To determlnethe
natureofthlsslngu|arpolntwewrltethedlerentla|equatlonln theformofEq.,;
Thus

,:- x
:
; }, i +x; ,
,z+x; }, i +x;
y + y +
:
y =
x x
: - x
:
p,x; =
i + x
z + x
and , ,x; = -
i + x
Becauseaquotlentofpo|ynomla|slsana|ytlcwhereverthedenomlnatorlsnonzero,
we see that p,x; and , ,x; are both ana|ytlc atx = c Hencex = ls a e/.
slngu|arpolntoftheglvendlerentla|equatlon.
Exampl e Z
3. 3 Regul ar Si ngul ar Poi nts 221
Itmayhappenthatwhen webeglnwlth adlerentlalequatlonlnthegeneral
formlnEq. , i ; andrewrlteltlntheformln,;, thefunctlonsp,x; and,,x;asglven
ln,:;arelndetermlnateforms atx =c In thls casethesltuatlon ls determlnedby
thellmlts
and
;
=p,c;= |.m p,x; = |.m xi,x;

,,=,
tc;= llm, ,x; = llmx
:
_,x;

,;
(6)
If
;
= = ,

, thenx = maybean ordlnarypolntofthedlerentlalequatlon


x
:
y+xp,x; y+, ,x; y=ln,; Otherwlse.
Ifboth the llmlts ln ,;and (6) exlstand are]/e, then x = ls aregular
slngularpolnt.

Ifeltherllmltfallsto exlstor ls lnnnlte, then x = ls an lrregular slngular


polnt.
Remark: Themostcommoncaselnappllcatlons,forthedlerentlalequa-
tlon wrlttenlntheform
,,
+
p,x;
,
+
, ,x;
y

y y = .
X x
:
,;
lsthatthe functlons p,x; and , ,x; arepo/yo-/./s Inthls case
;
= p,c;and
,,= , ,c;areslmply the constantterms ofthesepolynomlals, so there ls no need
toevaluatethellmlts lnEqs. ,;and(6).
Tolnvestlgatethenatureofthepolntx=forthedlerentlalequatlon
x
-
y

+,x
:
.a x; y+, i - .a x; y=.
wenrst wrlteltlntheformln,;

,.a x; }x
,
, i - .a x; }x
:
y + y +
:
y=c
x x
Thenl ' Hpltal ' sruleglvesthevalues
and

.a x , cosx
;= hm= hm= i

i
i- cosx .a x i
,,= llm = llm= -

x
:

zx z
forthellmltsln,;and(6). Slncetheyare notboth zero, weseethatx =ls not
anordlnarypolnt. Butboth llmlts arennlte, sotheslngularpolntx =lsregular.
Altematlvely,wecouldwrlte
222 Chapter 3 Power Seri es Methods
and
I - cos x I x
:
x
-
x


, ,x; = = I - I - + - +
. . .
x
:
x
:
2' 4' 6'
I x
:
x
-

- " + " -
2' 4' 6'
.
These(convergent)powerserlesshowexp|lclt|ythatp,x; and,,x;areana|ytlcand
moreoverthat
;, = p,c; = I and,, = , ,c; = | , thereby verlfylngdlrect|ythat
x= 0ls aregu|arslngu|arpolnt.
The Method of Frohenius
Wenowapproachthetaskofactua||ynndlngso|utlonsofasecond-order|lneardlt-
ferentla|equatlonneartheregu|arslngu|arpolntx= c Theslmp|estsuchequatlon
lstheconstant-coemclente,/a/-es/o./e,./o
: /
'
0 x y +
;,xy +,,y (7)
towhlchEq. ,;reduceswhenp,x; = ;,and, ,x;= ,,areconstants. Inthl scase
wecan verlfybydlrectsubstltutlonthattheslmp|epower functlony,x; ~ x ls a
so|utlonofEq. (7)lfandon|ylfls arootofthequadratlcequatlon
,- I ) +p,+,,= c (S)
Inthegenera| case, lnwhlch p,x; and, ,x; arepowerserlesratherthancon-
stants, ltlsareasonab|econecturethatourdlerentla|equatlonmlghthaveaso|u-
tlonoftheform
O O
ylx) =x.
-
x
-
= .
-
x
-
= .,x+.
. x
.
+.
:
x
:
+
. . .
(9)
-
-

the product ofx and a power serles. Thls tums outto be a very frultfu| con-
ecture, accordlngtoTheorem I (soontobestatedforma||y), everyequatlonofthe
form ln ( I ) havlngx = 0asaregu|arslngu|arpolntdoes, lndeed,haveat|eastone
such so|utlon. Thls fact ls thebaslsfor the method of Frobenius, namedforthe
GermanmathematlclanGeorgFrobenlus( I S4SI 9I 7) , whodlscoveredthemethod
lnthe I S70s.
Anlnnnlte serles oftheform ln(9)l sca||eda Frobenius series. Notethat
aFrobenlus serles ls genera||yoapowerserles. Forlnstance, wlth = -| the
serlesln (9)takesthe torm
lt ls not aserlesln /e,m/powersofx
To lnvestlgate theposslb|eexlstenceofFrobenlus serles so|utlons, webegln
wlththeequatlon
x
:
y
+xp,x; y+, ,x; y= 0 ( I 0)
3. 3 Regul ar Si ngul ar Poi nts 223
obtalnedbymu|tlp|ylngtheequatlonln ,;byx
:
Ifx = ls aregu|arslngu|ar
polnt,thenp,x; and, ,x; areana|ytlcatx=, so
p,x; =
;,+;, x+;
:
x
:
+;
x

+

,
, ,x; =,,+,, x+,
:
x
:
+,

SupposethatEq. , i ;hastheFrobenlusserlesso|utlon
O
y= .
-
x
-

-

, i i ;
, i z;
We may (and a|ways do) assume that., = becausethe serles must have anrst
nonzeroterm. TermwlsedlerentlatlonlnEq. , i z;|eadsto
and
O
y= .
-
,+;x
-

-
,
-
O
y

= .
-
,+) (+- i ;x
-

-
:

-
( i )
, i :;
SubstltutlonoftheserleslnEqs. , i i ; through, i :;lnEq. , i ;nowyle|ds
,- i ; .,x + ,+i ; ., x
.
+
.

+;,x+;, x
:
+

.,x
-
,
+,+i ; ., x+
+,

+,, x+
. . .

.
.,x+., x
.
+ =c , i ;
Upon mu|tlp|ylnglnltla| termsofthe twoproductsonthe |eft-hand sldehere and
thenco||ectlngcoemclentsofx , we see that the |owest-degreeterm lnEq. , i ;ls
., ,- i ) +p, +,,}x If Eq., i ;lstobesatlsnedldentlca||y,thenthecoemclent
ofthlsterm (as we|| as thoseofthehlgher-degreeterms) mustvanlsh. Butweare
assumlngthat., =, soltfo||owsthatmustsatlsfythequadratlcequatlon
,- i ; +p,+,,- , i :;
ofpreclse|ythesameformasthatobtalnedwlththeequldlmenslona|equatlonln,:;
Equatlon, i :;lsca||edtheindicial equation ofthedlerentla|equatlonln, i ;, and
lts two roots (posslb|yequa|) aretheexponents ofthedlerentla|equatlon (atthe
regu|arslngu|arpolntx=;
OurderlvatlonofEq. , i :;showsthatqtheFrobenlusserlesy=x_.
-
x
-
ls
tobeaso|utlonofthedlerentla|equatlonln, i ; , /etheexponentmustbeone
oftheroots,and
:
ofthelndlcla|equatlonln, i :; If, =
:
,ltfo||owsthatthere
are two posslb|eFrobenlus serles so|utlons, whereas lf, =
:
there ls on|y one
posslb|eFrobenlusserlesso|utlon, thesecondso|utloncannotbeaFrobenlusserles.
Theexponents , and
:
lntheposslb|eFrobenlus serles so|utlons are determlned
(uslngthelndlcla|equatlon)bytheva|ues
;, = p,c;and,, = ,,c;thatwehave
dlscussed. Inpractlce,partlcu|ar|ywhenthecoefnclentslnthedlerentla|equatlon
224 Chapter 3 Power Seri es Methods
cXump| e
lntheorlglna|formln, i ; arepo|ynomla|s, theslmp|estwayofnndlng

and,,ls
oentowrltetheequatlonlntheform
+
+

+

, ,,+ +

o
+ +

= c , i :;

Thenlnspectlonoftheserlesthatappearlnthetwonumeratorsrevea|stheconstants
and,,

Flndtheexponentslntheposslb|eFrobenlusserlesso|utlonsoftheequatlon
Sol uti on We dlvldeeachtermby

,i+ torecastthedlerentla|equatlonlntheform
and thusseethat

= and,,= - Hencethe lndlcla|equatlonls


,- i;+- =

+ - = ,+

i ; ,- =.
wlth roots , = and

= -i Thetwoposslb|eFrobenlus serles so|utlons are


thenoftheforms
O
=

Frohenius Series Solutions


O
and

Oncetheexponents , andr
:
areknown,thecoefnclents lnaFrobenlus serlesso-
|utlonaredetermlnedbysubstltutlonofthe serles ln Eqs. , i z;through , i :;lnthe
dlerentla| equatlon, essentla||y the same method as was used to determlne coef
nclents ln power serles so|utlons ln Sectlon z If the exponents , and r
:
are
comp|exconugates, thentherea|waysexlsttwo|lnear|ylndependentFrobenlusse-
rles so|utlons. Wewl||restrlctourattentlonheretothecaselnwhlch, andr
:
are
both rea| . We a|sowl|| seekso|utlons on|y for > c Once such a so|utlon has
been found, we need on|y rep|ace wlth toobtaln a so|utlon for < c
The fo||owlngtheorem ls proved ln Chapter:ofCoddlngton' s~ia./oo
oa/.n_ ee/./i,./os
cXump| e4
3. 3 Regul ar Si ngul ar Poi nts 225
THEOREM 1 Frobeni us Series Sol uti ons
Supposethatx l saregu|arslngu|arpolntoftheequatlon
x
:
y

+xp,x y+, ,x y , i ;
Letp > denotethemlnlmumoftheradllofconvergenceofthepowerserles
O O
p,x
;-
x
-
and , ,x = ,
-
x
-

Let, and
:
bethe(rea|)roots,wlth,
:
,ofthelndlcla|equatlon,- i ; +
;,+,, cThen
(a) Forx > , thereexlstsaso|utlonofEq., i ;oftheform
O
y, ,x; x.
-
x
-
-

conespondlngtothe|argerroot,
,.

= ; , i s;
(b) If, -
:
ls neltherzero nor a posltlve lnteger, then there exlsts a second
|lnear|ylndependentso|utlonforx > oftheform
O
y
:
,x; x/
-
x
-
-

correspondlngtothesma||erroot
:

,/,= ; , i ;
The radll ofconvergence ofthe power serles ln Eqs. , i s;and , i ;are each at
|east p. The coefnclents ln these serles can be determlned by substltutlng the
serlesln the dlerentla|equatlon
x
:
y

+xp,x y+, ,x y c
Wehavea|readyseenthatlf, =
:
,thentherecanexlston|yoneFrobenlus
serles so|utlon. Ittums out that, lf, -
:
ls aposltlve lnteger, theremay ornay
notexlstasecondFrobenlusserlesso|utlonoftheformlnEq. , i ;correspondlngto
thesma||erroot
:
Theseexceptlona|casesaredlscussedlnSectlon :Exanp|es
:through6l||ustratetheprocessofdetermlnlngthecoefnclentslnthoseFrobenlus
serlesso|utlonsthatare guaranteedbyTheorem i
FlndtheFrobenlusserlesso|utlonsof
zx
:
y

+xy
- ,x
:
+i ; y= ,z;
Sol uti on Flrstwedlvldeeachtermby2x
:
toputtheequatlonlntheformln, i :;
2 al a i
x
:
y

+=y+
: :
y=c
x x
:
,zi ;
We now seethatx = lsaregu|arslngu|arpolnt,andthat
;, = and,, = -
Becausep,x = and, ,x -- x
:
arepo|ynomla|s, theFrobenlusserleswe
226 Chapter 3 Power Seri es Methods
obtalnwl | | convergefora||x > cThelndlcla|equatlonls
, I)+ = , ,+i ) = 0,
so the exponents are , = and
:
= i They donotdlerbyanlnteger, so
Theorem I guarantees the exlstence oftwo |lnear|y lndependentFrobenlus serles
so|utlons. Ratherthanseparate|ysubstltutlng
O
y, =x
,

:
_.
-
x
-
-

O
and
y
:
=x

,
/
-
x
-
-
ln Eq. (20), lt ls more emclentto beglnby substltutlng y = x_.
-
x
-
We wl||
then getarecurrencere|atlonthatdependson Wlththe va|ue , = ltbecomes
a recurrence re|atlon for the serles for y, , whereas wlth
:
= -i lt becomes a
recunencere|atlonfortheserlesfory
:

and
Whenwesubstltute
O
y= _.
-
x
-
,
-
O
O
y
= +; .
-
x
-

-
,
,
-

y
= +; ,+- i ; .
-
x
-

:
-
lnEq. (20)theorlglna|dlerentla|equatlon,ratherthan Eq. (2I )weget
O O
2+; ,+- i ; .
-
x
-
+J+; .
-
x
-

-
-

O O
.
-
x
-

:
.
-
x
-
=c (22)
- -
Atthlsstage therearesevera| ways toproceed. A good standardpractlcels toshln
lndlces sothateachexponentwll|bethe sameas the sma||estonepresent. Inthls
examp|e, we shlft the lndex of summatlon ln the thlrd sum by -2 to reduce lts
exponentfrom++2to+Thl sglves
O O
2+; ,+- i ; .
-
x
-
++; .
-
x
-

-
-
O O
.
-:
x
-
.
-
x
-
=c (2J)
-:
-

The common range ofsummatlon ls 2, sowemusttreat = 0 and = i


separate|y. Fo||owlngourstandardpractlce,thetermscorrespondlngto= 0wl||
a|ways glvethelndlcla|equatlon
2 - i ; +- i } .,= 2,
:
+, .,=c
3. 3 Regul ar Si ngul ar Poi nts 227
Thetermscorrespondlngto= iyle|d
z,+i ; + ,+i ;- i } ., =,z
:
++z;., =0,
Becausethecoemclentz
:
++zof., ls nonzerowhether= or=-i .lt
fo||owsthat
., = ,z:;
lnelthercase.
Thecoefnclentofx
-
lnEq. ,z;ls
z,+; ,+- i ; .
-
+ ,+; .
-
- .
-
-
:
- .
-
=c
Weso|vefor.
-
andslmp|lfyto obtalntherecurrencere|atlon
.
-

:
.
-
=


z,+ ;
:
+,+ ; - i
forz ,z;
CASE 1 : , = Wenowwrlte.
-
lnp|aceof.
-
andsubstltute=lnEq.,z;
Thlsglvestherecurrencere|atlon
forz ,z:;
Wlththlsformu|awecandetermlnethe coemclents lnthenrstFrobenlusso|utlon
y, InvlewofEq. ,z:;weseethat.
-
=wheneverlsodd. Wlth=z, :,and:
lnEq. ,z:;,weget
.
:
.,
.

-

::

: i :

HencethenrstFrobenlusso|utlonls
and
y, ,x;=.,x
'
:
i+- +-++
x
:
x
-
x

i : : i : . ::
CASE 2:
:
= -i

Wenow wrlte /
-
ln p|ace of.
-
and substltute = i ln
Eq.,z; Thls glvestherecurrencere|atlon
/
_ /
-
-
:
-

z/
:
-
forz ,z:;
Agaln,Eq.,z:;lmp|lesthat/
-
=forodd.Wlth=z, :, and:ln,z:;, weget
/,
/
:
=

,
z
HencethesecondFrobenlusso|utlonls
y
:
,x; =/,x
-'
i+ +- +-+
x
:
x
-
x

z : z i :

228 Chapter 3 Power Seri es Methods


Exampl e
. .. . _. _ _ _ _ _ _ _______ __ H N ____ __ _ _ _ _ _ ____ ...... . ...
Flnd aFrobenlusso|utlonofBesse| ' sequatlonoforderzero,
(2S)
Sol ution Inthefomof, i :; , Eq. (2S)becomes

i
,

y +y + y = c
x x
:
Hencex=0 ls aregu|arslngu|arpolntwlth
;
,x;= iand,,x;=

soourserles
wl||convergefora||x > Because
;,= iand,,=, thelndlcla|equatlonls
, i;+=

=
Thusweobtalnon|ytheslng|eexponent=, andsotherel son|yoneFrobenlus
serlesso|utlon
O
y,x; =

ofEq. (2S), ltls lnfactapowerserles.


Thuswesubstltutey=_

ln(2S) , theresu|tl s
O O O

=

We comblnethenrsttwosumsandshlftthelndexofsummatlonlnthethlrdby -2
toobtaln
O O



Thetemconespondlngto

glves0 nolnfomatlon. Thetemcorrespondlng


tox

glves=,andthetemfor

yle|dstherecunencere|atlon

for2. (29)
Because=, weseethat
=0 wheneverlsodd. Substltutlng=2, 4,and
:lnEq. (29), weget
Evldent|y,thepatternls
The cholce i glves usoneofthe most lmportant specla| functlons lnmath-
ematlcs, the Bessel function of order zero of the frst kind, denoted by 1, ,x;
Thus
,;
Inthlsexamp|ewehavenotbeenab|etonndasecond|lnear|ylndependentso|utlon
ofBesse| ' sequatlonoforderzero.Wewl||derlvethatso|utlonlnSectlon:, ltwl||
notbeaFrobenlusserles.
Exampl e
Wen Tl - T
2
Is an Integer
3. 3 Regul ar Si ngul ar Poi nts 229
Recall that, lf, - r
:
ls a posltlve lnteger, thenTheorem I guarantees only the
exlstenceoftheFrobenlusserlessolutloncorrespondlngtothe larger exponentrl .
Example : lllustrates the fortunate case ln whlch the serles method nevertheless
yleldsasecondFrobenlusserlessolutlon. Thecaselnwhlchthesecondsolutlonls
notaFrobenlusserleswlllbedlscussedlnSectlonJ. 4.
FlndtheFrobenlusserlessolutlonsof
xy

+zy+xy=0. (J I )
Sol uti on Instandardformtheequatlonbecomes

z
,
x
:
) + y + y=.
x x
:
so we see that x = ls a regular slngular polntwlth
;
= zand

= 0. The
lndlclalequatlon
,- I ) +z= ,+ I ) =
has roots , =and
:
=-I , whlchdlfferby anlnteger. Inthls casewhen,

ls an lnteger, lt ls better to departfrom the standard procedure ofExample :and


beglnourworkwlththes-.//eexponent. As you wlllsee,therecurrencerelatlon
wlllthentell us whether ornotasecondFrobenlusserlessolutlonexlsts. Ifltdoes
exlst, ourcomputatlons wlll slmultaneouslyyleld/o/Frobenlus serles solutlons.
Ifthesecondsolutlondoesnotexlst,webeglnanewwlththelargerexponent=rl
toobtalntheoneFrobenlusserlessolutlonguaranteedbyTheorem I .
Hencewebeglnb y substltutlng
lnEq.(J I ) . Thlsglves
O O
y=x
-
,
.
-
x
-
=.
-
x
-
-
.
-

-
O O O
,- i ; ,- z;.
-
x
-
-
:
+z,- i ; .
-
x
-
-
:
+.
-
x
-
=0.
- - -
Wecomblnethenrst twosumsandshlftthelndexby -zlnthethlrdtoobtaln
O O
,- i ; .
-
x
-
-
:
+.
-

:
x
-
-
:
=0.
- -
Thecases=and= I reduceto
0
.
.,= and
.
., =0.
(J2)
Hencewehavemoarbltraryconstants.,and.,andthereforecanexpecttonnda
generalsolutlonlncorporatlngtwollnearlylndependentFrobenlusserlessolutlons.
If,for= I , wehadobtalnedanequatlonsuchas0 ., =J,whlchcanbesatlsned
forocholceof., .thl swouldhavetoldusthatnosecondFrobenlusserlessolutlon
couldexlst.
230 Chapter 3 Power Seri es Methods
"
".
FIGURE 3.3. 1. The solutions
cos x sin x
Yl (x) and Y
2
(X) = -
x x
in Example
Nowknowlngthata||lswe||, from,z;wereadtherecurrencere|atlon

,
for
Thenrstfewva|uesofglve


: :


: : :

evldent|ythepattemls

,z

i ;
for Therefore,agenera|so|utlonofEq.(J I ) l s
Thus
O
,(x) x
,

x
+

. . .

x z : x

,(x) cos x

smx) .
x
,;
Wehavethusfoundagenera|so|utlonexpressedasa|lnearcomblnatlonofthetwo
Frobenlusserlesso|utlons
cos x
,, (x)
x
sln x
and

.
x
,:;
AslndlcatedlnFlg. J . J . I , oneoftheseFrobenlusserlesso|utlonslsboundedbutthe
other lsunboundedneartheregu|arslngu|arpolntx 0acommonoccurrenceln
thecaseofexponentsdlfferlngbyanlnteger.
Summary
Whenconfrontedwlth a|lnearsecond-orderdlerentla|equatlon
A(x) ,

B(x) ,
'

C(x) ,
wlthana|ytlccoefnclentfunctlons,lnordertolnvestlgatetheposslb|eexlstenceof
serlesso|utlonswe nrstwrltetheequatlonln the standardform
,

P(x) ,'

Q(x) ,0.
IfP(x) and Q(x) arebothana|ytlcatx . thenx lsanordlnarypolnt,and
theequatlonhastwo|lnear|ylndependentpowerserlesso|utlons.
3. 3 Regul ar Si ngul ar Poi nts 231
Otherwlse, x =ls aslngu|arpolnt,andwe next wrltethedlerentla|equa-
tlon lntheform
__
+
p,x;_
+
,,x; _
=c
x x
:
Ifp,x; and,,x; areboth ana|ytlcatx =,thenx =ls aregu|arslngu|arpolnt.
Inthlscasewenndthetwoexponentsl| andl (assumedrea|,andwlthl l) by
so|vlngthelndlcla|equatlon
l (l - i ; +p,+,,=,
where;, = p,c;and,,= , ,c; Therea|waysexlstsaFrobenlusserlesso|utlon
_
= x _.
-
x
-
assoclated wlth the |arger exponent l| , and lfl| l ls notan
lnteger,theexlstenceofasecondFrobenlusserles so|utlony
:
=x

_/
-
x
-
lsa|so
guaranteed.
.............= 0 ....
..,..........,...........
,...........,.....,....
., ..,.....=
1. .+ . .

+ (sin x) y = 0
2. . .

+ = 0
3. .

+ (cos x) y' + xy = 0
4. .

+ .

+ ( 1 .

= 0
5. x( 1 + . + + .= 0
6. x
2
( 1 .

+ . = 0
7. x
2
y
"
+ (6 sin x) y' + 6y = 0
8. (6x
2
+ .

+ . + .

= 0
.= .= 0 .......,.........
..,...........= . ....
..., ..,.......= 0 ........,.
........,.....= ..
.....,....= ...,.....
.........,...., ..,.....
.....
9. ( 1 _ X) Y
"
. .

= 0
10. ( 1 .

+ . + = 0
11. ( 1 .

.+ = 0
12. .

+ .

+ .

= 0
13. (x
2
.+ . + .+ = 0
14. (x
2

+ (x
2
+ + (x
2
+ .= 0
15. .

(x
2
- . + .+ = 0
16. .

. + .+ + .= 0
......,............,
J ~ 0) ...., ..,......
...
17. ..+ + = 0
18. .+ - y = 0
19. . = 0
20. .+ + = 0
21. .

+ . ( 1 + .

= 0
22. .

+ . .

= 0
23. .

+ . (x
2
+ = 0
24. .

+ . + .

= 0
25. .( 1 . + = 0
26. .+ ( 1 .

..= 0
.......,.......,..
..........., ..,...
..... ........,.......
.,...~
27. .+ + .= 0
28. .+ ..= 0
29. .. .= 0
30. . + ..

= 0
31. ..

..+ ..

= 0
.............
.......,...........
32. .

+ ..+ .+ = 0
33. .

+ .

+ . .

( 1 + . = 0
34. .

+ (sin x) y' - (cos x) y = 0


35. Note that .= 0 is an irregular point of the equation
.

+ . + = .
(a) Show that = ._

can satisfy this equation


only if = (b) Substitute = _

to derive
the "formal" solution = _ .

What is the radius of


convergence of this series?
232 Chapter 3 Power Seri es Methods
36. (a) Suppose that .and B are nonzero constants. Show
that the equation .

+ . + B0 has at most one


solution of the form x
r
_

(b) Repeat part (a)


with the equation .

+ ..+ B (c) Show


that the equation .

+ ..

+ B0 has no Frobe
nius series solution. .....In each case substitute
x
r
_

in the given equation to determine the pos


sible values of
37. (a) Use the method of Frobenius to derive the solution
Y
x of the equation .

.+ (b) Verify
by substitution the second solution Y
2
.

Does Y
2
have a Frobenius series representation?
38. Apply the method of Frobenius to Bessel's equation of
order
.

+ .+

x
2
- 0,
to derive its general solution for x 0,
cos x sin x
. ,_ + _ .
Figure 3. 3. 2 shows the graphs of the two indicated solu
tions.
}
FIGURE 3.3.2. The solutions
cos x sin x
.
YI (x) _ and

._ In
Problem 38.
39. (a) Show that Bessel ' s equation of order 1 ,
.

+ .+ (x
2
- 0,
has exponents

1 and

-1 at x 0, and that the


Frobenius series corresponding to i s
x
-

J
I (x)
. .+


(b) Show that there is no Frobenius solution correspond
ing to the smaller exponent

-1 ; that is, show that it


is impossible to determine the coeffcients in
-

..

40. Consider the equation .

+ . + ( 1 - . (a)
Show that its exponents are . so it has complex-valued
Frobenius series solutions
-

-
and x-
i
_,

with .,,,1 . (b) Show that the recursion formula


is

o
n
2
+ .
Apply this formula with to obtain .

then with
to obtain ,

Conclude that .

and ,

are
complex conjugates: .

and ,

where the numbers .

and

are real . (c) Deduce


from part (b) that the diferential equation given in this
problem has real-valued solutions of the form
YI (x) A(x) cos (1n x) - B(x) sin(1n x) ,

.A(x) sin(1n x) + B(x) cos (1n x) ,


where .. _.

and B(x) _

41. Consider the diferential equation


x(x - l ) (x +

+2x (x - 3) (x + - 2(x l ) U
that appeared in an adverti sement for a symbolic algebra
program in the March 1 984 issue of the .....
.... (a) Show that x 0 is a regular
singular point with exponents

1 and

(b) It
follows from Theorem 1 that this diferential equation has
a power series solution of the form
Substitute this series (with 1 ) in the diferential equa
tion to show that

-2,

3, and

+ (.
2
- 5n

- (.
2
+ .+ .

.+ 1 ) (n + 2)
for . 2. (c) Use the recurrence relation in part (b)
to prove by induction that

.for ._ 1
Hence deduce (using the geometric series) that
for 0 x 1 .
.
YI
(x)
( 1 + .

42. This problem is a brief introduction to Gauss' s bypergeo


metric equation
x( 1 . + [y - .+ + . .0, (35)
where . and y are constants. This famous equation has
wide-ranging applications in mathematics and physics.
(a) Show that x 0 i s a regular singular point of Eq. (35),
with exponents 0 and 1 - y. (b) If y i s not zero or a neg
ative integer, it follows (why?) that Eq. (35) has a power
series solution
- -
. .
,
_

3. 4 Method of Frobeni us: The Excepti onal Cases 233


with cq = Show that the recurrence relation for thi s where a

a(a I ) (a . . . (a .- I ) for .I ,
and f
n
and

are defned similarly. (d) The series in (36)


i s known as the hypergeometric series and is commonly
denoted by F(a, x) . Show that
series is
(a . .
c
.
c

. .

for . (c) Conclude that with cq the series in


part (b) i s
.
I
,i) F( l , I , I , x) (the geometric series);
I - x
-
a

(36)
(ii) x F( -x)
I
n( l .
(iii) x F , -x
2
, tan-
1
x;
y(x) I
_

(iv) F(
-
k, ! , -x) .

(the binomial series).


Method of Frobenius: The Exceptional Cases
We contlnueourdlscusslonoftheequatlon
,,
+
p,x; ,
+
, ,x;
y

y -y =
X x
:
( I )
wherep,x; and,,x;areana|ytlcatx=. andx=lsaregu|arslngu|arpolnt. lf
theroots,and
:
ofthelndlcla|equatlon
,; =,

i ; +p,+,,= ,z;
do not dlerby an lnteger, then Theorem I ofSectlon guarantees thatEq. ( I )
hastwo|lnear|ylndependentFrobenlusserlesso|utlons. Weconsldernowthemore
comp|exsltuatlon ln whlch, -
:
ls an lnteger. lf, =
:
.then there ls on|yone
exponentaval|ab|e, and thus there can beon|y one Frobenlus serlesso|utlon. But
wesawlnExamp|e:ofSectlon thatlf, =
:
+x,wlth xaposltlvelnteger,
then lt ls posslb|e that a secondFrobenlus serles so|utlonexlsts. Wewl|| a|so see
that lt ls posslb|e that sucha so|utlon does not exlst. In fact, the second so|utlon
lnvo|veslnxwhenltlsnotaFrobenlusserles. Asyouwl||seelnExamp|esand4,
theseexceptlona|casesoccurlntheso|utlonofBesse| ' sequatlon.Forapp|lcatlons,
thls ls the most lmportant second-order |lnear dlfferentla| equatlon wlth varlab|e
coefnclents.
The Nonlogarithmic Case with r_ r

^
ln Sectlon we derlved the lndlcla| equatlon by substltutlng the power serles
p,x; =_;-
x
-
and,,x;=_,
-
x
-
andtheFrobenlusserles
O O
y,x; =x
.
-
x
-
=.
-
x
-

-
-
lnthedlfferentla|equatlonlntheform
x
:
y

+xp,x; y+,,x; y=0.


,.,= ; ,;
(4)
The resu|tofthlssubstltutlon,afterco||ectlonofthecoefnclentsof|lkepowersot
x, ls anequatlonoftheform
O
i
-
,;x
-
= ,;
-
234 Chapter 3 Power Seri es Methods
cXump| e 1
lnwhlchthecoemclentsdependon Itturns outthatthecoemclentofls
= - I)+

,:;
whlchglvesthe lndlcla| equatlonbecause= 0by assumptlon, a|so, for I ,
thecoefnclentof

hastheform
,:;
Here

ls a certaln |lnear comblnatlon of

A|though the exact


formu|als notnecessaryforourpurposes,lthappensthat

=
+

,s;

Becausea||thecoemclentsln,;mustvanlshfortheFrobenlusserlestobea
so|utlonofEq. (4),ltfo||owsthattheexponentandthecoemclents


mustsatlsfytheequatlon
+

= c ,;
Thls lsafor
lntermsof


Now supposethat=

+xwlthxaposltlvelnteger. Ifwe usethe|arger


exponentlnEq. ,;. thenthecoemclent+of
wl||benonzeroforevery
I because= 0 on|ywhen= andwhen=

< Once

havebeen determlned,wethereforecan so|veEq. ,;for


andcontlnue
tocomputesuccesslvecoefnclentslntheFrobenlusserlesso|utloncorrespondlngto
the exponent
But when we use the sma||er exponent

there ls a potentla| dlmcu|ty ln


computlng Forlnthls case

+x; = 0,soEq. ,;becomes


( I 0)
Atthl sstage

havea|ready beendetermlned. Iflthappensthat


then we canchoosearbltrarl|y andcontlnuetodetermlnetheremalnlngcoefn-
clentslnasecondFrobenlusserlesso|utlon. Butlflthappensthat
thenEq. ( I 0) ls notsatlsned wlth anycholceof lnthls casetherecannotexlsta
secondFrobenlusserlesso|utloncorrespondlngtothesma||erexponent

Exam-
p|es I and2l||ustratethesetwoposslbl|ltles.
Consldertheequatlon
( I I )
Here

= :and= 0, s othelndlcla|equatlonl s
=
- I ) +=

+= 0 ( I 2)
3. 4 Method of Frobeni us: The Excepti onal Cases 235
wlthroots= and

= therootsdlerbythelntegerx= We substltute
theFrobenlusserles = _

andget
O O
+ +

+:+


O O
++

0,

Whenwecomblnethenrsttwoanda|sothe|asttwosums, andlnthe|attershlftthe
lndexby theresu|tl s
O O
+

++

++

= 0.

The terms correspondlngto= glve the lndlcla| equatlonln, i z;.whereas for
iwegettheequatlon
+

+ +

++

= . , i ;
whlch ln thl s examp|e corresponds to the genera| so|utlon ln ,; Note that the
coefnclentof
ls.+
Wenowfo||owtherecommendatlonlnSectlon forthe case

+x
We beglnwlththesma||erroot

= Wlth

= Eq., i ;reducesto

= c
If= wecanso|vethlsequatlonfor
toobtalntherecurrencere|atlon
Thlsyle|ds
=

=
--

for=
and
, i 1;

, i :;
Inthecase=

+x.ltls a|waysthecoemclent
thatrequlres specla|consld-
eratlon. Herex = andfor= Eq. , i :;takestheformc

+= cHence

ls a secondarbltrary constant, and wecancomputeaddltlona|coefnclents, stl||


uslngtherecurslonformu|aln
andsoon.

=
- -,
:
, i :;
236 Chapter 3 Power Seri es Methods
Exampl e Z
When wecomblnetheresu|tsln and weget
O
y=

ln terms of the two arbltrary constants c, and

Thus we have found the two


Frobenlusserlesso|utlons
and
of Eq.

..
Determlnewhetherornottheequatlon

has two|lnear|ylndependentFrobenlusserlesso|utlons.
Sol ution Here
;
= and

= sothelndlcla|equatlonls
1= =

=0
wlth roots = : and

= dlerlng by x = On substltutlon of y
_

lnEq. weget
O O
+ +


O O
+

c

Ifwe shlftthelndexby lnthethlrd sum andcomblnetheotherthreesums,we
get
O O
+

=c

Thecoemclentofglvesthelndlcla|equatlon,andthecoemclentof

glves
+

+ =c
3. 4 Method of Frobeni us: The Excepti onal Cases 237
Becausethecoefnclentoflsnonzerobothfor andfor ltfo||ows
thatlneachcase. Forwegettheequatlon
+

+;
+


whlchcorrespondslnthlsexamp|etothegenera|equatlonln notethatthecoet-
nclentof

ls+;
We worknrstwlththesma||erroot

ThenEq. becomes

forFor= we canso|vefortherecurrencere|atlon


Becausethls formu|aglves


NowEq. wlthreducesto
=

and

0.


But = byassumptlon, and hence there ls noway t ochoose

sothat thls
equatlon ho|ds. Thus there ls Frobenlus serles so|utlon correspondlng to the
sma||erroot

To nndthe slng|eFrobenlus serles so|utloncorrespondlng to the |argerroot


wesubstltutelnEq. toobtalntherecurrencere|atlon
Thlsglves



Thegenera|pattemls

+
Thlsyle|dstheFrobenlusserlesso|utlon
of Eq.

238 Chapter 3 Power Seri es Methods


Reduction of Order
When on|y a slng|eFrobenlus serlesso|utlonexlsts, weneedan addltlona| tech-
nlque. We dlscuss here the method of reduction of order, whlch enab|es us to
useoneknownso|utlony, ofasecond-orderhomogeneous|lneardlfferentla|equa-
tlon to nnd a second |lnear|y lndependent so|utlon y
:
Conslderthe second-order
equatlon
y

+P(x) y' +Q(x) y ,z;


onanopenlnterva|ionwhlchP andQ arecontlnuous. Supposethatweknowone
so|utlony, ofEq. ,z; By TheoremzofSectlonz i . thereexlstsasecond|lnear|y
lndependentso|utlony
:
,ourprob|em ls to nndy
:
Equlva|ent|y, we wou|d|lketo
nndthequotlent
vex)
y
:
,x;
.
y, ,x;
Onceweknowv ex) , y
:
wl||thenbeglvenby
,z1;
,z;
We begln by substltutlng the expresslon ln ,z; ln Eq. ,z;. uslng the
derlvatlves
Weget
[vy' +2v'
y +:

y, ] +P [vy +:
y, ] +_:y, .
andrearrangementglves
[
II
r Q ]

r : y
,
+ y
,
+ y, +: y, + :y,
+ :y,

Butthebracketedexpresslonlnthls|astequatlonvanlshesbecausey, ls aso|utlon
ofEq.,z; Thls|eavestheequatlon
,z:;
The key to the success ofthls method ls thatEq. ,z:;ls //e.ln v'. Thus the
substltutlonln,z;hasreducedthesecond-order|lnearequatlonln,z;tothenrst-
order(lnv' ) |lnearequatlonln ,z:; Ifwewrlteu v' andassumethaty,(x) never
vanlshesoni ,then Eq. ,z:;yle|ds
u

+z +P(xu c ,z:;
AnlntegratlngfactorforEq.,z:;ls
thus
p(x) y exp,P(x) ax
cXump| e
3. 4 Method of Frobeni us: The Excepti onal Cases 239
Wenowlntegratetheequatlonln(27)toobtaln
u, expP(x) dx,= C, so u' = = exp- ,P(x) dx,
Anotherlntegratlonnowglves
)
:
_ _
,
exp _- ,P(x) dx,
- ~ u * C
:
dx+x
,, )
,
Wlth the partlcu|ar cholces C = i and x = we get the reduction-of-order
formula
,
exp _- ,P(x) dx,
)
:
= ,,
:
dx
)
,
(2)
Thls formu|a provldes a secondso|utlon)
:
(x) ofEq (23) onanylnterva| where
,,(x) lsneverzero Notethatbecauseanexponentla|functlonnevervanlshes,)
:
(x)
ls anonconstantmu|tlp|eof), (x) , so,, and)
:
are|lnear|ylndependentsolutlons
Forane|ementaryapp|lcatlonofthereductlon-of-orderformu|aconslderthedler-
entla|equatlon
x
:
,

- 9x,'
+25, =c
In Sectlon we mentloned that the equldlmenslonal equatlon x
:
,

+,x,' +
q,) = has thepowerfunctlon,(x) = x' asaso|utlonlfandon|y lflsarootof
thequadratlcequatlon
:
+(,- i ; +q, = cHere, = 9andq,25,soour
quadratlcequatlonls
:
- i c+25 = ,-5)
:
=andhastheslng|e(repeated)root
= 5. Thlsglvestheslng|epowerfunctlonso|utlon), (x) =x

ofourdlerentla|
equatlon
Beforewecan app|y thereductlon-of-orderformulato nndasecondso|utlon,
wemustnrstdlvldetheequatlonx
:
,

- 9x,'+25, = bylts|eadlngcoefncleat
x
:
togetthestandardform
//
9
/
25
, , +, =
x x
:
lnEq (23) wlth |eadlng coefnclent i Thus wehave P(x) = -9Jx and Q(x) =
25Jx
:
, sothereductlon-of-orderformu|aln(2)yle|dsthesecond|lnear|ylndepea-
dentso|utlon
)
:
(x) =x

,
(x)
:
exp- ,dx,dx
=x

,x
, a
exp(9 ln x) dx =x

,x
| a
x
-
dx =x

ln x
forx > c Thus our partlcu|arequldlmenslona|equatlon has thetwolndependent
so|utlons),(x) x

and)
:
(x) =x

|n xforx > c

Slml|arapp|lcatlonsofthereductlon-of-orderformu|acanbefoundlnProb-
|ems:-::ofSectlon2. 2wherewelntroducedthe method ofreductlonoforder
lnProb|em:(thoughwlthoutderlvlngtherethereductlon-of-orderformu|altse|f
240 Chapter 3 Power Seri es Methods
The Logarithmic Cases
Wenowlnvestlgatethegenera|formofthesecondso|utlonoftheequatlon

+

= .

, i
underthe assumptlon that lts exponents , and

= , - xdlerbythe lnteger
xWeassumethatwehavea|readyfoundtheFrobenlusserlesso|utlon
O
)j=

= ; ,z
for> 0 correspondlngtothe|arger exponent , Letus wrlte for
and for

ThuswecanrewrlteEq. , i ; lntheform

+
+=0
of Eq. ,z;
Becausethelndlcla|equatlonhasroots,and

= , - x. l t canbefactored
easl|y.

+ i ;+= , - , ; ,- ,+x;
=

+,x- z, ;+,- , x;=.


soweseethat
thatls,

- z, = -i - x
Inpreparatlonforuseofthereductlonoforderformu|aln,zs;. wewrlte
Then
sothat
+ +

= =


= exp |n

exp

.
,;
In the |ast step we have used the fact that a composltlon ofana|ytlc functlons ls
ana|ytlc andtherefore has a power serles representatlon, thelnltla| coefnclentof
thatserlesln , i ; ls ibecause

= i
We now substltute,z;and , i ; ln,zs; , wlth thecholce= iln,z;.thls
yle|ds
3. 4 Method of Frobeni us: The Exceptional Cases Z41
Weexpandthedenomlnatorandslmp|lfy.


,z;
(Herewehavesubstltuted,;andlndlcatedtheresu|tofcarrylngout|ongdlvlslon
ofserles as l||ustrated lnFlg. i i . notlng lnpartlcu|arthatthe constant term of
the quotlentserlesls i ; We now conslderseparate|y the casesand> 0.
Wewanttoascertalnthegenera| formof
wlthoutkeeplngtrackofspeclnccoef-
nclents.
CASE 1 : EQUAL EXPONENTS (rt = r2) ' Wlth. Eq. ,z;glves

, i


.
Consequent|y,lnthecaseofequa| exponents,thegenera|formof
ls
O

ln

Notethe|ogarlthmlcterm,ltls a|wayspresentwhen
,;
CASE 2: POSITIVE INTEGRAL DIFFERENCE (rt = r2 + N) . Wlth ~ .
Eq. ,z;glves
sothat

i
,

i
,

,:;
wherecThlsglvesthegenera|formof
lnthecaseofexponents
dlfferlngbyaposltlvelnteger. Notethecoefnclent
thatappearsln,:;butnot
ln,; Iflthappensthat .then there ls no|ogarlthmlcterm, lfso, Eq. , i ;
hasasecondFrobenlusserlesso|utlon(asl nExamp|e i ;
242 Chapter 3 Power Seri es Methods
InourderlvatlonofEqs. (33)and(34 )whlchexhlbltthegenera|formofthe
secondso|utlon ln the cases, = r
:
and , r2 = x > 0, respectlve|ywehave
saldnothlngabouttheradllofconvergenceofthevarlouspowerserlesthatappear
Theorem i (next) ls asummatlonoftheprecedlngdlscusslonanda|sote||swhere
theserlesln(33)and(34)converge. AslnTheorem iofSectlon3. 3, werestrlctour
attentlontoso|utlonsfor> 0,
THEOREM 1 The Exceptional Cases
Supposethat= 0l saregu|arslngu|arpolntoftheequatlon

, () y= c (4)
Letp > 0denotethemlnlmumofmeradllofconvergenceofthepowerserles
O O
=

and =


Let,andr
:
betheroots,wlth,

,ofthelndlcla|equatlon
,- 1 ) ;,,,= c
(a) If, =

,thenEq. ,:hastwoso|utlonsy, and)


:
oftheforms
and
O

= 0)
O

=
y,|n x

(35a)
(35b)
(b) If, r2 = x,aposltlvelnteger,thenEq.(4)hastwoso|utlonsy,andy
:
of
theforms
and
O
y, ,x; =

= 0)
O

= Cy, () |n

(36a)
(36b)
InEq. (36b) , b, = 0 but C maybeeltherzeroor nonzero, so the |oganthmlc
termmay ormaynotactua||ybepresentlnthlscase. Theradllofconvergenceof
thepowerserlesofthlstheoremarea||at|eastp. Thecoefnclentslntheseserles
(andtheconstantC lnEq.(36b) maybedetermlnedbydlrectsubstltutlonofthe
serleslnthedlerentla|equatlonln(4).
Exampl e 4
3. 4 Method of Frobeni us: The Excepti onal Cases 243
Wewl||l||ustratethecase=

byderlvlngthesecondso|utlonofBesse| ' sequa-


tlon oforderzero,
,:;
forwhlch=

=c InExamp|eofSectlon wefoundthenrstso|utloa
,s;
AccordlngtoEq. ,o;thesecondso|utlonwl||havetheform
O

=
|n +

,;

The nrst twoderlvatlvesof

are
O
_

n +

and
WesubstltutetheselnEq. ,:;andusethefactthat a|sosatlsnesthlsequatlon
toobtaln
c

+
andltfo||owsthat
O O O
+

+ +

Theon|ytermlnvo|vlnglnEq. ,:;ls so= c But

=lf
ls odd, andltfo||owsthata||thecoefnclentsofoddsubscrlptln

vanlsh
Now weexamlnethe coefnclents wltheven subscrlpts lnEq ,:; Flrstwe
seethat
=

:
Forwereadtherecurrencere|atlon

244 Chapter 3 Power Seri es Methods


from ,:; Note the 'nonhomogeneousterm (notlnvo|vlng the unknown coefn-
clents)ontherlght-hand slde ln,:z; Suchnonhomogeneousrecurrencere|atlons
are typlca| ofthe exceptlona| cases ofthemethodofFrobenlus, andthelrso|utlon
oftenrequlresabltoflngenulty. Theusua|strategydependsondetectlngthemost
consplcuousdependenceof

onWenotethepresenceof

ontherlght-
hand slde ln ,:z; , lnconunctlon wlth the coefnclent (z)

on the |e-hand slde,


wearelnduced tothlnkof

as so-e//,dlvldedby z

( )

. Notlng a|so the


a|ternatlonofslgn,wemakethesubstltutlon
,:;
lntheexpectatlonthattherecurrencere|atlonfor

wl||be slmp|erthantheone
for
We chose

ratherthan ( i ) because

> , wlth iln


,:;. weget

i Substltutlonof,:;ln,:z;glves
whlch bol|sdowntotheextreme|yslmp|erecurrencere|atlon
Thus
andsoon. Evldent|y,
i

i i

i+
_
i i i

+ _ i+
_
+ _ ,
i i i i

+ i+
_
+ _ +

i i i

where wedenoteby

thenthpartla| sumoftheharmonlcserles_( i J)
,::;
Flna||y,keeplnglnmlnd thatthecoemclentsofoddsubscrlptarea||zero,we
substltute,:;and,::;ln,;toobtalnthesecondso|utlon
O

= |n x +

= + - -+
.
: i zs i sz:
,:;
ofBesse| ' sequatlonoforderzero. Thepowerserlesln,:;convergesfora||The
mostcommon|yused|lnear|ylndependentof secondso|utlonls
z z
-(y |n z) y+

T T
cXump| e
3. 4 Method of Frobeni us: The Excepti onal Cases 245
thatl s,

where y denotes
y |lm ,u
-
ln

- -
Thls partlcu|arcomblnatlon ls chosenbecauseoflts nlce behavlor as
ltlsca||edtheBcssclluncti0n0l0rdcrzcr0 0lthc scc0ndkind.
As ana|ternatlveto the method ofsubstltutlon, we l|lustratethe case

x
by emp|oylng the technlque ofreductlon of order to derlve a second so|utlon ot
Besse| ' sequatlonoforder

the assoclated lndlcla| equatlon has roots and

Prob|emofSectlon

oneso|utlonof Eq. ls

Accordlng to
Thus
Wlth fromthereductlonoforderformu|alnylelds

bylong
dlvlslon

246 Chapter 3 Power Seri es Methods


:
h|

Notethatthetechnlqueofreductlonoforderreadl|yyle|dsthenrstsevera|termsof
theserles, but doesnotprovldearecurrencere|atlonthatcanbe usedtodetermlne
thegenera|termoftheserles.
Wlthacomputatlonslml|artothatshownlnExamp|e:(butmorecomp|lcated
seeProb|emzi ;.themethodofsubstltutloncanbeusedtoderlvetheso|utlon
where

lsdennedln,::;forn i , = cThereadercanverlfythattheterms
shownln Eq. (50)agreewlth
,z;
Themostcommon|yused|lnear|ylndependentof1, } so|utlonofBesse| ' sequatlon
oforder ilsthecomblnatlon
= ln

___
_
I

, - i ; T

n . n .

Examp|es:and
_
l||ustratetwomethodsofnndlngtheso|utlonlnthe|ogarlth-
mlccaseslrectsubstltutlonandreductlonoforder. Athlrda|tematlvelsout|lned
ln Prob|em i
. , , , . ,, . . -... , ,,, . , ,,--, , , , ..... , , . .... ,, , ... ..... . ,, , . ... ....... ............... . .. . ... .... ..
........,,......,
.......,............
.....................,...
..............
12. .


13. .

.=
14. .

. . .=
15. Begin with
1. . .
2. . .
3. ..
4. .. =
5. . . .
6. . .
7. .

..

=
8. . . =
...... .........
.............., ..,..
.............,.......,
...............
.........,......
9. . .=
10. .

. .

=
11. .

. .
.

, . = -

. . .
Using the method of reduction of order, derive the second
linearly independent solution
.

., .ln x
_


of Bessel ' s equation of order zero.
16. Find two linearly independent Frobenius series solutions
of Bessel ' s equation of order ,
17. (a) Verify that

.= .i s one solution of
.

. .
3. 4 Method of Frobeni us: The Excepti onal Cases 247
(b) Note that =

= Substitute
C

= ln .
_

in the difrential equation to deduce that


that
and

=
-
for .
.
(c) Substitute

. i n this recurrence relation and


conclude from the result that

Thus the second


solution is
C
.

.= .. .
_

.
18. Consider the equation . = which has exponents
= and

at .= (a) Derive the Frobenius


series solution
(b) Substitute
C

ln .
_

in the equation .
relation
to derive the recurrence
.
..

=
. .
Conclude from this result that a second solution is

.. . .

. .
19. Suppose that the diferential equation
. .

.,. , . = (54)
has equal exponents =

at the regular singular point


. so that its indicial equation is
Let = and defne

for .by using Eq. (9) ;


that is,

o
..
Then defne the function . of .and to be
C
. =
_

(55)

(a) Deduce from the discussion preceding Eq. (9) that


(57)
Hence deduce that
C

= . =
_

58

i s one solution of Eq. (54). (b) Diferentiate Eq. (57)


with respect to to show that
Deduce that

. is a second solution of
Eq. (54) . (c) Diferentiate Eq. 58with respect to to
show that
C

ln . .

_
.

20. Use the method of Problem to derive both the solu


tions in (38) and (45) of Bessel ' s equation of order zero.
The following steps outline this computation. (a) Take
= show that Eq. (55) reduces in this case to

for .

(b) Next show that = = and then deduce


from that

= = for .odd. Hence you


need to compute

and only for .even. (c) De


duce from that

.

.

With = = in 58 this gives . (d) Diferenti


ate to show that

Substitution of this result in (59) gives the second solution


in (45).
21. Derive the logarithmic solution i n of Bessel
'
s equa
tion of order by the method of substitution. The follow
ing steps outline this computation. (a) Substitute
in Bessel ' s equation to obtain
C
.
_
.

. .

248 Chapter 3 Power Seri es Methods


(b) Deduce from Eq. that = -1 and that

= 0
for .odd. (c) Next deduce the recurrence relation

.1 )
.

. .

for .1
.
Note that if

is chosen arbitrarily, then

is
determined for all . 1 . (d) Take

= and substitute

. .
Bessel' s Eqation
in Eq.

to obtain
1 1


. .
(e) Note that

= 1 =

and deduce that

Wehave a|ready seensevera| casesofBesse| ' sequatlonoforder .


, i ;
Its so|utlons are now ca||ed Besse| functlonsoforder . Suchfunctlons nrstap-
peared ln the I 730s ln the workofDanle| Bemou||l and Eu|eron the oscl||atlons
ofa vertlca||y suspended chaln. The equatlon ltse|fappears ln a i :::artlc|eby
Eu|eronthevlbratlonsofaclrcu|ardrumhead,andFourlerusedBesse|functlonsln
hlsc|asslca| treatlseonheat( I 22). Butthelrgenera| propertles werenrststudled
systematlca||y ln an i sz: memolr by the German astronomer and mathematlclan
FrledrlchW.Besse|, i :s:-i s::;, whowaslnvestlgatlngthemotlonofp|anets.The
standardsourceoflnformatlononBesse|functlonslsG. N. Watson' s~
2nded. (Cambrldge. CambrldgeUnlverslty Press,
i ::; Its : pages ofreferences, whlch cover on|y the perlod up to i zz.glve
some ldeaofthevast|lteratureofthl ssubect.
Besse| ' sequatlonln, i ;haslndlcla|equatlonr
:
-
:
=,wlthroots=+p
Ifwe substltutey = _

lnEq. , i ;.wennd lntheusua|mannerthat=


andthat
(2)
form 2. TheverlncatlonofEq. (2) ls |etothereader(Prob|em:;
The Case r = ~
If we use = and wrlte ln p|ace ofthen Eq. (2) yle|ds the recurslon
formu|a

.
m(2+m)
,;
Because=, ltfo||owsthat=0 fora||oddva|uesofm. Thenrstfew even
coemclentsare

-
2(2+2)

-
2
:
(+ i ;

+
= -
:,zp+:;
=

2
+
.

)(

)
,

= - = -
6(2+:; 2
- .
2
.
3 (+ I ) ( +2) (+;
3. 5 Bessel ' s Equati on 249
Thegenera|patternls

.

.

sowlththe|argerrootr =wegettheso|utlon
O

y, ,x;=

. .
(4)
If
; = thl s ls the on|y Frobenlus serlesso|utlon, wlth = as we||, ltlsthe
functlon wehave seenbefore.
The Case r = *
Ifweuse r =andwrlteln p|aceofEq. takesthefom
..

= ,;
for. whereas= We seethattherel sapotentla|dlmcu|tylflthappens
thatlsaposltlvelntegerthatls, lflseltheraposltlvelntegeroranoddposltlve
lntegra|mu|tlp|eof . Forthenwhen.= Eq.,;lsslmp|y

0,
Thuslf

=.thennova|ue ofcansatlsfythl sequatlon.


But lfl s an odd posltlve lntegra| mu|tlp|e of , we can clrcumvent thls
dlfncu|ty. For supposethat= kwherek ls an oddposltlvelnteger. Thenwe
needon|ychoose=fora||oddva|uesof.Thecrucla|step lsthekthstep,
k(k

=,
andthlsequatlonwl||ho|dbecause

=
Hencelfl soaposltlvelnteger, wetake=for.oddanddennethe
coefnclentsofevensubscrlptln temsofbymeansoftherecurslonfomu|a

=- ,
..
. ,:;
Incomparlng,:;wlth,;.weseethat,:;wl|||eadtothesameresu|tasthatln(4),
exceptwlthrep|acedwlth Thus lnthl scaseweobtalnthesecond so|utlon
O

y
:
,x;= ,:;

. .
Theserlesln(4) and ,:;converge fora||> because=ls the on|y slngu|ar
polntofBesse| ' sequatlon. If> .thenthe|eadlngtermlny, ls whereas
the|eadlngtermlny
:
ls

Hence y,,;=. buty


:
+>as ,so
lt ls c|ear that y, andy
:
are |lnear|ylndependentso|utlonsofBesse| ' sequatlonof
order>
250 Chapter 3 Power Seri es Methods
The Gamma Function
Theformu|aslnand(7)canbeslmp|lnedbyuseofthegamma function
whlchlsdennedfor> by
=

c
-

a
Itl snotdlmcu|ttoshowthatthl slmproperlntegra|convergesforeach> c The
gammafunctlon ls a genera|lzatlonfor> ofthe factorla| functlon n' , whlch
ls denned on|y lf ls a nonnegatlvelnteger. To see the way ln whlch lsa
genera|lzatlonofn' , wenote nrstthat
!

-
= c
-
a = |lm
.
c
-
| =

-- .
Then welntegratebypartswlth= andac = c
-a
thatls,
+=
Thl sl sthemostlmportantpropertyofthegammafunctlon.
If wecomblneEqs. ,;and weseethat
=

= =

= =

=
andlngenera|that
+ = foranlnteger.
Anlmportantspecla|va|ueofthegammafunctlonl s
wherewehavesubstltuted

forlnthenrstlntegra| ,thefactthat
,;



ls known,butlsfarfromobvlous. See,forlnstance,Examp|elnSectlon of
Edwards andPenney, c././s i./yms.cac./s.7thedltlon(UpperSadd|e
Rlver,NJ. PrentlceHa||,
A|though l s denned ln on|y for > we canusethe recurslon
formu|aln todenne wheneverlsneltherzeronoranegatlvelnteger. lf
-I < < then
=
+

FIGURE 3.5. 1. The graph of the


extended gamma function.
3. 5 Bessel ' s Equati on 251
therlght-handsldels dennedbecause< x + <

Thesameformu|amaythen
beusedtoextendthedennltlonof|(x) tothe open lnterva| ,-z, then to the
openlnterva|, -,-z;. and soon. The graphof thegammafunctlonthusextended
ls shownlnFlg.

Thestudentwho wou|d|lketo pursue thlsfascinatlngtoplc


furthershou|dconsu|tArtln' s/co.--.i./o(NewYork. Ho|t, Rlnehartand
Wlnston,

In on|y39pages,thlsl soneofthe nnestexposltlonsln theentlre


|lteratureofmathematlcs.
Bessel Functions of the First Kind
Ifwechoose.,= i } z

|,p+ lnwherep> .andnotethat


|,p+-+ i ; = ,p+-; ,p+-
,p+z ,p+ i ; |,p+
byrepeated app|lcatlonofEq. wecan wrltethe Bessel function of the frst
kind of order pveryconclse|ywlththealdofthegammafunctlon.
O

,x=
- |,p+-+

, i
Slml|ar|y, lfp > ls notanlnteger, wechoose/, = i }z
-
|, -p+ ln ,:to
obtalnthe|lnear|ylndependentsecondso|utlon

ofBesse| ' sequatlonoforderp lfplsnotanlnteger,wehavethegenera|so|utlon

forx > ,x

mustberep|acedwlth x
lnEqs. , i ;through . stogetthecorrect
so|utlonsforx <
Ifp= . anonnegatlvelnteger, then Eq. , i ;glves

,x- _ _

-,-+ z
fortheBesse|functlonsofthe nrst klndoflntegra| order. Thus
and

Thegraphs of J
a
(x) and J,(x) are shownlnFlg. 3. 5. 2. lna general way they re-
semb|edampedcoslneandslneoscl||atlons,respectlve|y(seeProb|emz:;lndeed,
lfyou examlne the serles ln you can see part of the reason why J
a
(x) and
cosx-/,/beslml|aron|ymlnorchangeslnthedenomlnatorsln, i :;areneeded
to produce theTay|orserlesforcos x. AssuggestedbyFlg.

thezerosofthe
252 Chapter 3 Power Seri es Methods
}
-:aZero
(- + g -
of / (x)
1 2. 4048 2. 3562 3. 83 1 7 3. 9270
2 5. 5201 5. 4978 7. 01 56 7. 0686
X 3 8. 6537 8. 6394 1 0. 1 735 1 0. 21 02
4 1 1 . 79 1 5 1 1 . 78 1 0 1 3. 3237 1 3. 35 1 8
5 1 4. 9309 1 4. 9226 1 6. 4706 1 6.4934
FIGURE 3.5.2. The graphs of the Bessel
functions

.and

.
FIGURE 3.5.3. Zeros of

.and

.
functlons 1, ,x; and 1

,x; are /nter/acedbetween any two consecutlve zeros ot


1, ,x; there ls preclse|y one zero of 1

(x) (Prob|em z:;and vlce versa. The nrst


four zeros of 1, ,x; are approxlmate|y z :c:s. zci . s ::. and i i :i For
n |arge, the nth zero of 1, ,x; l sapproxlmate|y n - , , the nth zero of 1

,x;
ls approxlmate|y n + _. Thusthelnterva|betweenconsecutlve zeros ofelther
J, (x) or 1

(x) ls approxlmate|y another slml|arlty wlth cos x and sln x. You


can seethe way the accuracy oftheseapproxlmatlonslncreaseswlth lncreaslngn
byroundlngtheentrleslnthetab|e lnFlg. totwodeclma|p|aces.
Ittums outthat 1 _ (x) ls ane|ementaryfunctlonlftheorder ls ha|fanodd
lnteger. Forlnstance,onsubstltutlonof = and = - ln Eqs. , i ;and, i :,
respectlve|y,theresu|tscanberecognlzed(Prob|emz;as

:
,x; =

smx
x
and 1

:
,x; =
z
cos x.
x
Bessel Functions of the Second Kind
, i
The methods ofSectlon :mustbe usedto nnd |lnear|y lndependent second so-
|utlons oflntegra| order. A very comp|lcatedgenera|lzatlonofExamp|e ln that
sectlonglvestheformu|a
z x i
-

,
2
,

:-

n - m - i ;
}
. (x;=

y +ln _
. (x;

- -

-=a
wlththenotatlonusedthere. Ifn = cthen the nrst sum ln,zc;ls takentobezero.
Here, }
.
(x) ls ca||ed the Bessel function of the second kind of integral order
n
Thegenera|so|utlonofBesse| ' sequatlonoflntegra|ordern l s
,zi
Itl slmportanttonotethat r
-
(x) as x c(Flg. :; Hence.
:
= cln
Eq. ,zi ; lf,(x) ls contlnuous atx = c Thus lf,(x) ls a cont/nuous so|utlonof
Besse| ' sequatlonofordern, ltfo||owsthat
}
3. 5 Bessel ' s Equati on 253
for some constant. Because 1

,c; we see lnaddltlonthatlf c.then


cy,c; InSectlon
we wl||see thatthlsslng|efactregardlngBesse|functlons
hasnumerousphyslca|app|lcatlons.
Flgure l||ustratesthefactthatfor > thegraphsof

and

|ookgenera||y|lkethoseof

andI, Inpartlcu|ar,

,c;cwhl|eI

as- c

,andbothfunctlonsundergo dampedoscl||atlonas
}
FIGUR 3.5.4. The graphs of the Bessel functions
Yo (x) and Y1 (x) .
FIGUR 3.5.5. The graphs of the Bessel functions
}(x) and Y
2
(x) .
Bessel Function Identities
Besse|functlonsareana|ogoustotrlgonometrlcfunctlonslnthattheysatlsfya|arge
numberofstandardldentltlesoffrequentutl|lty,especla||ylntheeva|uatlonoflnte-
gra|slnvo|vlngBesse|functlons. Dlrentlatlonof
O


_. . :
lnthecasethatls anonnegatlvelntegerglves

O

. .
O

. .
O

. .

andthuswehaveshownthat
Slml|ar|y,

IfwecarryoutthedlerentlatlonslnEqs.andandthendlvldetheresu|tlng
ldentltlesbyandrespectlve|y, weobtaln(Prob|ems;theldentltles

254 Chapter 3 Power Seri es Methods


cXump| e 1
cXump| e Z
cXump| e
and
,:
ThuswemayexpressthederlvatlvesofBesse|functlonslntermsofBesse|functlons
themse|ves. SubtractlonofEq. ,z;fromEq. ,z:;glvestherecurslonformu|a
,::
whlch can be used to express Besse| functlons of hlgh orderln terms ofBessel
functlonsof|owerorders. Intheform
,::
ltcanbeusedtoexpressBesse|functlonsof|argenegatlveorderlntermsofBessel
functlonsofnumerlca||ysma||ernegatlveorders.
The ldentltleslnEqs. ,zzthrough ,z:;ho|dwherevertheyaremeanlngful
thatls, whenevernoBesse|functlonsofnegatlve lntegra|orderappear. lnpartlcular,
theyho|dfor a|lnonlntegra|va|uesof.
Wlth = c. Eq. ,zz;glves
,x1, ,x; ax=x1, ,x;+c
Slml|ar|y,wlth = c. Eq.,z;glves
. ..
,1,,xax= -1, ,x; +c
Uslngnrst =zandthen = .lnEq. ,z:;. weget
sothat
1, ,x; = -1, ,x;+- . ,1,,x;

Wlth slml|armanlpu|atlonsevery Besse|functlonofposltlvelntegral ordercan be


expressedln termsof1, ,x;and 1,,x;
Toantldlerentlatex) ,x; , we nrst note that
,x
,
),x; ax= -x
,
1,,x;+c
byEq ,z;wlth = i Wethereforewrlte
3. 5 Bessel ' s Equati on 255
andlntegratebypartswlth
a = zx ax,
Thlsglves
and
ac=

ax,
c=


, = z, = C,
wlththealdofthesecondresu|tofExamp|e i
The Parametric Bessel Equation
Theparametric Bessel equation of order ls

,zs
wherel saposltlveparameter. Aswewl||seelnChapter. thlsequatlonappearsln
theso|utlonofLap|ace' sequatlonlnpo|arcoordlnates. Itlseasytosee(Prob|em
thatthesubstltutlon=transformsEq.,zs;lntothe(standard)Besse|equation

- + ,

=c
a

a
,:
wlth genera| so|utlon y, ; =

Hence the genera| solutlonof


Eq. ,zs;l s
=


Nowconsldertheelgenva|ueprob|em

=c,
,i=c
,
, i
onthe lnterva| c. i} We seek thepos//.cva|ues offor whlch there exlsts a
nontrlvla| so|utlonof, i ; that ls .o/oson c, i} Ifwe wrlte = .

then
thedlerentla|equatlonln, i ; lsthatlnEq. ,zs;. solts gener so|utlonlsglvenln
Eq. ,c;Because

- -as- cbut

lsnnlte,thecontlnultyof
requlresthat

= c Thus =

Theendpolntcondltlon y,i =
now lmp|lesthat =imustbea(posltlve)rootoftheequatlon

=c ,:
For > i .

oscl||atesrather|lke lnFlg. zandhence hasan lnnnlte


sequenceofposltlvezeros

. . . (seeFlg. : ltfo||owsthatthekth
posltlveelgenva|ueoftheprob|emln, i ; ls

,;
andthatltsassoclatedelgenfunctlonl s
,:
The roots

of Eq. ,:;for s and / zcare tabu|ated ln Tab|e of


M.AbramowltzandI. A.Stegun,u.a/oo/o)M./c-./../i./o-s(NewYork.
Dover, i :;
256 Chapter 3 Power Seri es Methods
}
FIGUR 3.5.6. The positive zeros
Yn h Y

Yn3
, . . . of the
Bessel function

.

h
'

~ . . ... . ... . . .... . ,.. .. . . . . . . _ _ ..... ... .,_ ___ _ . _ ._ .____ _ _ _ ___ ____ _ _ __ ~ ~
1. Diferentiate termwise the series for , . ) t o show directly
that .= .(another analogy with the cosine and
sine functions).
2. (a) Deduce from Eqs. and that
, ,

.
.

(b) Use the result of part (a) to verify the formulas in


Eq. for

.and

. and construct a fgure


showing the graphs of these functions.
3. (a) Suppose that is a positive integer. Show that

(b) Conclude from part (a) and Eq. that

.=
.

, ]

4. Apply Eqs. and to show that


and

.=

(SIl. .cos x)
.

.=

(cos . . .. .
.
Construct a fgure showing the graphs of these two func
tions.
S. Express

.in terms of

.and .
6. Derive the recursion formula i n Eq. for Bessel ' s
equation.
7. Verify the identity in by termwise diferentiation.
8. Deduce the identities in Eqs. .and from those in
Eqs. and
9. Verify that the substitution = ..transforms the para
metric Bessel equation in into the equation in
10. Show that
..=

.
1 1 . Use the relation . = .. to deduce from
Eqs. and .that if ,is not a negative integer, then

.=
.

.
,

, ,

,

This form is more convenient for the computation of

.
because only the single value , of the gamma func
tion is required.
12. Use the series of Problem to fnd = lim . if

.
Use a computer algebra system to graph . for .near
Does the graph corroborate your value of
.......

.........
....................
, ... ..........,...
..... , .........
......... ......,....
..............
13. .

... 14. .

, ...
15. .

, . .. 16. . . ..
..

. .. 18. .

. ..
19. .

. .. 20. .. ..
.

...
22. Prove that

,.) = cos (x sin .


,
by showing that the right-hand side satisfes Bessel ' s equa
tion of order zero and has the value , when .=
Explain why this constitutes a proof.
23. Prove that

.= cos(e - .sin e) de

_
by showing that the right-hand side satisfes Bessel ' s equa
tion of order and that its derivative has the value
when .= Explain why this constitutes a proof.
24. It can be shown that

(.) = cos(ne - x sin e) de.

_
With . show that the right-hand side satisfes
Bessel ' s equation of order .and also agrees with the val
ues

and Explain why this does .sufce to


prove the preceding assertion.
25. Deduce from Problem that

_
(.) =

c
| s| a 9
de.

_
.....Show frst that

_
_ c
| s| a 9
de =
_
_

c
|x s|a 9
+ c
|x s|a 9 ,
de;
then use Euler' s formula. )
26. Use Eqs. and and Rolle' s theorem to prove that
between any two consecutive zeros of

.there i s pre
cisely one zero of

. Use a computer algebra system


to construct a fgure illustrating this fact with .= (for
instance) .
27. (a) Show that the substitution = .

in Bessel ' s
equation of order ,
Applications of Bessel Functions
3. 6 Appl i cati ons of Bessel Functi ons 257
yields
+ -
,

j =
(b) If .is so large that

is negligible, then
the latter equation reduces to + Explain why this
suggests (without proving it) that if . is a solution of
Bessel ' s equation, then
. .

.cos x + .sin x)
= .

cos(x - o)
with and o constants, and .large.

Asymptotic Approximations It is known that the choices


= and o = (.+ .in yield the best ap
proximation to

.for .large:
:, - cos

.+

Similarly,
r, - sin

.+

In particular,

_
(.) cos

. -

,
\
and
r, - sin

,
\
if .i s large. These are ..,.,,....in that the
ratio of the two sides in each approximation approaches unity
as . +0.
ThelmportanceofBesse|functlonsstemsnoton|yfromthefrequentappearanceof
Besse| ' sequatlon ln app|lcatlons, but a|sofrom the fact thatthe so|utlonsofmany
othersecond-order|lneardlerentla|equatlonscanbe expressedln termsofBessel
functlons. To seehowthlscomesabout,webeglnwlthBesse| ' sequatlonoforder]
lntheform
( I )
andsubstltute
U = x
-
y, z = /x" ,:
Thenaroutlnebutsomewhattedloustransformatlon(Prob|em I 4)ofEq. ( I ) yle|ds
258 Chapter 3 Power Seri es Methods
Exampl e 1
thatls,
,;
wheretheconstants s.c. and,areglvenby
= i- z:, s=a

, c=;

. and , z; ,:
Itls aslmp|emattertoso|vetheequatlonsln,:;for
i -
a =
z

/=
z,
,
,
and
,
;=
2`
,, i -

- :s
p =
,
,;
Undertheassumptlonthatthesquarerootsln,;arerea| ,ltfo||owsthatthegenera|
so|utlonofEq. ,;l s
where
=

(assumlngthatpls notanlnteger) ls the genera| so|utlonoftheBesse|equatlonln


, i ; Thlsestab|lshesthefo||owlngresu|t.
THEOREM 1 Sol utions in Bessel Functions
Ifc > c., ,c. and,i-

:s.thenthegenera| so|utlon(forx > c;of


Eq.,;l s
,:;
where O, ;, /.and areglvenbytheequatlonsln(5). Ifpl sanlnger, then
J, lstoberep|acedwlth I, .

So|vetheequatlon
:x

+sxy+ ,x
-
- ; y= (7)
Sol ution To compareEq.,:;wlthEq. ,;.werewrltetheformeras
x

+zxy+
,- + x
-
)) =c
and see that = z. s - .c =

.and, : Then theequatlonsln,;glve


a=-t , ;=z./=

.andp= t . Thusthegenera| so|utlonln ,:;of Eq. (7)l s


y,x; =x
-
,

., 1,

)+.

1,

) ]
Ifwereca||fromEq. , i ;ofSectlon3. 5that

= sln z and

=

weseethatagenera|so|utlonofEq.,:;canbewrlttenlnthee|ementaryform
y,x; =x
-

+ssln

Exampl e Z So|vetheAlryequatlon
3. 6 Appl i cati ons of Bessel Functi ons 259

o
+x =c ,s;
Sol ution Flrstwerewrltetheglvenequatlonlntheform
x - L
FIGUR 3.6. 1. The buckling
column.
Thls lsthespecla|caseofEq. ,;wlth ~ = s = c,c = . andg = Itfo||ows
fromthe equatlons ln,;thata = , f = , k = z. and = Thusthegenera|
so|utlonofEq. ,s;ls

Bucking of a Vertical Column


Fora practlca| app|lcatlon, we now consldertheprob|emofdetermlnlng whena
unlform vertlca| co|umn wl|| buck|e under lts own welght (after, perhaps, belng
nudged|atera||yusta bltbyapasslngbreeze). Wetakex = catthefreetopend
oftheco|umnandx = L > catlts bottom, weassumethatthe bottom ls rlgld|y
lmbedded lnthe ground, perhaps lnconcrete, see Flg. : i Denote the angu|ar
deectlon ofthe co|umn at the polntx by 0 (x) . From the theory ofe|astlclty lt
fo||owsthat
d
z
J
ii
:
+gpx0 =c,
dx
,;
where i ls the Young' s modu|us of the materla| of the co|umn, ls lts cross-
sectlona|momentoflnertla, p l sthe|lneardensltyoftheco|umn,andg lsgravlta-
tlona|acce|eratlon. Forphyslca|reasonsnobendlngatthefreetopoftheco|umn
andnodeectlonatltslmbeddedbottomtheboundarycondltlonsare
,c;=c, 0(L) = c , i c;
Wewl||accept,;and, i c;asanapproprlatestatementoftheprob|emandattempt
toso|veltlnthlsform. Wlth

:

gp
ii

wehavetheelgenva|ueprob|em
+ y
z
x0 = c, ,c;= c, 0(L) = c
, i i ;
, i z;
Theco|umncan buck|eon|ylfthere ls anontrlvla| so|utlonof, i z; , otherwlsethe
co|umnwl||remalnln lts undeectedvertlca|posltlon.
Thedlerentla|equatlonln, i z;lsanAlryequatlonslml|artotheonelnEx-
amp|e z Ithas theformofEq. ,;wlth ~ = s = c,c = y
z
, andg = The
equatlonsln,;glvea= , f = , k = y, and = Sothegenera|so|utlonls
, i ;
260 Chapter 3 Power Seri es Methods
. .
FIGURE 3.6.2. The graph of
'-l j3
(
Z
) .
,
:
lnordertoapp|ythelnltla|condltlons,wesubstltutep ln
O
(

, ;
-

z-

'

_- ,
, -
,
:
andnnd aersomeslmp|lncatlonsthat
c
z

. /
)

y
z
x
)
y
|
x
,


y
,
/
)
,

:

i sc

Fromthls ltls c|earthattheendpolntcondltlon,c; clmp|lesthatc, c, so


( l 4)
Theendpolntcondltlon ,i; cnowglves
( l ;
Thusthe co|umn wl||buck|eon|ylfz
J,
/

(z) = cThegraphof
yi
)
/
z
ls a root ofthe equatlon
( l :;
(seeProb|emofSectlon ; lsshownlnFlg : z,whereweseethatthesma||est
posltlvezeroz , lsablt|essthanzMosttechnlca|computlngsystemscannndroots
|lkethlsone Forlnstance,eachofthecomputersystemcommands
fso1ve ( Besse1J ( - 1/3 , x ) =O , x , 1 4 2 )
FindRoot [ BeSse1J [ - 1 / 3 , x ] ==O , { x , 2 } ]
fzero ( ' besse1j ( - 1/3 , x ) ' , 2 )
,M.p|c;
,M./c./..;
,Mt.s,
yle|dtheva|uez , i s::(roundedaccuratetonvedeclma|p|aces)
Theshortest|engthi,forwhlchtheco|umnwl||buck|eunderltsownwelght
ls
i
'

lfwesubstltutez, i s::andp oA, whereolsthevo|umetrlcdensltyofthe


materla|oftheco|umnandA ls lts cross-sectlona| area, wenna||yget
LI
. /
)
i
,
, i s:;
goA
, i :;
3. 6 Appl i cati ons of Bessel Functi ons 261
forthecrltlca|buck|lng|ength. Forexamp|e,wlthastee|co|umnorrodforwhlch
L = z s Z i c

lbJln.
:
andgo = c zslbJln.

, theformu|aln, i :;glves theresu|ts


shownlnthetab|elnFlg. :
Circular with in.
Circular with in.
bh0rlcslBucklingLcnglhLl
Annular with gg
er
in. and rout
er
i n.
ft in.
ft in.
ft in.
FIGURE 3.6.3.
We haveusedthe faml|larformu|as ~ = -`and i = -'for aclrcu|ar
dlsk. Thedatalnthetab|e show whyagpo|esareho||ow.
Problems
......,..........
..., ..,............
1. .

. .


2. .+ + .
3. . .

4. .

. + .
5. .

. .


6. .

..+ ...


7. .

. + .


8. ..

. + .
9. .

...


10. .

. . .


11. .

12. ..

]
13. Apply Theorem to show that the general solution of
..
is . .

.cos x .sin x) .
14. Verify that the substitutions i n i n Bessel ' s equation
(Eq. yield Eq.
15. (a) Show that the substitution
..

...
transforms the Riccati equation ....

into
..

. (b) Show that the general solution of


....

is
.....Apply the identities in Eqs. and of
Section
16. (a) Substitute the series of Problem of Section in
the result of Problem here to show that the solution of
the initial value problem
is
.
..
.

,
. .
,

(b) Deduce similarly that the solution of the initial value


problem
is
.
.

,
..

,
. .

, , , ,

Some solution curves of the equation ....

are
shown in Fig. . The location of the asymptotes where
. can be found by using Newton' s method to
fnd the zeros of the denominators in the formulas for the
solutions as listed here.

Z
- I
- Z
\
.
FIGURE 3.6.4. Solution curves of @
.

..
262 Chapter 3 Power Seri es Methods
17. Figure shows a linearly tapered rod with circular
cross section, subject to an axial force of compression.
As in Section its defection curve = . satisfes
the endpoint value problem
= . = = .
:
x
x - a x - o
FIG URE 3.6.5. The tapered rod of Problem
Here, however, the moment of inertia = .of the
cross section at .is given by
. =

..

,
where

= the value of at .= Substitution of


.in the diferential equation in yields the eigen
value problem
.

AY = e. = =
where A =

(a) Apply the theo


rem of this section to show that the general solution of
.

= is
. = ..cos .sin ,
(b) Conclude that the nth eigenvalue is given by

...where .= .is the length of the rod, and


hence that the nth buckling force is
. | COl O
Note that i f .= this result reduces to Eq. of Sec
tion
18. Consider a variable-length pendulum as indicated i n
Fig. Assume that its length is increasing linearly
with time, . = . It can be shown that the oscil
lations of this pendulum satisfy the diferential equation
.. .e =
under the usual condition that e is so small that sin d
is very well approximated by e: e sin e. Substitute
.= .+ to derive the general solution
For the application of this solution to a discussion of the
steadily descending pendulum ("its nether extremity was
formed of a crescent of glittering steel, about a foot in
length from hor to hor; the hors upward, and the under
edge as keen as that of a razor . . . and the whole hissed
as it swung through the air . . . down and still down it
came") of Edgar Allan Poe' s macabre classic "The Pit and
the Pendulum," see the article by Borrelli, Coleman, and
Hobson in the March issue of .......
.(Vol. pp.
FIGURE 3.6.6. A variable-length pendulum.
ARlccatlequatlonlsoneofthefom
ay
= ~,x; ,+ s,x; y+c,x;
ax
ManyRlccatlequatlons|lketheones|lstednextcanbeso|vedexp|lclt|ylntemsof
Besse|functlons.
, i ;
,z;
3. 6 Appl i cati ons of Bessel Functi ons 263
ay
ax
= , - x

,
ay
ax
= x + ,,
ay
- = x - ,,
ax
ay

- =y - x
ax
(3)
(4)
(5)
(6)
Forexamp|e,Prob|em ilnthlssectlonsaysthatthegenera|so|utlonof Eq. . ls
glvenby
(7)
Seewhetherthesymbo|lcDEso|vercommandlnyourcomputera|gebrasys-
tem,suchastheM.p/ccommand
dsolve ( diff ( y( x ) , x ) xA2 + y ( x ) A 2 , y ( x
ortheM./c-./..command
DSolve [ y ' [ x)
xA2 + y [ x) A 2 , y [ x ) , x )
agrees wlth Eq. ,:; lfBesse| mnctlons other than those appearlng ln Eq. ,:;are
lnvo|ved, you may need to app|y the ldentltles ln (26) and (27) ofSectlon 3. 5to
transformthe computer' sanswerto (7) . Then see whetheryoursystemcantake
the |lmltasx -- cln (7)to show that thearbltraryconstant.ls glvenln terms ot
thelnltla| va|uey,c;by
y,o; |
. = -
z|
,s;
Now you shou|dbe ab|etouse bul|t-ln Besse| functlons top|ottyplca| so|utlon
curves|lkethoseshownlnFlg. 3. 6. 4.
Next,lnvestlgateslml|ar|yoneoftheotherequatlonsln(2)through(6). Each
has agenera| so|utlonofthe samegenera| form ln (7)aquotlentof|lnearcom-
blnatlonsofBesse|functlons. lnaddltlonto 1, ,x;and r, (x) , theseso|utlonsnay
lnvo|vethe-oa;cascssc/)./os
and
P
x, ,x;=
_
/
-
1, ,/ x; +r, ,/ x;]
thatsatlsfythe-oa;cascssc/c,./o
x

+xy- ,x

+p

; y=c
oforderp Forlnstance,thegenera| so|utlonofEq. ,;ls glvenforx >cby
i

) i

x - c

x
y,x; = x

i,

- .i
,

,;
264 Chapter 3 Power Seri es Methods
4
Z
0
-Z
-4
-a
X
FIGURE 3.6.7. Solution curves
.
of A

..
where
y,o; |,
. =- .

|
( I 0)
Flgure 3. 6. 7showssometyplca|so|utloncurves,togetherwlththeparabo|a,=x
that appears to bear an lnterestlng re|atlon to Eq. (6)we see a funne| neary =
+andaspoutneary=-
TheBesse|functlons wlth lmaglnary argumentthatappear ln thedennltlons
ofI _ (x) and x_ (x) may|ookexotlc, butthepowerserlesofthemodlnedfunctlon
I
.
(x) ls slmp|ythatoftheunmodlnedfunctlonJ
.
(x) exceptwlthoutthea|tematlng
mlnusslgns. Forlnstance,
and
x

x
-
x

i, (x) = I +- +- ++

4 64 2304
x x

I
.
(x) =
2
+
I 6
+
3S4
+
I S432
+

Checkthesepowerserlesexpanslonsuslngyourcomputera|gebrasystem|ookat
BesselI ln eltherM.p/corMathemat/ca~andcomparethem wlthEqs. ( I 7) and
( I S) lnSectlon3. 5.
The second-order dlfferentla| equatlons ofthe form y = )(x . y;wlth the
samerlght-hand sldes as ln Eqs. ( I ) through (6) have lnterestlng so|utlons whlch,
however, cannot be expressed ln terms of e|ementary functlons andJor known
specla|functlons such as Besse|functlons. Neverthe|ess, they canbelnvestlgated
uslngan ODEp|otter. Forlnstance,thelnterestlngpattemlnFlg. 3. 6. Sshowsso|u-
tloncurvesofthesecond-orderequatlon
y=y

-x ( I I )
wlththesamelnltla|va|uey,c; =0butdlfferents|opesy ,c; = -3. 3, -3. I , . . . ,
0. 7. Equatlon( I I ) lsaformof thefsi.//c.cms.ca.,anequatlonthatarose
hlstorlca||ylnthec|asslncatlonofnon|lnearsecond-order dlerentla| equatlonsln
terms ofthelr crltlca| polnts (see Chapter I4 ofE. L. Ince, oa/.n_ cc/./
i,./os.New York. Dover Pub|lcatlons, I 956). Flgure 3. 6. Swas suggestedby
anartlc|ebyAnneNoonburgcontalnlngaslml|arngurelntheSprlng I 993lssueof
thec- oni- i xcs/cc
~
~
~
~
~
~
~
Z

-
-
~ -
~
}
\
.
-
-
-
\
FIGURE 3.6.8. The frst Painleve transcendant = .

A,
= = 3e 3,3 , . .
3. 6 Appl i cati ons of Bessel Functi ons 265
Flna||y, here' sare|atedexamp|ethatwaslnsplredbyaM.p/cdemonstratlon
package. TheM.p/cdsolve commandyle|dsthegenera| so|utlon
y,x; =x
-
,
,., 1,

,x;+.

r,

,x
+x
-
, ,
, i s:::cc+ i :c:ccx

+sc::ccx
-
+:ccx

+ i ccx

+x
, ,
; , i z;
ofthenonhomogeneoussecond-orderequatlon
x

+xy+ ,x

- ; y=x , i ;
ShowthatTheorem i ln thls sectlonexp|alnstheBesse|partofthea||egedso|u-
tlon lnEq. , i z; Canyouexp|alnwheretheratlona|functlonpartcomesnom, or
at |eastverlfylt! Forfurtherexamp|es ofthls sort, youcanrep|acethecoemclent
lnEq. , i ;wlth

- i . wherelsan c.clnteger,and/orrep|acethexonthe
rlght-handsldewlthx

.wheres ls anoaalnteger. (Wlth parltles other thanthese,


moreexotlc specla|functlonsarelnvo|ved. )
lapace1ranslorm
Nethods

Laplace Transforms and Inverse Transforms


f(t)
_
D{ f(t ) f' (t)
f(t)
_
:{ f(t ) l(a)
FIGURE 4.1. 1. Transformation
of a function: 1 in analogy with
.
266
_
n Chapter 2 wesaw that|lnear dlerentla| equatlons wlthconstant coelhclents
havenumerousapp|lcatlonsandcanbeso|vedsystematlca||y.Therearecommoa
sltuatlons,however,lnwhlchthea|ternatlvemethodsofthlschapterareprelerab|e.
Forexamp|e,reca||thedlerentla|equatlons
-x+.x+/x= i, ; and tt+ t+ t = i ,;
c
correspondlngto amasssprlngdashpot systemanda serlestcclrcult, respec
tlve|y. It often happens ln practlce that the forclng term, i, ; or i , ; . has
dlscontlnultlesfor examp|e, when the vo|tage supp|ledto an e|ectrlca| clrcult ls
turnedoffandonperlodlca||y. In thls case the methods ofChapter2 canbequlte
awkward,andtheLap|acetransformmethod lsmoreconvenlent.
Thedlfferentlatlonoperatorncanbevlewedasatransformatlonwhlch,whea
app|ledtothefunctlon), ; . yle|dsthenewfunctlonn, ), ; |= ) ,; TheLap|ace
transformatlon1 lnvo|ves theoperatlonoflntegratlonandyle|dsthenewfunctloa
1, ), ; | = i,s; ofanewlndependentvarlab|es Thesltuatlonls dlagrammedla
Flg. 4. I . I . After|earnlnglnthlssectlonhowtocomputetheLap|acetransformi(s)
ofa functlon ), ; . we wl|| see ln Sectlon4. 2that theLap|acetransform converts
aa_ cc/./equatlon lnthe unknown functlon ),; lntoan ./,c/m/.equatlonln
i,s) Because a|gebralc equatlons are genera||y easler to so|ve than dlerentla|
equatlons, thls ls one method that slmp|lnes the prob|em ofnndlng the so|utlon
), ;

cXump| e 1
cXump| e2
4. 1 Lapl ace Transforms and I nverse Transforms 267
DEFI NITI ON The Lapl ace Transform
Glven a function ), ; dennedfor a|| 0, theip/..cms)oof)isthe
functionidennedasfo||ows.
i,s; = 1, ), ; , = `c

),; a
fora||va|uesofsforwhlchtheimproperlntegra|converges.
, i ;
Reca||thatanimproper integral overanlnnnltelnterva|l s dennedasa|lmlt
oflntegra|soverboundedlnterva|s , thatls,

,,; a= |lm
-
,,;a

--

(2)
Ifthe|lmltln(2)exlsts,thenwesaythatthelmproperlntegra|converges; otherwlse,
ltdiverges orfal|sto exlst Notethatthelntegrandofthelmproperlntegral ln, i ;
contalnstheparametersl naddltlontothevarlab|eoflntegratlon Therelore,when
thelntegra|ln, i ; converges,ltconvergesnotmere|ytoanumber,butto+]./o
iofs Asln thefo||owlngexamp|es,ltls typlca|forthe lmproper lntegral lnthe
dennltlonof1, ),; |toconvergeforsomeva|uesofsanddlvergeforothers
Wlth),; = ifor 0,thedennltlonoftheLap|acetransformln, i ; glves
andtherefore

, i |

! 0 or s >
s
(3)
As ln (3), lt' sgoodpractlce to speclfy thedomaln oftheLap|ace transformln
prob|emsaswe||aslnexamp|es. A|so,ln thls computatlonwehaveusedthecom-
mon abbrevlatlon
,,;

= |lm
,,;
-

--

(4)

Remark: The|lmltwecomputedln Examp|e i wou|dnotexlstlfs < 0,


forthen , i }s; c

wou|dbecomeunboundedas/-- +c. Hence1, i |lsdehned


on|yfors > 0. Thlslstyplca|ofLap|acetransforms, thedomalnofatransformls
norma||yoftheforms >.forsomenumber.
Wlth ),; = c

for 0, weobtaln

- -

,
|
-
1, c

|= c

a = c
- -

a =

o
s- .
,

268 Chapter 4 Lapl ace Transform Methods


Exampl e
Ifs- a > c.thenc
-

- cas- +c, soltfo||owsthat

, c

i

or s >a.
s - a
(5)
Notehere thatthelmproperlntegra| glvlng1, c

, dlverges lfs a. It ls worth


notlnga|sothattheformu|a ln (5)ho|dslfa ls acomp|ex number. Forthen, wlth
a =a+/;,
c
- -

=c
,

c
- -
-

- c
as- +c, provldedthats >a=Re[a] , reca|| thatc
,
=cos; +/sln;
The Lap|ace transform 1,

, ofa powerfunctlon ls most convenlent|y ex-


pressedlnterms ofthegamma function |,x; . whlchls dennedforx > cbythe
formu|a
(6)
Forane|ementarydlscusslonof|,x; . seethe subsectlononthegammafunctlonln
Sectlon3. 5, whereltls shownthat
|, i ; = i (7)
andthat
|,x+ i ; =x|,x; (S)
forx >cItthenfo||owsthatlfl saposltlvelnteger,then
|,+ i ; =|,;
=

, - i ; |,- i ;
=

,- i ;

,- z; |,- z;
=,- i ; ,- z;

z |,z;
=,- i ; ,- z;

z i

|, i ; ,
thus
|,+ i ; = (9)
lf lsa posltlve lnteger. Therefore, the functlon |,x+ i ; . whlch ls dennedand
contlnuousfora||x > -I g agrees wlththefactorla|functlonforx = . aposltlve
integer.

Supposethat],;=

wherea ls rea|anda > -i Then


1,

,=

a
Ifwesubstltute=s . =}s.anda=a}slnthlslntegra|,weget
1 = - c = ,

|
i
,

~N
a
|,.+ i ;
s
. _
s

, i c;
Exampl e 4
4. 1 Lapl ace Transforms and I nverse Transforms 269
fora||s > 0(sothat = s > 0). Because |,+ I) = lflsaaoaaegatlve
lateger,weseethat
Forl astance,
I
1, |=
'
s

z
1, |= _,
s
)
6
aad 1, |= _.
s
( I I )
As l aProb|ems I aadz. theseformu|ascaabederlvedlmmedlate|yfromtheden-
nltloa, wlthouttheuseofthegammafuactlon.

Linearit of Transforms
It ls aot aecessary forus to proceed muchfurther la the computatloa ofLaplace
traasforms dlrect|y fromthe denaltloa. Oacewe kaow the Lap|acetraasforms of
severa|functlons,wecaacomblnethemtoobtalatransformsofotherfunctloas. The
reasoalsthattheLap|acetransformatloals a//c.operatloa.
THEOREM 1 Linearity of the Lapl ace Transform
Ifcand/areconstaats,then
1,c), ; +/,,; ,=.1, ), ; , +/1, ,, ; | , i z;
fora||ssuchthattheLap|acetransforms ofthefunctlons)aad,bothexlst.
TheproofofTheorem I fo||owslmmedlate|yfromthe|laearltyoftheopera-
tloasoftaklng|lmltsandoflategratloa.
1,.),;+/,, ; | =

c
-
.),;+/,,;} a
= |lm c
-
.),;+/,,;} a
C=O

=. |lm c
-
),; a+/ |lm c
-
,,;a
C=O

C~O

=.1, ), ; | +/1, ,,; |


Thecomputatlonof1,
-

|lsbasedontheknownspecla|va|ue
ofthegammafuactlon. Forlnstaace,ltfo||owsthat
( I 3)
270 Chapter 4 Lapl ace Transform Methods
Exampl e
Exampl e
uslagtheformu|a|(x +I ) =x|(x) ln (9), nrst wlthx = andthen wlthx = |
Nowtheformu|asl n( I 0)through( I 2)yle|d
z :|, ) :
1, )

+:

|=)

_
+

=

+
_

s
Reca|| that cosh/ =
togetherglve
thatl s,
Slml|ar|y,
1, .aa / |=

1,e

,+-1,c

|= - -+ ,
I I I I I
z z z s - / s + /
s
1,.aa / |=

fors >/>c
s - /

/
1, ma / |=

for s >/>c
s - /
Becausecos/=,e

+e
-

;}z,theformu|aln,;(wlth.=/ /;yle|ds
I I I I zs
1cos/ |=

-
s

/
+
s

/
=

aadthus
s
1, .a / |=

for s >c
s + /
(Thedomalafo||owsfroms >n- / /} =0. )Slml|ar|y,
/
1, m/ |=

for s >c
s +

, i:;
, i ;
, i :;
( I 7)

App|ylag|laearlty,theformu|ala , i :;. aadafaml|lartrlgonometrlcldentlty,weget


1, )c

+ z sln

) |=1, )c

+i - .a : |
I s
=
s- z s s

+:
)s

+i ::s- 7z
= for s >c
s ,s- z; ,s
:
+:;

Exampl e
Jl/J
(s
s
(s
(s
t
a
(a
r(a +
(s
s

q
j
(s
s - a
cos kt
s
S
2
+ k
2
(s
sin kt
k
s
2
+ k
2
(s
cosh kt
S
S
2
- k
2
(s I kl )
sinh kt
k
S
2
- k
2
(s I kl )
u(t - a)
_s
(s
s
FIGUR 4. 1.2. A short table of
Laplace transforms.
4. 1 Lapl ace Transforms and I nverse Transforms 271
Inverse Transforms
Accordlng to Theorem 3 ofthls sectloa, ao two dlereat fuactloasthatareboth
contlnuousfora|| t 0 can havethesameLap|acetraasform. Thus lfi,s;ls the
transformofsomecoatlnuousfuactlon), ) , then),; lsunlque|ydetermlaed. Thls
observatlona||owsustomakethefo||owlagdenaltloa. Ifi,s; = 1, ), ; , , theawe
ca||),; theinverse Laplace transform ofi,s; aadwrlte
),;= 1
-
,
, i,s; ,
UslngtheLap|acetraasformsderlvedl nExamp|es2, 3, aad5we seethat
aadsooa.
1

.



=
_-
z

s + z
:
( I S)

NOTATION: FUNCTIONS AND THEIR TRANSFORMS. Throughoutthlschapter


we denotefunctlonsoft by|owercase|etters. Thetraasformofafuactloa wl|| a|-
waysbedeaotedbythatsame|ettercaplta|lzed. Thus i,s; lstheLap|acetraasform
of),; andx(t ) ls the lnverseLap|acetransformofx,s;
A tab|eof Lap|ace transforms serves a purpose slml|ar to thatof a tab|e of
lntegra|s. Thetab|e ln Flg. 4. I . 2 |lststhe transforms derlved la thls sectloa, maay
addltloaa|transforms caa bederlvedfromthesefew, uslagvarlousgeaera|proper-
tlesoftheLap|acetransformatlon(whlchwewl||dlscussla subsequeatsectloas).
Piecewise Continuous Functions
Asweremarkedatthebeglnalngofthl ssectlon,weaeedtobeab|etohaad|ecertala
typesofdlscontlnuousfuactloas. Thefunctloa)(t)ls saldtobepiecewise contin
uous ontheboundedlaterva|. t /provldedthat../}caabesubdlvlded lato
nalte|ymanyabuttlng sublnterva|sln suchawaythat
1. )lscontl auousla thelnterlorofeachofthese sublnterva|s , aad
Z. ),; has annlte|lmltast approacheseachendpolatofeachsublaterva|from
ltslnterlor.
We say that ) ls plecewlsecoatlauousfort 0 lflt ls plecewlse contlauous oa
everyboundedsublaterva| of[0,+~; Thus aplecewlsecoatlauousfuactloahas
on|y slmp|edlscoatlnultles (lfany) andoa|yat lso|atedpolats. Atsuchpolatsthe
va|ueofthefunctloaexperlencesanalteump,aslndlcatedlnFlg. 4. I . 3. Thejump
in ),; at the point .lsdenaedtobe),.+;- ),.-; , where
),.+;= |lm ),. + -; aad ),.-; = |lm ),. - -;
.a- .a-
Perhaps theslmp|estplecewlsecoatlnuous(butdlscoatlauous)fuactloalsthe
unit step function, whosegraph appears la Flg. 4. I . 4. Itlsdenaedasfo||ows.
0 for t < 0,
a(t ) =
I fort 0.
( I 9)
272 Chapter 4 Lapl ace Transform Methods
:
-
X
o
'

'

'

| ,


FIGURE 4.1.3. The graph of a
piecewise continuous function; the
solid dots indicate values of the
function at discontinuities.
FIGURE 4.1.4. The graph of the
unit step function.
FIGURE 4. 1.5. The unit step
function .

has a jump at = u.
cXump| Cb
Because ,; = l for aad becausethe Lap|ace traasform lavolvesoa|ythe
va|uesofafuactloafor.weseelmmedlate|ythat
l
1, , ; , =

(s >;
s
,z;
Thegraphoftheualtstepfuactloa

,;= ,- .;appearsl aFlg. 4. l . 5. Its ump


occursat=.ratherthaaat=, equlva|eat|y,

for < ..

,;= ,- .; =
l for.
Flad1,

, ; , lf. >o
(2l )
Sol ution We beglawlththedenaltloaoftheLap|acetraasform. We obtala
1,

, ; , = e
-

,; a = e
-

a = llm
e
,
!

-
-
o

-- S
j

coasequeat|y,
e
-

1,

, ; , = (s > . . >;
s
General Properties of Transforms
Itlsafaml|larfactfromca|culusthatthel ategra|
!
-
,,; a
,zz;

exlsts lf , ls plecewlse coatlauous oa the bouaded laterva| .,/} Heace lf)ls


plecewlsecoatlauousfor . ltfo||owsthatthelategra|
exlsts for a|| / < +0
.
But la orderfor F(s)the llmlt ofthls last lategra| as
/- +ctoexlst,weaeedsomecoadltloato |lmlttherateofgrowthof),;as
4. 1 Lapl ace Transforms and I nverse Transforms 273
t -- +c. Thefunctloa ] l ssaldtobeof exponential order ast -- +lfthere
exl staoaaegatlvecoastaats x,.,aad suchthat
](t) x-

fort (23)
Thus afuactlonlsofexpoaeatla|orderprovldedthatltgrowsaomorerapldly(as
t -- +) than a coastant mu|tlp|e of some expoaeatla| fuactloa wlth a |laear
expoaeat. The partlcu|ar va|ues of x, ., aad are aot so lmportaat. What ls
lmportaatlsthatso-esuchva|uesexlstsothatthecoadltloala(23)lssatlsned.
Thecoadltlonl a(23)mere|ysaysthat),}-

|lesbetweea -MaadMaad
lsthereforeboundedlnva|uefort sufncleat|y|arge. Iapartlcu|ar,thlslstrue(wlth
. = c;lf](t) ltse|flsbouaded. Thusevery/oaeafuactlonsuchascosktor
slaktlsofexpoaeatla|order.
If(t ) ls apo|ynomla|, thea thefaml|larfactthat ;, -

-- cast -- +
lmp|les that (23) ho|ds (for sumcleat|y |arge) wlth x = . = i Thus every
po/yo-/./functloalsofexponeatla|order.
For aa examp|e ofaa e|emeatary functloa that ls coatlauous aad therefore
bouadedoa every (nalte)laterva|,butneverthe|essls aot ofexpoaeatla|order,coa-
slderthefuactloa](t ) = -

.
= exp(t
:
) . Whatevertheva|ueof.. weseethat
_
](t )
_
-

.
_
.

rm = rm = rm -
.
= +c
-
-

-
-

-
because t
:
. -- +cas t -- +c. Heacethe coadltloa la (23)caaaotho|d
for any (nalte) va|ue x, so we coac|ude that the fuactlon ](t ) = -

.
ls oof
exponeatla|order.
Slml|ar|y,because-

.
-- +cast -- +c, weseethatthelmproperlate-
gra|]
-
-

.
athatwou|ddenael -

.
does aot exlst(foraays;, aadtherefore
thatthefuactloa-

.
doesohaveaLap|acetraasform. Thefo||owlagtheoremguar-
aateesthatplecewlsefunctloasofexpoaentla|orderaohaveLap|acetraasforms.
THEOREM Z Existence of Lapl ace Transforms
Ifthemnctlon]l splecewlsecontlnuousfort caadl sofexponentla|orderas
t -- +c, thenlts Lap|acetransform i,s)= l ](t ) exlsts. Morepreclse|y,lf
)lsplecewlsecontlnuousandsatlsnesthecondltlonla (23), then i,s;exlstsfor
aII s >.
Proof: Flrst we aote that wecaatake = c la (23). Forbyplecewlse
coatlnulty, ](t ) lsbouadedoa c.} Iacreaslagxl a(23) lfaecessary, wecaa
thereforeassumethat ] (t) xlfct Because-

ifort c.ltthea
fo||owsthat ](t ) x-

fora||t o
Byastandardtheoremoaconvergeaceoflmproperlategra|sthefactthatab-
so|uteconvergeacelmp|lesconvergeaceltsufncesforustoprovethatthelategra|

](t ) a
exlstsfors >. To dothl s, ltsumcesl atumtoshowthattheva|ueofthelategral
274 Chapter 4 Lapl ace Transform Methods
remalnsboundedas/- +c. Butthefactthat ),; Me
,

fora|| 0lmp|les
that
M e

-
,

a=
!

M
lfs > . ThlsprovesTheorem2.
o
s- .
Wehaveshown, moreover,that
!

M
i,s; e
-
),; a
o
s - .
lfs >. Whenwetake|lmltsass- +c, wegetthefo||owlngresu|t.
COROLLARY F(s) for s Large
If),;satlsnesthehypothesesofTheorem2, then
|lm i,s; = o
s-O
(24)
(25)
Thecondltlonln(25)severe|y|lmltsthefunctlonsthatcanbeLap|acetrans-
forms. Forlnstance, thefunctlono,s;=sJ,s+I ) cannotbetheLap|acetransform
ofany reasonab|efunctlonbecauselts |lmltass - +c ls I , notc Moregen-
era||y, aratlona| functlona quotlentoftwo po|ynomla|scanbe (and ls, as we
sha|| see)aLap|acetransformon|ylfthedegreeofltsnumeratorls|essthanthatof
ltsdenomlnator.
Ontheotherhand,thehypothesesofTheorem 2aresufnclent,butnotneces-
sary, condltlonsforexlstenceoftheLap|acetransform of), ; Forexamp|e, the
functlon ),; IJfal|stobeplecewlsecontlnuous (at = 0), butneverthe|ess
(Examp|e 3 wlth.= - > -I ) ltsLap|acetransform
1,
-
.

:
,=
|
' :
=

s
,

bothexlstsandvlo|atesthecondltlonln(24),whlchwou|dlmp|ythats i,s;remalns
boundedass- +c.
The remalnderofthls chapterl sdevoted |arge|yt otechnlques for so|vlnga
dlerentla|equatlonby nrstnndlngtheLap|acetransformoflts so|utlon. Itls then
vlta| for
_
s to know that thls unlque|y determlnes the so|utlon ofthe dlerentla|
equatlon, thatl s, thatthefunctlonofswehavefoundhason|y onelnverseLap|ace
transform that cou|d be the deslred so|utlon. The fo||owlng theorem ls proved ln
Chapter6 ofChurchl| | ' sopem/o./M./e-./.s, 3rd ed. (New York. McGraw-
Hl||, I 972).
THEOREM 3 Uni queness of I nverse Lapl ace Transforms
Supposethatthefunctlons ),; d,,; satlsfy thehypothesesofTheorem 2,
sothatthelrLap|acetransfoms i,s; ando,s; bothexlst. Ifi,s; = o,s; for
a|| s > .(forsome.;, then), ; = ,, ; whereveron [0, +c)both)and,are
contlnuous.
4. 1 Lapl ace Transforms and I nverse Transforms 275
Thustwoplecewlsecontlnuousfunctlonsofexponentla|orderwlththesame
Lap|ace transform candler on|y at thelr lsolatedpolnts ofdlscontlnulty. Thls ls
ofnolmportancelnmostpractlca| app|lcatlons, sowemayregardlnverseLap|ace
transformsasbelngessentla||yunlque. In partlcu|ar, twosolutlonsofadlerentla|
equatlonmustbothbecontlnuous, andhencemustbethesameso|utlonlftheyhave
thesameLap|acetransform.
Historical Remark: Lap|ace transforms have an lnterestlng hlstory.
The lntegra| ln the dennltlon ofthe Lap|ace transform probab|y appeared nrst la
theworkofEu|er. Itlscustomary lnmathematlcstonameatechnlqueortheorem
for the next person afterEu|erto dlscoverlt (e|se there wou|dbe severa| hundred
dlerent examp|es of Eu|er' s theorem) . In thls case, the next person was the
FrenchmathematlclanPlerreSlmondeLap|ace( I 749I S27), who emp|oyedsuch
lntegra|s lnhl s workonprobabl|ltytheory. The so-cal|ed operatlona| methodsfor
so|vlngdlerentla|equatlons,whlchare basedon Lap|acetransforms, werenot ex-
p|olted by Lap|ace. Indeed, they were dlscovered and popu|arlzed by practlclng
englneersnotab|y theEng|lshe|ectrlca|englneerO|lverHeavlslde( I S50I 925).
These technlques were successfu||y and wlde|y app|ledbefore they hadbeen rlg-
orous|yustlned, and around the beglnnlng ofthe twentleth century thelrva|ldlty
wasthe subect ofconslderab|econtroversy. One reason ls thatHeavlsldeb|lthe|y
assumedtheexlstenceoffunctlonswhoseLap|acetransformscontradlctthecondl-
tlonthati,s; -- 0ass-- 0,therebyralslngquestlonsastothemeanlngandnature
offunctlonsln mathematlcs. (Thlsls remlnlscentoftheway Lelbnlztwocenturles
ear|lerhadobtalnedcorrectresu|tslnca|cu|ususlnglnnnlte|ysma|lrealnumbers,
therebyral slngquestlonsastothenatureandro|eofnumbers lnmathematlcs. )
.,,.... .......,....
.............,..
.....
9.

1. =
3. =

5. = sinh t
7.

FIGURE 4.1.6.
8.

(
2
,

FIGURE 4. 1.7.
2. =
2
4. = cos
6. = sin
2

FIGURE 4.1.8.
10.
_
...... .. .....,.....
....... .....,...
..,.......
11. = .+ 3
13. =

15. = + cosh 5t
17. = cos
2

19. = +

21. = cos
12. = 3t
5
/2
.

14. =
/2

16. = sin + cos


18. = sin 3t cos 3t
20. =
22. = sinh
2
3t
276 Chapter 4 Lapl ace Transform Methods
..... .. .......,.
................

23. ._
.

25. .-

. .

27. .
. .
.
29. .
.

.
31. .
.

24. . .

26. .
.
.
28. .
.

.
.
. .
. .

33. Derive the transform of sin .by the method used


in the text to derive the formula in
34. Derive the transform of sinh .by the method used
in the text to derive the formula in .
35. Use the tabulated integral
!

cos ...

.cos .sin .
.
to obtain l{cos . directly from the defnition of the
Laplace transform.
36. Show that the function sin(et
2
) is of exponential
order as but that its derivative i s not.
37. Given . let if . if
. First, sketch the graph of the function making
clear its value at .Then express in terms of unit
step functions to show that . .


38. Given that . let i f .
i f either .or First, sketch the graph
of the function making clear its values at .and
Then express in terms of unit step functions to
show that . .


39. The .........i s defned as follows:
. if . . .
(a) Sketch the graph of to see why its name is appropri
ate. (b) Show that
C

_
. .

,
o
for all 0. (c) Assume that the Laplace transform of
the infnite series in part (b) can be taken termwise (it can).
Apply the geometric series to obtain the result
l

.
40. (a) The graph of the function is shown in Fig. .
Show that can be written in the form
C

_

. .

,
o
(b) Use the method of Problem to show that
l

.
C
+
o
6
t
FIGURE 4. 1. 10. The graph of the function of
Problem .
41. The graph of the .,........ is shown in
Fig. . Express .in terms of the function of Prob
lem .and hence deduce that

.
. . - tanh - .
. .
-
' : + -
FIGURE 4. 1. 11. The graph of the function of
Problem .
42. Given constants .and defne . for by
.
. if . n and .is odd;
if . .and .is even.
Sketch the graph of .and apply one of the preceding prob
lems to show that
.

. .
.
4. 2 Transformati on of I ni ti al Val ue Probl ems 277
Transformation of Initial Value Problems
y
_Ceat|aaeastaaet|ea
'
'
'
'
'
'
' '
'
'
' '
'
'
' '
'

'
u
'
'
'
'
'
'
'
o
'
'
'
'
'
y

'
'

'
'
'
'
' '
'
'

'
' .
u
' '
'
'
'
'
'

o
X
X
|eeew|seeeat|aaeaseet|vat|ve
FIGUR 4.2. 1. The
discontinuities of f' correspond to
"corers" on the graph of f.
We nowdlscusstheapp|lcatlonofLap|acetransformsto so|vea|lneardlfferentla|
equatlonwlthconstantcoemclents,suchas
.x

,;+/x ,;+cx,;= ( I )
wlth glven lnltla| condltlons x,c; = x,and x ,c; = x Bythe |lnearlty ofthe
Lap|acetransfomatlon,wecan transform Eq. ( I ) byseparate|ytaklng theLaplace
transformofeachtermlntheequatlon. Thetransformedequatlonls
.1,x
, ; | +/1,x ,; |+c1,x,; |= 1, ), ; | , (2)
ltlnvo|vesthetransformsofthederlvatlvesxandx

oftheunknownfunctlonx

Thekeytothemethod ls Theorem I , whlchte||sushowtoexpressthetransformof


theofafunctlonln termsofthetransformofthefunctlonltse|f.
THEOREM 1 Transforms of Derivatives
Supposethatthemnctlonl scontlnuousandplecewlsesmoothfor0and
l sofexponentla|orderas - +c, so that there exlstnonnegatlveconstantsM,
c,and1 suchmat
.for
Then

exlstsfora > c. and


= a1,
|
;
c= s i,s; - c .
(3)
(4)
Thefunctlonlsca||edpiecewise smooth ontheboundedlnterval/}lf lt
ls plecewlsecontlnuouson /}anddlerentlab|eexceptat nnlte|y many polnts,
wlth belng plecewlsecontlnuous on /} Wemay asslgn arbltrary va|ues
to at the lso|ated polnts at whlch ls not dlerentlable. We say that ls
plecewlsesmoothfor 0lfltlsplecewlsesmoothoneveryboundedsublnterva|
of[0,+) . Flgure4. 2. I lndlcates how comerson the graph ofcorrespondto
dlscontlnultlesln ltsderlvatlve

Themaln ldeaoftheproofofTheorem I lsexhlbltedbestbythecaselnwhlch

ls contlnuous (notmere|y plecewlsecontlnuous)for c Then, beglnnlng


wlththedennltlonof1, ) ,; |andlntegratlngbyparts, weget
1, ) ,; |=

=

,

+ s


Becauseof(3), thelntegratedterm approacheszero (when s >.;as-
+c, and lts va|ue at the |ower |lmlt= 0 contrlbutes -ctotheeva|uatlonof
thcprecedlngexpresslon. The lntegra|thatremalnsls slmply1, ), ; | ,byTheorem
2 ofSectlon4. I , the lntegra| converges whens > c Then 1, ) ,; |exlsts whea
s > c.andltsva|uels thatglvenlnEq. (4) . Wewl||deferthecaselnwhlch
haslso|ateddlscontlnultlestotheendofthlssectlon.
278 Chapter 4 Lapl ace Transform Methods
Exampl e 1
Solution of Initial Value Problems
Ia order to transform Eq. , i ;, we aeedthetraasform ofthesecoadderlvatlve as
we||. Ifweassume thaty,) = ) ,)satl snesthehypothesesofTheorem i , thea
thattheoremlmp|lesthat
aadthus
1, ) , ) ,= 1, y ,) ,=s 1, y,) ,- y,c)
=s 1, ) , ) , - )
,c)
=ss1, ), ) , - ),c) }- ) ,c) ,
1l ), ) ,=s

I,s) - s),c)- ) ,c)


Arepetltloaofthl sca|cu|atloaglves
(5)
1, ) , ) ,=s 1, ) , ) ,- ) ,c)=s

I,s) - s

),c)- s) ,c)- ),c) ,:)


Afternalte|ymanysuchstepsweobtalathefo||owlagexteasloaofTheorem i
COROLLARY Transforms of Hi gher Derivatives
Supposethatthefunctlons ), ) , ), , )

are contlauous andplecewlse


smoothfor < 0, andthateachofthesefunctlonssatlsnesthecondltlonsln(3)
wlththesame
-
va|uesofMand. Then1, )
-
, ) , exlstswhens >.. and
1, )

, ) ,=s
-
1, ), ) - s
-

,
),c)- s
-

) ,c)-

- )

(0)
=s
-
I,s) - s
-

,c)-

- s
-

(0) - )

,c)
So|vethel altla| va|ueprob|em
x- x- :x= 0, x ,c)=z, x ,c)= -i
(7)
Sol uti on Wlththeglvenl altla|va|ues, Eqs. (4)aad(5)yle|d
1,x ,) ,=s 1,x , ) ,- x(0) =sx,s) - z
aad
1,x,) ,=s

1,x , ) , - sx,c)- x(0) =s

x,s)- zs+ i .
where(accordlngtoourconveatloaaboutaotatlon)x,s)denotestheLap|acetraas-
formofthe(unkaown)fuactloax , ) Heacethetraasformedequatloals
[s

x,s)- zs+ i]- s x,s)- z}- :x,s) } =0,


whlchwequlck|yslmp|lfyto
Thus
,s

- s- :) x,s)- zs+= c
zs-
x,s) =
s

- s- :
zs- 3
,s- ) ,s+z)
Exampl e Z
4. 2 Transformati on of I ni ti al Val ue Probl ems 279
Bythemethodofpartla|fractlons(oflntegra|ca|cu|us), thereexlstconstaats~aad
ssuchthat
zs - n s
-- = - -,
,s- ; ,s+z) s- s+z
andmu|tlp|lcatlonofbothsldesofthlsequatlonby,s- ;,s+z)yle|dstheldeatlty
zs- = n,s+z)+s,s- ;
Ifwe substltute s = , we nnd that n = . substltutlon ofs = -z shows that
s= Hence
3 7
x,s; = 1,x , ) ,= +
s
z s- s+
Because1

, l},s- a) , = e

,ltfo||owsthat
ls the so|utlonoftheorlglna| lnltla| va|ue prob|em. Notethatwedldnotnrstnnd
the genera| so|utlon ofthe dlerentla| equatlon. The Lap|ace transform method
dlrect|yyle|dsthedeslredpartlcu|arso|utlon, automatlca||y taklng latoaccount
vlaTheorem I andlts coro||arytheglvenlnltla|condltlons.
Remark: InExamp|e I we foundtheva|uesofthepartla|-fractloncoefn-
clentsnandsbythetrlckofseparate|ysubstltutlngtheroots s= ands= -z
oftheorlglna|denomlnators
:
- s- := ,s- ; ,s+z)lntotheequatloa
zs- = n,s+z)+s,s- ;
thatresu|tedfrom c|earlng fractlons. In|leuofany such shortcut, the sure-nre
method lstoco||ectcoefnclentsofpowers ofsontherlght-handslde,
zs- = ,n+s;s+,zn- ;
Thenuponequatlngcoemclentsoftermsof|lkedegree,wegetthe|lnearequatloas
n+ s= z,
zn- s= -
whlcharereadl|yso|vedforthesameva|uesn= ands=
So|vethelnltla|va|ueprob|em
x++x= sln , x ,c)= x (0) = c

Such a prob|em arlses ln the motlon of a mass-and-sprlng system wlth extema|


force,asshownlnFlg.+ z z
280 Chapter 4 Lapl ace Transform Methods
Solution Becausebothlnltla| va|ues are zero, Eq. ,;yle|dslx(t ) = s

x,s) We read
thetransformofsln3t fromthetab|elnFlg.4. I . 2(Sectlon+ l ) aad therebygetthe
FIGURE 4.2.2. A mass-and
spring system satisfying the initial
value problem in Example The
mass i s initially at rest in its
equilibrium position.
\
FIGURE 4.2.3. The position
function . in Example
transformedequatlon
Therefore,
3
s

x,s) + +x,s) =

s +
3
x,s)
,s

++) ,s

+)

Themethodofpartla|fractloasca||sfor
3 ns+s cs+b
= +
,s

++) ,s

+) s

++ s

+
Thesure-nreapproachwou|dbetoc|earfractlonsbymu|tlp|ylagboth sldesbythe
commondenomlnator,andthenco||ectcoemclentsofpowersofsontherlght-hand
slde. Equatlngcoemclentsof|lkepowersonthetwosldesoftheresu|tlagequatlon
wou|dthenyle|dfour|laearequatlonsthatwecou|d so|veforn, s, c,aadb
However,herewecanantlclpatethatn= c= c,becauseneltherthenumer-
atornorthedenomlnatoronthe|eftlnvo|vesanyoddpowersofs, whereasnonzero
va|uesfornorcwou|d|eadtoodd-degreetermsontherlght. Sowerep|acenand
cwlthzerobeforec|earlngfractlons. Theresu|tlstheldentlty
3 = s,s

+ )+b,s

+ +)= ,s+b) s

+,s+ +b)
Whenweequatecoefnclentsof|lkepowersofswegetthe|lnearequatlons
s+ b = c,
s+ :b= 3,
whlcharereadl|yso|vedfor s= and b= - Heace
3 2 l 3
x,s) = lx(t ) = -
.

.
.
l c s

++ s

+
Becausel sln2t = z},s

++)andl sln3t = },s

+) , ltfo||owsthat
x(t ) = sln 2t sla 3t .
Flgure4. 2. 3showsthegraphofthlsperlod2posltlonfunctlonofthemass. Note
that the Lap|ace traasform method agaln glves the so|utlon dlrect|y, without the
necessltyofnrstnndlngthecomp|ementaryfunctlonandapartlcu|arso|utlonofthe
orlglna| nonhomogeneous dlerentla| equatlon. Thus nonhomogeneousequatlons
areso|vedlnexact|ythesamemaanerasarehomogeneousequatlons.
Examp|esI and2l||ustratetheso|utlonprocedurethatlsout|lnedlnFlg. 4. 2. 4.
Exampl e
4. 2 Transformati on of I ni ti al Val ue Probl ems 281
Ohcrcntal
cguaton
nx(/)
Alcbrac
cguaton
n a)
5o| utonx(/)
oldhcrcntal
5olutona)
) olalgcbrac
cguaton
FIGURE 4.2.4. Using the Laplace transform to solve an initial
value problem.
Linear Systems
Lap|acetransforms arefrequent|yusedlneaglaeerlngprob|emstoso|veasystea
oftwo or more constant-coemcleat |lneardlerentla| equatloas lnvo|vlagtwoor
moreunknownfunctloasx , ) , y,) , ofthe lndependentvarlab|e Whealaltla|
condltlonsarespeclned,theLap|acetransformreducessucha|lnearsystemofdll-
ferentla|equatlonstoa|lnearsystemofa|gebralcequatloaslawhlchthecakaowas
arethetraasformsoftheso|utlonfunctlons. AsExamp|el||ustrates,thetechalqce
forasystemls essentla||ythesameasforaslng|e|lneardlfferentla|eqcatloawlth
constantcoemclents.
So|vethesystem
zx= -:x+zy,
y= zx- zy+ +c sla ,
(S)
subecttothelnltla|condltlons
x ,c)=x ,c)= y,c)= y ,c)=c ,)
Thl slnltla|va|ueprob|emdetermlnesthelndlcateddlsp|acementfuactloasx,)aad
y, ) ofthetwomassesshownlnFlg. + z , assumlngthattheforce),)=:slaJt
l ssudden|yapp|ledtothesecondmassatthetlme = cwhenbothmassesare at
restln thelrequl|lbrlumposltlons.
J(t ) as n3t
FIGURE 4.2.5. A mass-and-spring system satisfying the initial
value problem in Example Both masses are initially at rest in their
equilibrium positions.
Sol ution Wewrlte x,s) = 1,x , ) , aadl,s) = 1, y,) , Thenthelaltla|coadltloasla,)
lmp|ythat
1,x'
, ) , =s

x,s, aad 1, y,) ,=s

:,s ,
282 Chapter 4 Lapl ace Transform Methods
}'
/

FIGURE 4.2.6. The position


functions . and in
Example
se.aase 1, s|a , = },s

+) , :|e ::aasie:msei:|e eaa:|eas| a( S) a:e :|e


eaa:|eas
zs

x,s)= -:x,s) +zl,s ) ,


l zc
s

r,s) = zx,s) - zl,s) +

s +
1|as:|e::aasie:measys:em|s
,s

+) x,s) - l,s) =c,


l zc
-zx,s) +,s

+z) l,s) =

s +
, l c)
1|eae:e:m|aaa:ei:||sa|:eii|aea:eaa:|eas|ax,s)aaal,s)|s
aaawe:eaa|iyseive-as|a,C:ame:s:aie,ie:|as:aa.e-:aesys:em|a, l c)ie:
aaa
l zc S
x,s) =
,s

+ l ) ,s

+ +) ,s

+ )
=
s

+ l
-
s

+ +
+
s

+
, l l a)
l zc,s

+) l c S l S
l,s) =
,s

+ l ) ,s

+ +) ,s

+ )
=
s

+ l
+
s

+ +
-
s

, l l o)
1|ea::|aii:a.:|eaae.emes|:|eas|as, l l a)aaa, l l |)a:e:eaa|iyieaaaas|a,
:|eme:|eaeisamie z te:|as:aa.e, ae:|a,:|a::|eaeaem|aa:e:ia.:e:s a:e
i|aea:|as

,we.aaw:|:e
l zc n s c

,s

l)

) ,

)
=

l
+

+
+

aaa|:ieiiews:|a:
l zc= n,s

++) ,s

+)+s,s

+t ) ,s

+)+c,s

+t ) ,s

++) , l z)
sa|s:|:a:|eaeis

= -l ,:|a:|s, s= / , aze:eei :|eia.:e:s

+l ) |a, l z),|ve
l zc = n

S, sen = s|m|ia:iy, sa|s:|:a:|eaeis

= -+|a , l z)y|eias
s = -S, aaasa|s:|:a:|eaeis

= -y|eiasc = 1|aswee|:a|a:|ea::|ai
i:a.:|eaae.emes|:|eas|ewa|a, l l a)
A: aay:a:e, :|e|ave:seLaia.e::aasie:msei:|ees:es|eas| as, l la)
aaa, l l |),|ve:|eseia:|ea
x,) = s|a- +s|az+ s|a,
y, ) = i c s|a ++s|az- : s|a
t|,a:e+ z :s|ews:|e,:a|sei:|ese:wee:|eaz-es|:|eaiaa.:|easei:|e:we
masses



FIGUR 4.2.7. A mass
spring-ashpot system with
exteral force J(t ) .

Exampl e 4
4. 2 Transformati on of I ni ti al Val ue Probl ems 283
The Transform Perspective
Le:as:e,a:a:|e,eae:ai.eas:aa:.eeia.|ea:se.eaae:ae:eaa:|eaas:a:eaa:|aa
eime:|ea
-x
+cx+/x=),;
ei:|e iam|i|a: mass-s:|a,-aas|e: sys:em ,t|, + z ) 1|ea :ae ::aasie:mea
eaa:|ea|s
-s
:
x,s;- sx,c;- x,; ]+.sx,s;- x,c;}+/x,s;=i,s; , l )
Ne:e:|a:, l )|saa./,e/m/.eaa:|ea-|aaeea,ai|aea:eaa:|ea-|a:aeaa
|aewax,s; 1||s|s:|esea:.eei:|eewe:ei:|eLaia.e::aasie:mme:aea
L|aea:dif erentl eaa:|easa:e::aasie:mea
|a:e:eaa|iyseiveaalgebraic eaa:|eas
iiweseive, l )ie:x,s; , we,e:
w|e:e
i,s; i,s;
x,s; =
z,s;
+
z,s;
.
z,s; = -s
:
+ .s + / aaa i ,s; =-x,c;s+ -x ,c;+ .x,o;
, l :)
Ne:e:|a:z,s; aeeaaseaiyea:|e|ys|.aisys:em|:seii 1aas, l +):es:a:s
x,s;=1,x, ; , as:|esameia:e:maeeaa|a,eaiyea:aees:enaiie:.eaaaeae
aeeaa|a,eaiyea:|e|a|:|ai.eaa|:|eas ia:|e.aseeiaaaaae:aameasys:em,
:|ese:we:e:msa:e:|e::aasie:ms
i,s;
1,x, , ; , =
z,s;
i ,s;
aaa 1,x,, , ; , =
z,s;
ei:|es:eaaye:|ea|.seia:|eaaaa:|e::aas|ea:seia:|ea,:ese.:|veiy1aeeaiye
:ea:|aia|u.ai:y|aaaa|a,:|eseseia:|eas|s|aaaa|a,:|e|ave:seLaia.e::aasie:m
ei:|e:|,|:|aaas|ae|a , l +) Ha.|ei:|e:ema|aae:ei:||s.|a:e:|saeve:ea
:eaaa|a,Laia.e::aasie:msaaa|ave:se::aasie:ms iaa::|.aia:,esee|:aae
me:|eas:|a:a:esau.|ea:iyewe:iai:eeaa|ieas:eseive:e|iems:aa:-aai||:
:|ese|asamieslaaaz-.aaae:|eseivea:eaa|iy|y:|eme:|easeiCaa:e:z
Additional Transform Techniques
s|ew:|a:
Sol ution ii),; =e

,:|ea),c;=aaa) ,; =e

+.e

uea.e1|ee:emi,|ves
1,e

+.e

,=1,
) , ; , =s 1, ), ; , =s 1, e

,
i : ieiiewsi:em:|ei|aea:|:yei:|e::aasie:m:|a:
284 Chapter 4 Lapl ace Transform Methods
uea.e
( l 5)
|e.aase1,
,

, l},s- .;
Exampl e
t|aa1, s|a / ,
Sol uti on Le:),) s|a / 1|ea),c; caaa
) , ) s|a /+ / .es /
1|eae:|va:|ve|aveives:|eaewiaa.:|ea.es/ , seweae:e:|a:),c) caaa
a|ne:ea:|a:ea,a|a 1|e:esai:|s
),) z/.es/- /

s|a/
sa: 1,) , s

1, ) ,|y:|eie:maia|a( 5) ie::|e::aasie:mei:|ese.eaa
ae:|va:|ve,aaa1,.es / , s},s

+/

) , sewe|ave
z/s

- /

1( s|a / , s

1, s|a / ,
s + /
t|aaiiy,weseive:||seaa:|eaie:

1||s:e.eaa:e|s.eas|ae:a|iyme:eieasaa::|aa:|eai:ena:|veeievaiaa:|a,:a:
|a:e,:ai
1, s|a/ ,

s|a/a

samies+aaa5 esie|::|eia.::|a:|i),c) c,:|eaa|ne:ea:|a:|eaei)


.e::eseaas:emai:|i|.a:|eaei|:s ::aasie:m|ys i:|s:easeaa|ie:e ese.::ae
|ave:seee:a:|eaei|a:e,:a:|ea,aa:|a|ne:ea:|a:|ea):e.e::eseaa:ea|v|s|eaei:a:
::aasie:m|ys
THEOREM Z Transforms of I ntegral s
ii),; |sa|e.ew|se.ea:|aaeasiaa.:|eaie: caaasa:|saes:ae.eaa|:|eaei
eseaea:|aie:ae: ), ; M,

ie: 1, :|ea


1 ),:;a:

1,) ,


ie:s > c. a|vaiea:iy,
1

,

),:; a: , l s)
Exampl e
4. 2 Transformati on of I ni ti al Val ue Probl ems 285
Poof: se.aase)|s|e.ew|se.ea:|aaeas,:|eiaaaamea:ai:|ee:emei.ai
.aias|mi|es:aa:
,, ; =

t
),:;a:
|s.ea:|aaeasaaa:|a:, , ; = ),;w|e:e)|s.ea:|aaeas, :|as,|s.ea:|aaeasaaa
|e.ew|sesmee:|ie: c ta::|e:me:e,
,
t
,
t M M
,,; ),:; a: M ecr
a: =
_
(e
c
t - i ;
_ _
e
c
t
,
o , c c
se,, ; |seieseaea:|aie:ae:as - +c. uea.ewe.aaaiy1aee:em:a,
:||s,|ves
1, ), ; , = 1,, ,; ,= s1, ,,; ,- ,,c;
New,,c;= c,sea|v|s|ea|ysy|eias
1

t
),:;a: = 1,,, ; , =
1, ),; ,
.
s
w||.|.emie:es:|e:eei
t|aa:|e|ave:seLaia.e::aasie:mei
o,s;
i
s
:
,s- .;

Sol uti on iaene.:, , i s;meaas:|a:we.aaaeie:eaia.:e:eisi:em:|eaeaem|aa:e:,aaa


:|e|ave:se::aasie:mei:|e:esai:|a,s|mie:es:ess|ea,aaaaaaiiy|a:e,:a:ei:em
o:e,:e.e::e.:ie::|em|ss|a,ia.:e:s; 1|as
1
-

i

1
-

a:=
t
e
a
r
a:=
!
(e
a
t


s ,s- .)
, s- . , .
weaew:eea::|e:e.|a|ae:ee|:a|a
1
-

= 1
-

a:= _(e
a
r
-
a:

,
t

,
t i
s
:
,s- .) , s ,s- .) , .
i i
a
r

,
t i
a
t =
-; -;
e - :
,
=
.
:
(e - .- i ;
1|| s:e.|a|ae| s eueaame:e.eavea|ea:way:aaa:|eme:|eaeia::|aii:a.:|eas
ie:aaa|a,aa|ave:se::aasie:meiai:a.:|eaei:|eie:mi,s;}s
-
_,s; }
286 Chapter 4 Lapl ace Transform Methods
Poof of Theorem I. we.ea.iaae:||sse.:|eaw|:|:ae:eeiei1aee:eml
|a:|e,eae:ai.ase|aw||.|]|sme:eiy|e.ew|se.ea:|aaeas weaeea:e:ev:
:|a::|ei|m|:
i|m

),;a

es|s:saaaaiseaeea:eaaa|:svaiae w|:|/asea,ie:, .

. .

-
, |e :|ee|a:s
|a:e:|e::e:|e|a:e:vai c,/}a:w||.|)|sa|s.ea:|aaeas Le:, = caaa

= /
1|ea we.aa|a:e,:a:e|ya::seaea.||a:e:vai ,
-
-
, ,
-
; wae:e]|s.ea:|aaea
1||sy|eias
New:|ea:s:samma:|ea
+

+

),

;+

),

-
, ;
+


),

I ;+

),

;
( l 9)
,zc)
|a( I 9):eies.eesaewa:e-),,;+

),

;= -),c;+

),/; . aaa:aese.
eaasamma:|eaaaasa:es:|mes:|e|a:e,:aii:em,=c:e

=/1ae:eie:e( l 9)
:eaa.es:e

) ,;a= -),c;+

),/;+s

),; a
sa:i:em,)we,e:
|is > c. 1|e:eie:e,aaaiiy:a||a,i|m|:s,w|:|sasea)as/- +c|a:ae:e.ea|a,
eaa:|ea, we,e::|eaes|:ea:esai:
1, )

,; ,= s 1, ), ; , - ),c;
Extension of Theorem 1
Newsaese:|a::|eiaa.:|ea)|seaiy|e.ew|se.ea:|aaeas,|as:eaaei.ea:|aa
eas), aaaie:, ,

, |e:|ee|a:s,ie: > c;wae:ee|:ae:)e:)|sa|s.ea:|a


aeas 1|eia.::|a:)|s|e.ew|se.ea:|aaeas|a.iaaes:|eassam:|ea:aa:-|:a|a
ea.||a:e:vai
-
-
, ,
-
}|e:weeasa..ess|vee|a:seia|s.ea:|aa|:y-)a,:eesw|:aa
iaa.:|ea:|a:|s.ea:|aaeasea:|ewaeie.iesea|a:e:vaiaaa|aseaae|a:vaiaes
:|a:mayae:a,:eew|:|:|ea.:aaivaiaes ),
-
-

; aaa),
-
; 1|evaiaeeiaa|a
:e,:aieaaa|a:e:vai|sae:ane.:ea|y.aaa,|a,:|evaiaesei:|e|a:e,:aaaa::ae
/(
t
)
6
5
4
3
2
0
cXump| C
0
1
2
3 4 5 6
t
FIGURE 4.2.8. The graph of the
unit staircase function of Example
7.
Problems
4. 2 Transformati on of I ni ti al Val ue Probl ems 287
eaae|a:s ueweve:,|i:|eiaaaamea:ai:|ee:emei.ai.aias|sa||ea:eaaa:a:
vaiae ei:ae |a:e,:ai, :|ea :|e aa:|ae:|va:|ve iaa.:|ea mas:oe .ea:|aaeaaa:a:
.iesea|a:e:vai we:|e:eie:ease:|e.ea:|aaeasi:em w|:a|a:ae|a:e:va|:aa
e|a:vaiaesaoeve |aevaiaa:|a,,|ya::s):|e|a:e,:aisea:ae:|,a:|a, l ) 1a:
:esai:|s
w|e:e
+
. . .
+

_
e
s
t
k
-2 ),_
2
;+ e-
s
tk -
I
),
,
-
j
;
+
_
e
s
t
k
- l
),
,
;+ e
-s
t
k
),
|
;
|~
j
= ),c

; ,(
tn
) + e-
s
b
),/; ,
n=1
,zc)
,zl )
aeae:es :|e ,aa|:e),-p| a),; a: = tn
. Assam|a,:aa:1, ) ,; | es|:, :
:|e:eie:e,e::|e,eae:ai|za:|ea
O
1, )
,; |= s i,s; ),c

;- e
-
s
tn
,(
tn
)
n=1
ei1, ) ,; |=s i,s; ),c;w|eaweaew:a|e:aei|m|:|a, l )as/ +.
,zz)
Le:), ; = l+ ;|e:aeaa|:s:a|:.aseiaa.:|ea,|:s,:a||ssaea| ai|,: z s
1aea),c;= l, ] ,;= 0,aaa,, = I ie:ea.a|a:e,e:= l , z, , u:a.e
,zz)y|eias
O
0 = s i,s; l e-
n
s ,
n=1
se:|eLaia.e::aasie:mei), ; | s
l
-
n
s
l
i,s; =

e = .
s
n=O
s , i - e-S )
ia:|eias:s:eweasea:|eie:maiaie::aesameia,eeme::|.se:|e,
l x

|:ax = e-
s
< I .
...,...........,..
.....
3. . . .= .= . =
4. ..+ .= .= .
1. .+ ..= .= . =
5. .+ .= sin .= = .
6. ...= cos t ; .= = .
2. .+ .= .= . = . 7. .+ . cos .= .
288 Chapter 4 Lapl ace Transform Methods
8. .+ .= .==.
9. .+ ..+ .= .==.
10. .+ . + .= .=. =
11. .=.+ =.+ .= =
12. . .+ =.+
.==
13. .+ + .=. + =.= =
14. ..+ .=+ X + = .==
. =
15. .+ . + . =+ . + + .. =
.== . = =
16. .=. =. =..= =
=
.,,.........,......
...... ....

17. .= 18. . =
. . . .+
.+
19. .=
S (S
2
+ .
20. .=
S (S
2
+

21. . =
2 2
22. . =
S .+ . .

23. . = 24. . =
S
2
(S
2
. .+ .+
25. Apply Theorem to derive { sin kt } from the formula for
{cos kt } .
26. Apply Theorem to derive { cosh kt } from the formula
for { sinh kt } .
27. (a) Apply Theorem to show that
.
{

} =
_
{

'
}
. .
(b) Deduce that {

} = .. .

_
for .=
. . .
.,,........,....,....
.......
S
2
- k
2
28. { t cos kt } =
(s
2
+ k
2
)
2
..
29. { t sinh . =
.

P
)
2
S
2
+ k
2
30. { t cosh kt } =
(S
2

P
)
2
31. Apply the results in Example and Problem to show
that
_
_

2 2 2
=
3
smk k cos k ) .
.+ k ) k
.,,.........,..
..,.............
32. . . . } =S -
I

for .
33. If = I on the interval .,(where . and
otherwise, then

} =

.
34. If =

'

is the square-wave function whose


graph is shown in Fig. . then
.
.

= - tanh - .
.
.....Use the geometric series. )
!|
)
| ,
. ;
s
.

FIGURE 4.2.9. The graph of the
square-wave function of Problem .
35. If i s the ..., ...whose graph is shown in
Fig. . then

} =
. +
!|

.

P

. ;

FIGURE 4.2. 10. The graph of the


on-of function of Problem
36. If get ) is the ..........whose graph is
shown in Fig. . then
.
{ . } = _ tanh - .
.
y|| )

FIGURE 4.2. 11. The graph of the


triangular wave function of Problem
37. If i s the .......whose graph i s shown in
Fig. . then

} = _
) . .
.....Note that = where it is defned. )
!'|
;
s
.

FIGURE 4.2. 12. The graph of the


sawtooth function of Problem
4. 3 Transl ati on and Parti al Fracti ons 289
Translation and Partial Fractions
As|iias::a:ea|ysamieslaaa2eise.:|ea4. 2, :|eseia:|eaeiai|aea:a|ne:ea:|ai
eaa:|eaw|:|.eas:aa:.eeu.|ea:s.aaei:ea|e:eaa.ea:e:|ema::e:eiaaa|a,:ae
|ave:seLaia.e::aasie:meia:a:|eaaiiaa.:|eaei:|eie:m



, l )
w|e:e :|e ae,:ee ei | s iess:|aa :|a: ei 1|e :e.|a|ae ie: aaa|a,

|s|aseaea:|e same me:|ea eia::|ai i:a.:|eas:|a:wease|aeie


mea:a:y.ai.aias:e|a:e,:a:e:a:|eaaiiaa.:|eas 1aeieiiew|a,:we:aiesaes.:||e
:|epartial fraction decomposition ei |a:e:msei:|eia.:e:|za:|eaei:aeae
aem|aa:e:|a:ei|aea:ia.:e:saaa|::eaa.||ieaaa:a:|.ia.:e:s.e::eseaa|a,
:e:|e:eaiaaa.emiesze:es, :ese.:|veiy,ei
RULE 1 Linear Factor Partial Fractions
1|ee::|eaei:|ea::|aii:a.:|eaae.emes|:|eaei .e::eseaa|a,:e:|e
i|aea:ia.:e: eimai:|i|.|:y |s asam ei a::|aii:a.:|eas, |av|a,:|e
iem

"
+ +
. . .
+ ,

w|e:e

. . . ,aaa

a:e.eas:aa:s.
RULE Z Quadratic Factor Partial Fractions
(2)
1|ee::|eaei:|ea::|aii:a.:|eaae.emes|:|ea.e::eseaa|a,:eme|::eaa.||ie
aaa:a:|.ia.:e:

eimai:|i|.|:y |sasameia::|aii:a.:|eas,
|av|a,:|eie:m
w|e:e

. ,aaa

a:e.eas:aa:s
t|aa|a,

|aveives:wes:es. t|:s:wemas:aaa:|ea::|aii:a.:|ea
ae.emes|:|eaei aaa:|eawemas:aaa:|e|ave:seLaia.e::aasie:meiea.|
ei:|e|aa|v|aaaia::|aii:a.:|easei:|e:yes:|a:aea:|a(2) aaa,) 1|eia::e:
s:e|s|aseaea:|eieiiew|a,eiemea:a:y:ee::yeiLaia.e::aasie:ms
THEOREM 1 Transl ation on the s- Axls
iiF(s) ex|s:sie:s > ..:|ea

es|s:sie:> +.,aaa
La|vaiea:iy,


(4)
,
1|as:|e::aasia:|eas |a:|e::aasie:m.eneseaas:emai:|i|.a:|eaei
:|ee:|,|aaiiaa.:|eaei|y

290 Chapter 4 Lapl ace Transform Methods


Exampl e 1
Sol ution
FIGUR 4.3. 1. The mass
spring-dashpot system of Example

X
L

FIGUR 4.3.2. The position


function (t) in Example
Proof: iiwes|miy:eia.esw|:|s -a|a:|eaeaa|:|eaeiI,s) =1, ), ; , ,
wee|:a|a
I,s- a) =

'

'),; a=

' ),; ]a=


_| '),; ,
1||s|s,+),aaa|:|s.iea::|a:,)|s:|esame
iiweaiy:|e::aasia:|ea:|ee:em:e:|eie:maiasie::|eLaia.e::aasie:ms
ei
-
,.es/ , aaas|a/:|a:weai:eaay|aew-mai:|iy|a,ea.|ei:|eseiaa.:|eas
|y' aaa:eia.|a,sw|:|s- a|a:ae::aasie:ms-we,e::|eieiiew|a,aaa|:|ea
:e:|e:a|ie|at|,+ l z
q
j
_n
.
(s ~ a)
(s - a)
n+
1
s - a
q
j
cos kt
( s - a)
2
+ k
2
(s ~ a)
k
q
j
sin kt
(s - a)
2
+ k
2
(s ~ a) (8)
te::eaay:eie:ea.e, aii :|eLaia.e::aasie:ms ae:|vea |a:||s .aa:e:a::
i|s:ea|a:|e:a|ieei::aasie:ms:|a:aea:s|a:|eeaaae:s
Ceas|ae:amassaaas:|a,sys:emw|:|m = / = I 7,aaac = |am|saa|:s
,t|, + l ) Asasaai, ie:x , ) aeae:e:|ea|sia.emea:ei:|e mass m i:em |:s
ea|i||:|ames|:|eaii:|emass|sse:|ame:|eaw|:|x(0) =aaax(0) = I , aaa
x , ) ie::|e:esai:|a,aameai:eees.|iia:|eas
1|ea|ne:ea:|aieaa:|ea|sx+ x+ l x = 0, seweaeea: eseive:|e|a|:|ai
vaiae:e|iem
x+:x++x=0, x,c)= , x ,c)= I .
we:a|e:aeLaia.e::aasie:meiea.|:e:mei:|ea|ne:ea:|aieaa:|ea se.aa:
,e|v|easiy)1, c, = 0, we,e::|eeaa:|ea
s

x,s) - s- l j+:s x,s) - [++x,s) =0,


w||.|weseiveie:
s+I 9 s+
x,s) =
s

+ :s + +
=

,s + )

+
z
+
z

,s + )

+ z

Aiy|a,:|eie:maias|a(7)aaa(S)w|:|a=-aaa/=, weaewsee:|a:
x ,)=

' ,.es+zs|a )
t|,a:e+ zs|ews:|e,:a|ei:||s:a|aiyae.ay|a,aameaes.|iia:|ea
Exampl e Z
4. 3 Transl ati on and Parti al Fracti ons 291
samiez|iias::a:esaaseiai:e.|a|aeie:aaa|a,:|ea::|aii:a.:|ea.e:m
.|ea:s|a:|e.aseeiaea:eea:eai|aea:ia.:e:s
. ~....... .._ --.-- .... . . _ - -------
t|aa:|e|ave:seLaia.e::aasie:mei
s

+ l
k,s) =

s
Sol ution Ne:e:|a::|eaeaem|aa:e:eik,s) ia.:e:sas_,s) =s ,s+z) ,s- +) uea.e
Exampl e
s

+ l n s c
- = +-+
s

- zs

- ss s s+z s- +
Hai:|i|.a:|eaeiea.|:e:mei:||seaa:|ea|y _,s) y|eias
s

+l =n,s+z) ,s- +)+ss,s- +)+cs ,s+z)


w|eawesa..ess|veiysa|s:|:a:e:|e:|:eeze:ess = 0, s = -z, aaas =+ei:ae
aeaem|aa:e:_,s) |a:||seaa:|ea,we,e::|e:esai:s
-sn= l , l zs=. aaa z+c= l
1|asn=- , s= _, aaac= _, se
aaa:|e:eie:e
s

+l
_ _
- =-- + -+-,
s

- zs

- ss s s+z s- +
1 =- +e +e
_
j

+ l l

, l
.
,
s

- zs

- ss s l z z+

samie|iias::a:esaa|ne:ea:|a:|ea:e.|a|aeie:aaa|a,:|ea::|aina.:|ea
.eeu.|ea:s|a:|e.aseei:eea:eai|aea:ia.:e:s
seive:|e|a|:|aivaiae:e|iem
y

+:y++y=

,
y,c;=y ,c;=c
Sol ution 1|e::aasie:meaeaa:|ea|s
1|as
z
s

:,s)++s ,s)++,s) =
s
z n s c n i
,s)=
s

,s+z)

=
s

+
s

+
,
+
,s+z)

+
s+z
,;
1eaaan, s, aaac, wemai:|iy|e:|s|aes|ys

:ee|:a|a
z

=n+ss+cs

+s

I,s) ,
, s+z)
, i
292 Chapter 4 Lapl ace Transform Methods
cXump| C4
w|e:e I,s) = b,s+z)

+I,s+z)

| :ae sam ei:|e:wea::|ai i:a.:|ea


.e::eseaa|a,:e,s+z)

sa|s:|:a:|eaeis = c|a, l c)y|:iasn = 1eaaa


s aaac,wea|iie:ea:|a:e, l c):w|.e:ee|:a|a
aaa
-+
- = s+zcs+s

I,s) +s

I ,s)
, s+z)
l z
= zc+:s I,s) +:s

I ,s)+s

I,s)
,s+ z)
, l l ,
, l z,
Newsa|s:|:a:|eaeis = c|a, l l ) y|eiass = - , aaasa|s:|:a:|eaeis = |a
, l z)y|eiasc=
1eaaabaaaI, wemai:|iyea.|s|ae|a(9)|y,s+z)

:e,e:
z
= b+I,s+z)+,s+z)

o,s) ,
s
w|e:eo,s) = ns

+ss

+cs

,
,aaa:|eaa|ii::ea:|a:::ee|:a|a
:
_= I+z,s+z) o,s) +,s+z)

o ,s)
s
, l )
, l :,
sa|s:|:a:|eaeis = -z|as , l )aaa, l +)aewy|eiasb = - aaaI =
~
v
1|as
(

:,s) = +
"

s ,s+z)

s+z

se:|eseia:|eaei:|e,|vea|a|:|aivaiae:e|iem|s

samies+,, aaa:|iia::a:e:e.aa|aesie:aea||a,w|:|aaa:a:|.ia.:e:s|a
a::|aii:a.:|eaae.emes|:|eas
Ceas|ae::aemass-s:|a,-aas|e:sys:emas|asamie l, |a:w|:a|a|:|ai.eaa|
:|easx ,c)= x ,c)= caaaw|:|:ae|meseaes:enaiie:.eI,)= l s|a z t|aa
:|e:esai:|a,::aas|ea:me:|eaaaas:eaaye:|ea|.me:|eaei:|:mass
Sol ution 1|e|a|:|aivaiae:e|iemweaeea:eseiv:|
x+:x++x= c s|a z , x ,c)= x ,c,= c
1|e::aasie:meaeaa:|ea|s
uea.e
:c
s

x,s)+:s x,s)++x,s) =
_
:

s +
:c ns+s cs+b
x,s) =
,s

+ +) ,s + )

+ z}
=
s

+ +
+
,s + )

+ z

FIGURE 4.3.3. The periodic


forced oscillation xsp (t ) , damped
transient motion Xtr (t ) , and
solution x(t ) xsp (t) + Xtr (t) in
Example .
4.3 Transl ati on and Parti al Fracti ons 293
....,,..,.......,
= .. +,
...... ....= .,.....+
., ..
= ..
.....,,
= +.. . .
.,...,.....,.,,.....,...
...,..
.0s= .. . .=
.....,...=

...=
.... ....= .,.....
., ..,
= - -
.....,,


,..,...,.....,.,,. .,....,..
= .. =
.....,...= =
....... . .. .,...
..,x,s

, - +

..,.....,... .,.,.
. = ,


.....,.,....,,...
.......,.,...,....,.,.
.......,,..,.,.,,+ , ...
.....,....
294 Chapter 4 Lapl ace Transform Methods
cXump| C
Resonance and Repeated Quadratic Factors
1|eieiiew|a,:we|ave:seLaia.e::aasie:msa:easeiai|a|ave::|a,a::|aii:a.
:|eas:|a:.e::eseaa:e:|e.aseei:eea:eaaaa:a:|.ia.:e:s
.|

s I

/ J

-sm ,
, s +/ ) z/
_

I I
.
J

=
_
,sm/- / .es /)
,s +/ ) z/
( l 6)
( I 7)
1|eseieiiewi:emsamieaaar:e|iem3 I eise.:|ea+ z, :ese.:|veiyse.aa:
ei:|e:esea.e|as ( I 6) aaa( I 7)ei:|e:e:mss|a/ aaa.es/ , a:eea::a
aaa:a:|.ia.:e:e:a|aa:|iys|,aais:ae|eaemeaeaei:eseaaa.e|aaaaaaam:a
me.|aa|.aie:eie.::|.aisys:em

UseLaia.e::aasie:ms:eseive:|e|a|:|aivaiae:e|iem
x+ox I,s|ao , x ,c) 0 x(0)
:|a:ae:e:m|aes:|eaaaameaie:.eaes.|iia:|easeiamasseaas:|a,
Sol uti on w|eawe::aasie:m:aea|iie:ea:|aieaa:|ea,we,e::aeeaa:|ea
4
-4
FIGURE 4.3.4. The resonance
solution in with (a = and
Fa = together with its envelope
curves = .

I,o I,o
s x,s) +o
,
x,s)

se x,s)

s +o ,s

+o

) (s

+o
,
)
iio= o,,weaaaw|:|ea:a|ia.ai:y:|a:
se|:ieiiews:|a:
x,s)
I,o I I
o

- o s

+o s

+o

I,o I

x , )

smo,-

smo
o

- o
,
o, o
sa:|io o,,we|ave
I,o,
x,s)

,s +o
,
)
se( I 7)y|eias:|e:eseaaa.eseia:|ea
I,

x,)

,smo,- o,.eso,)
zo
,
( I S)

Remark: 1|eseia:|ea.a:ve aeaaea|a ( I S) |eaa.es|a.| aaaie::a


,seet|, 4. 3. 4) |e:weea :|e eaveiee.a:vesx c, ) :|a: a:ee|:a|aea|y
w:|:|a,( I S)|a:|eie:m
x ,)= n, ) .es o,+s, ) s|a o,
aaa:|eaaeaa|a,:|easaaiami|:aaec , n

+ s

ia:||s.aseweaaa:aa:
c, )
I,
o

+ I .
zo
1||s:e.|a|aeie:.eas::a.:|a,eaveiee.a:vesei:eseaaa.eseia:|eas|s|iias::a:ea
ia::|e:|a:|eai|.a:|eama:e:|aiie::||sse.:|ea
cXump| e
seive:|e|a|:|aivaiae:e|iem
4. 3 Transl ati on and Parti al Fracti ons 295
y
-
+zy

+y= :e

. y,c;= y ,c;= y

,c;= y

,c;= c
Sol uti on r|:s:wee|se:ve:|a:
uea.e:|e::aasie:meaeaa:|ea|s
:
,s
-
+2s
:
+ I ) I(s) =
:
.
(s - I )
1|asea::e|iem| s :eaaa:|e|ave:se::aasie:mei
:
I(s) =
(s- I )
:
(s
:
+ I )
:
A B Cs + D Es + F
= + -+ + . ( I 9)
(s - I )
:
s - I (s
:
+ I )
:
s
:
+ I
iiwemai:|iy|y:|e.emmeaaeaem|aa:e:(s - I )
:
(s
:
+ I )
:
,we,e::|eeaa:|ea
A(s
:
+ I )
:
+ B(s - I ) (s
:
+ I )
:
+Cs (s- I )
:
+ D(s - I )
:
+ Es (s - I )
:
(s
:
+ I ) + F(s - l )
:
(s
:
+ I ) = : (20)
Ueasa|s:|:a:|a,s = I weaaa:|a:A = I .
aa:|ea(20)|saa|aea:|:y:|a:|eiasie:aiivaiaeseis . 1eaaa:|evaiaesei
:|e:ema|a|a,.eeu.|ea:s,wesa|s:|:a:e|asa..ess|ea:|evaiaess = 0, s = -I ,
s =2 , s = -2, aaas = 3 |a(20). 1||sy|eias:|esys:em
-B + D + F = 3,
-SB - 4C + 4D - SE+ SF = 0,
25B + 2C + D + I 0E + 5F= -2I , (2I )
-75B - I SC + 9D- 90E+ 45F = -2I ,
200B + I 2C+4D+ I 20E+ 40F = -96
eiavei|aea:eaa:|eas|a B, C, D, E, aaa F. w|:|:|ea|aeia.ai.aia:e::e
,:ammea:eseivei|aea:sys:ems, weaaa:|a:B = -2, C = 2, D = 0, E = 2,aaa
F = I .
weaewsa|s:|:a:e|a( I S):|e.eeu.|ea:swe|aveieaaa,aaa:aase|:a|a
I 2 2s 2s + I
I(s) = - -+ + -.
(s - I )
:
s - I (s
:
+ I )
:
s
:
+ I
ke.aii|a,, I 6), :|e::aasia:|ea:ee::y,aaa:|eiam|i|a:::aasie:msei.esaaa
s|a , weseeaaaiiy:|a::|eseia:|eaei:|e,|vea|a|:|aivaiae:e|iem|s
y,; = ,- 2) e

+ ,+ I)s|a +2 cos t .
296 Chapter 4 Lapl ace Transform Methods
Problems
.,,...........,.....
...........
1. =

2.

3. =

sin
4. =

cos , ,
.,,..............,....
............
.
5. .
.
6. .
. . .

7. .=
.

...
.
8. .
.

..
.
9. .
.
.
.
.
10. .
.
.
.
.,............,.....
....... ....

1 1. .

. .
.
13. .=
.

15. .=

. .

17. .=

.
.

.
19. .=

.. .
.


21. .=
.

.....
.
12. .

. .
. .
14. .
.
.

16. .
.

18. .
. .

...
. .

22. .
..

..

......,.........
....
23. -
1

..

cosh .cos .
24.
- 1

..

sinh .sin .
25.
- 1

..

(cosh .sin .sinh .cos .


26. -
1

(cosh . sin at - sinh . cos .


. .. ..
...,...........,..
.....
27. .+ .... =
28. . ... . =
29. . .. . . =
30. ....=
..
31. .

. .= .. . =
32. .

.. . . = .

=
33. .
.= . . = . .

=
34. .

.... .
.

=
35. .

. .= . . = . =
.


36. .

..

.. .= .

=
37. ... .= .. =
38. .. .cos . . =
........,.......
....,......,........
.............. = 0
39. Suppose that k and Los
Use the inverse transform given in Eq. t o derive the
solution . sin Construct a fgure that illustrates
the resonance that occurs.
40. Suppose that = k . . and =

cos Derive the solution


.

.
Show that the maximum value of the amplitude function
. =

is . Thus (as indicated in


Fig. . the oscillations of the mass increase in am
plitude during the frst s before being damped out as

Z
-Z
FIGUR 4.3.5. The graph of the damped
oscillation in Problem .
4. 4 Derivati ves, I ntegral s, and Products of Transforms 297
1|eLaia.e::aasie:mei:|e,|a|:|aiiyaa|aewa)seia:|eaeiaa|ne:ea:|aieaa:|ea
|sseme:|mes:e.e,a|za|ieas:|e:eaa.:ei:|e::aasie:msei:we/oiaa.:|eas
te:esamie,w|eawe::aasie:m:|e|a|:|aivaiae:e|iem
we,e:
x+ x = .es , x(0) x

|0) = 0,
X(s)

= _

"
l

= 1,.e ,

1, s|a ,
(s + I ) s + I s + I
1||ss::ea,iysa,,es:s:|a::|e:eea,|::e|eawayei.em||a|a,:|e:weiaa.:|ea
s|a aaa .es :e e|:a|a a iaa.:|ea x(t ) w|ese::aasie:m |s :aepoaecei/e/
::aasie:ms sa:e|v|easiyx,)|sos|miy:|e:eaa.:ei.esaaas|a,|e.aase
1,.ess|a ,= 1
s|a2t j =

"
l


+4 (s + l )
1|as1,.es s|a , = 1,.es , 1, s|a ,
1|ee:em I ei:||sse.:|eaw|ii:eiias:|a::|eiaa.:|ea
/,) ](r) g(t - r) ar
|as:|eaes|:ea:ee::y:|a:
1, / ,) , H(s) F(s)
.
G(s ) .
( I )
(2)
1|eaewiaa.:|eaeiaeaaeaas:|e|a:e,:ai|a( I ) aeeaaseaiyea]aaa,aaa|s
.aiiea:|e.o.o//oei]aaa, i:|saeae:ea|y]* ,,:|e|aea|e|a,:aa:|:|sa
aew:yeei:eaa.:ei]aaa,, se:a|ie:ea:|a:|:s::aasie:m|s:ae:eaa.:ei:ae
::aasie:msei]aaa,
DEFI NI TI ON The Convol uti on of Two Functions
1|econvolution ]* ,ei:|e|e.ew|se.ea:|aaeasiaa.:|eas]aaa,|saeaaea
ie: 0asieiiews .
,)* ,) , ) ]|r) g,t - r) ar*
(3)
wew|iiaisew:|:e
),)* ,( ) w|ea.eavea|ea: ia:e:msei:|e.eaveia:|ea
:eaa.:,1|ee:emI ei:||sse.:|easays:aa:
1, )* y, 1, ),

1, y,
iiwema|e:|esa|s:|:a:|ea= - r |a:|e|a:e,:ai|a(3), wesee:aa:
),)* ,() ],r) g,t - r)a:
],- ; ,,; , -a;
,,; ),- ; a ,,)* ), )
1|as:|e.eaveia:|ea| s .o--./.e ]* ,= ,* ]
298 Chapter 4 Lapl ace Transform Methods
cXump| C 1
cXump| CZ
Theconvo|utlonofcost and slnt ls
(cost) * (slnt) = cosrsln(t- r) dr
Weapp|ythetrlgonometrlcldentlty
cosAslnB = sln(A+ B) - sln(A- B) ]
t oobtaln
(cos t ) * (sln t ) =sln t - sln(2r - t ) ] dr
=

rsl nt+ cos (2r t )

a
thatl s,
(cost ) * (sl nt ) = tsl nt.
Andwereca||fromExamp|e5ofSectlon4. 2thattheLap|acetransformoftslnt
ls lndeedsJ(s
:
+ I )
:
.
Theorem l ls provedattheendofthl ssectlon.
THEOREM 1 The Convol ution Property
Supposethat ](t ) and g(t ) areplecewlsecontlnuousfor 0 andthat ),;
and ,,; areboundedby Me' ast +c. ThentheLap|acetransformofthe
convoIutlon](t ) * ,, ; exlstsfors > c, moreover,
l ](t ) * g,t ) =1, ), ; ,

lg(t ) (4)
and
l
(
F(s)
.
G(s ) = ](t ) * g(t) . (5)
Thuswecan nndthelnversetransformoftheproduct F(s)

G(s) , provlded
thatwecaneva|uatethelntegra|
l
(
F(s)

G(s) =](r) g(t r) dr. (5')


Examp|e 2 l||ustrates thefact that convo|utlon often provldes a convenlent
a|ternatlvetotheuseofpartla|fractlonsfornndlnglnversetransforms.
Wlth](t ) =sln2tandg(t ) = e' ,convo|utlonyle|ds
,

2
:

= (sln 2t)* e' =

e'

sln 2rdr
( s- I ) (s +4)
]
,

]
,
e

sln 2r dr = e
'
( - sln 2r - 2 cos 2r) _ ,
so
l = e - sm2t- cos2t.
__

2
'
I

2
(s - I ) (s
:
+4) 5 5 5

cXump| C
4. 4 Deri vati ves, I ntegral s, and Products of Transforms 299
Diferentiation of Transforms
Accordlng toTheorem I ofSectlon4. 2, lf= 0 then dlfferentlatlon of],t)
correspondstomu|tlp|lcatlonofl tstransformbys Theorem 2,provedattheend
of thls sectlon, te||s us that dlfferentlatlon ofthe transform F,s) corresponds to
mu|tlp|lcatlonof theorlglna|functlon](t ) by t .
THEOREM Z Differentiation of Transforms
If lsplecewlsecontlnuousfor0and Me

as- +c,then
l - = F
'
(s) (6)
fors >C. Equlva|ent|y,
(7)
Repeatedapp|lcatlonofEq. (6)glves
(S)
for= I , 2, 3, . . . .
Flndl

,
sln kt .
Sol uti on Equatlon(S)glves
cXump| C4
,9)

Theform ofthedlfferentlatlonproperty lnEq. (7) ls often he|pfu| lnnndlng


anlnversetransformwhentheae/../.eofthetransformlseaslertoworkwlththan
thetransformltse|f.
Flndl
.
tan
,
( l Js ) .
Sol uti on Thederlvatlveof tan
|
( l Js ) lsaslmp|eratlona|functlon,soweapp|yEq,7).
l tan = l tan
_

.
I

l
.
a _

s t ds s
300 Chapter 4 Lapl ace Transform Methods
cXump| e
Therefore,
-
(
slnt
l tan =
s t

Equatlon(S)canbeapp|ledtotransforma|lneardlerentla|equatlonhavlng
po|ynomla|,ratherthanconstant,coefnclents. Theresu|twl||beadlerentla|equa-
tlon lnvo|vlngthe transform, whether thls procedure |eads to success depends, of
course,onwhetherwecan so|vethenewequatlonmorereadl|ythantheo|done
Letx (t ) betheso|utlonofBesse| ` sequatlonoforderzero,
tx+x'+tx = 0,
suchthatx (0) = l andx'(0) = 0. Thls so|utlonofBesse| ` sequatlonlscustomarl|y
denotedbyJ, (t ) . Because
lx
'
(t ) =s x,s) - l and lx (t ) =s

x,s) - s,
and becausex and x are each mu|tlp|ledbyt , app|lcatlonofEq. (6) yle|ds the
transformedequatlon
a a
-- s

x,s) - sj+sx,s) - l }- - x,s) } = 0.


as as
Theresu|tofdlerentlatlonandslmp|lncatlonlsthedlfferentla|equatlon
,s

+l ) x ,s)+ s x,s) = 0.
Thl sequatlonlsseparab|e
ltsgenera|so|utlon ls
x ,s) s
=
x,s)
-
s

+ l

c
x,s) = _
s

+l
InProb|em 39 weout|lnethe argumentthat C = l Becausex,s) = l J, (t ) , lt
fo||owsthat
l
l J, (t ) =
s

+l
, l c)

Exampl e
4. 4 Deri vati ves, I ntegral s, and Products of Transforms 301
Integration of Transforms
Dlerentlatlon of F(s ) corresponds to mu|tlp|lcatlon of ](t ) by t (together wlth
a change of slgn). It ls therefore natura| to expect that lntegratlon of F(s) wl||
correspondtodlvlslonof)(t) byt . Theorem , proved at the endofthls sectlon,
connrms thls, provldedthatthe resu|tlng quotlent ](t )Jt remalns we|| behavedas
0fromtherlght, thatls, provldedthat
_
)(t)
rm

t
exlstsandls nnlte
THEOREM 3 I ntegration of Transforms
, l l )
Supposethat),;l s plecewlsecontlnuousfort 0, that),;satlsnesthecon-
dltlonln( I I ) , andthat

), ; Me

as+c. Then
fors > c. Equlva|ent|y,
. . . ....... ................
Flnd; (slnh t )Jt .

](t )

t

_
F(c) dc , l z)
, l )
Sol uti on We nrstverlfythatthecondltlonln, l l ) ho|ds .
Exampl e
slnht e- e e
+e
|lm= |lm |lm l ,

2t

2
wlththealdofl ` Hplta| ` sru|e. Then Eq. ( I 2), wlth )(t; slnht, yle|ds
slnht ` dc
; ; slnh t dc

t _ _ c l
Therefore,
becauselnl = 0.
l ` l l l c - l ,
`

_
_ c l
-
c + l
dc
_
ln
c + l

slnht

ln
s + l
,
t 2 s - l

Theformofthelntegratlonproperty lnEq. , l )ls oftenhe|pfu|lnnndlngan


lnverse transform when the lndennlte /e,m/ofthe transform ls easlerto hand|e
thanthetransformltse|f.
302 Chapter 4 Lapl ace Transform Methods
Sol uti on We cou|dusepartla|fractlons,butltlsmuchslmp|ertoapp|yEq. , i ; Thlsgives
1

.
a-
zs

z-

,s
:
- i ;
:

,-
:
- i ;
:
andtherefore

-i ,

.

= 1
-
:

= 1
s
:
- i

,s
:

i ;
:

= slnh t .
" Proofs of Theorems

Proof of Theorem I : The transforms i,s; and o,s; exlst when s ~ .o,
TheoremzofSectlon: i Forany : > 0 the dennltlon oftheLap|acetransforn
glves
o,s; =

e
-
,;a=

e

-
,- :;a ,= - :; .
andtherefore
o,s; =e

e
-
,- :;dt ,
becausewemayae]e],;and,;t obezerofor < 0. Then
F,s; o,s;= o,s;

],:;a:=

e
-
),:; o,s;a:
=

e

],:;

,- :; a a:
=

),:;,- :; aa:
Nowourhypotheseson]and lmp|ythattheorderoflntegratlonmaybereversed.
(Theproofofthls requlres adlscusslonofunlform convergence oflmproperlnte-
gra|s, andcanbefoundln Chapter zofChurchl|| ` sopem/o./M./e-./.s,Jrd
ed. (NewYork. McGraw-Hl||, i :z; ;Hence
andtherefore,
i,s; o,s; =

e
],:; ,- :;a:a
=

),:; ,- :;a:,a
=

e
-
),;
, ; ] a ,
F,s; o,s;= 1 ),; ,; .
Werep|acetheupper|lmltofthelnnerlntegra|wlthbecause, :;= 0whenever
: > Thlscomp|etestheproofofTheorem i A
Problems
4. 4 Deri vati ves , I ntegral s, and Products of Transforms 303
Poof of Theorem 2. Because
dlfferentlatlonunderthelntegra|slgnyle|ds
thus
a
,

i,s; = - c

)(t)a
as
0
i ,s;= l -t](t ) ,
whlchl s Eq.(6). We obtalnEq.(7)byapp|ylngl
|
andthendlvldlngby-t . The
va|ldltyofdlerentlatlonunderthelntegra| slgndepends onunlformconvergence
oftheresu|tlnglntegra| , thls lsdlscussedlnChapter2ofthebookbyChurchl||ust
mentloned. A
Proof of Theorem J: Bydennltlon,
i,-;=

c
-
)(t ) at .
Solntegratlonofi,-; fromsto+cglves
Underthehypothesesofthetheorem,theorderoflntegratlonmaybereversed(see
Churchl| | ` sbookonceagaln) ,ltfo||owsthat

i,-;a- =

c
-
),; a-a
=
c
-
),;a ,
-
|

o
t
-
ThlsverlnesEq. ( l 2), andEq. ( l 3)fo||owsupon nrstapp|ylngl
|
andthenmu|tl-
p|ylngby A
...... * . ...... 6. =

. =
.=
1. = . =
3. = . = sin
S. = . =

2. = . =

4. =

. = 0Os .,,..........,.
..............
304 Chapter 4 Lapl ace Transform Methods

7. .-
. .

9. . -
.

11. . -
.

.
13. . -
. .

8. . -
. .

10. . -
.

12. .-
. .

..
.
14. . -
..

.
.......,,.....
.....,...
15. .. 16. -

cos
17. =

cos 18. -

sin

sin

cos
19. - 20. -

21. = 22. -

................
....
.
23. .= In

.
.

25. .= In
. .
27. .= I n
.

,
.

24. . = In

. .

26. . -tan-
I

.
28 .

............., ..
,..................(0) -
29. . . .= 0
30. . . .-0
31. . . . .-0
32. . . .-0
33. . ..-0
34. ..

. ..= 0
35. Apply the convolution theorem to show that
.

..-

e.
. .
.....Substitute .= . )
...... .,,.....
.......... ...., ..
,............(0) = .(0) = 0.

36. ...= . - sin .

0
37. ...- . = . t)J
38. ... .-
. -

sin .

0
Termwise Inverse Tansformation of Series
...,... .Operational Mathematics, .
.....,..,,... ......
; 0, ....,...........
C
.

. -_
.

..0 . ............
. ...
C
.

-
..
.,,..........
39. In Example it was shown that

. -_ = + _
.

. .
Expand with the aid of the binomial series and then com
pute the inverse transformation term by term to obtain
Finally, note that = implies that =
40. Expand the function . -.

in powers of .

to show that
41. Show that
.

cos .

.

= .,
Mathematlca| mode|s ofmechanlca| or e|ectrlca| systems often lnvo|ve functlons
wlthdlscontlnultlescorrespondlngtoextema|forcesthataretumedabrupt|yonor
off. One such slmp|eonofffunctlonls the unlt stepfunctlonthatwelntroducedln
X

FIGUR 4.5. 1. The graph of the


unit step function at t = u,

FIGUR 4.5.2. Translation of


J(t ) u units to the right.
4. 5 Peri odi c and Pi ecewise Conti nuous I nput Functi ons 305
Sectlon4. I . Reca||thatthe/sep]./oat= .lsdennedby

0 lf < .,

,;= ,- .;=

f i . _ .
( I )
The notatlon

,;lndlcates succlnct|ywhere the unlt upward stepl nva|uetakes


p|ace(Flg.4. 5 . I ), whereas ,- .;connotesthesometlmes usefu| ldeaofa'tlne
de|ay .beforethesteplsmade.
InExamp|eSofSectlon4. I wesawthatlf.0,then
e
-

1, ,- .; ,=
s
(2)
Because1, , ; , = i }s,Lg. (2) lmp|lesthatmu|tlp|lcatlonof thetransformof,;
o, e
-

correspondstothetrans|atlon- -.lntheorlglna|lndependentvarlab|e.
Theorem l te||susthatthlsfact, whenproper|ylnterpreted,lsagenera|propertyof
theLap|acetransformatlon.
THEOREM 1 Transl ati on on the '-Axis
If1j, ; , exlstsfors > .,then
1, ,- .; ],- .; ,= e

i,s;
and
fors >.+.
Notethat
1

.
, e
-

i,s; ,= ,- .; ],- .;
,- .; ],- .;=

),- .;
lf < ..
l f.
(3a)
(3b)
(4)
ThusTheorem I lmp|lesthat1

,
, e

i,s; , ls thefunctlonwhosegraphfor.
lsthetrans|atlonby.unltstotherlghtofthegraphof),;for0. Notethatthe
part(lfany)ofthegraphof],;tothe|eftof=

ls 'cutoandlsotrans|ated
(Flg. 4. 5. 2). In some app|lcatlons the functlon ],;descrlbes an lncomlngslgna|
thatstartsarrlvlngattlme= 0. Then ,-.; ),-.;denotesaslgna|ofthesame
'shapebutwlthatlmede|ayof.,soltdoesnotstartarrlvlnguntl|tlme* .
ProoJoJTheorem I: Fromthedennltlonof1j,; , , weget
Thesubstltutlon= t +.thenyle|ds
e

i,s; =

],- .;a
306 Chapter 4 Lapl ace Transform Methods
cXump| e 1
cXump| CZ
FromEq. (4) weseethatthlsls the same as
,

i,s; =

,- ., ),- ., 1= .( ,- ., ),- .; , .
because , - ., ), - . = 0 for < . Thls completes the proof of
Theorem I . A
Wlth), = Theorem I glves
.

= ,- ., - ,

.,
-
=

,
I

s 2 ,- . -
lf< .,
lf t .
Flnd .( , , ,lf
lf < 3,
l f 3
(Flg.4. 5. 4).
(Flg.4. 5. 3).

Sol ution Before app|ylngTheorem I , wemustnrstwrlte,, lntheform ,-

, ),- 3) .
cXump| e

The functlon ),,whose trans|atlon 3 unlts to the rlght agrees (for 3) wlth
,, = -ls), = (t+

-because),- 3) = - Butthen
2 - -
i,s; = .,
-
+-+-,= + _ + ,
s s s
sonowTheorem I ylelds

-
.
.
, , ,= .
.
,-
, ),- 3) = ,
i,s;=,
+ _ + .
s s s
Flnd .
.
), , ,lf
](t ) =

s 2t
X
-

lf 0 t < 2,
lft 2
4
(Flg. 4. 5. 5).
X

FIGURE 4.5.3. The graph of


the inverse transfon of
Example I .
FIGURE 4.5.4. The graph of the
function . of Example
FIGURE 4.5.5. The function
of Examples and .
4. 5 Periodi c and Pi ecewi se Conti nuous I nput Functi ons 307
Sol uti on We notenrstthat
cXump| e4
](t ) = l - a (t - 2n) ] cos2t = cos2t- a (t - 2n)cos2(t- 2)
becauseoftheperlodlcltyofthecoslnefunctlon. HenceTheorem l glves
s ( l - e
:
` )
l ](t ) = lcos 2t - e
:
` lcos 2t =
:

s + 4
... . .. . .. .. . . ....... . . . . ... . ... . . .. .. . .
Amassthatwelghs32| b(massm = l s|ug)lsattachedtothefree endofa|ong,
|lghtsprlngthatls stretched lftbyaforceof4 |b,/=4 lbJ).Themasslslnltla||y
atrestlnlts equl|lbrlumposltlon. Beglnnlngattlmet = 0 (seconds), an extema|
force](t ) =cos2t lsapp|ledtothemass, butattlmet =2thlsforcelstumedo
(abrupt|ydlscontlnued) and themassls a||owedtocontlnuelts motlonunlmpeded
Flndtheresu|tlngposltlonfunctlonx (t ) ofthemass.
Sol uti on We needtoso|vethelnltla|va|ueprob|em
x''+4x = ](t ) , x(0) =x
'
(0) =0,
where]( t ) lsthefunctlonofExamp|e3. Thetransformedequatlonls
so
Because
s ( l - e
:
' )
(s
:
+4) X(s) = F(s) =
:

s + 4
-|

s
(
.
2 l
: :
=
1
tsm t
(s + 4)
byEq. ( l 6)ofSectlon4. 3, ltfo||owsfromTheorem l that
x -
x(t ) = tsln 2t- a (t - 2n) (t- 2n)sln 2(t- 2)
= t- a (t - 2n)
.
(t - 2n) ] sln2t .
x - x/l
x - -x/l
ln
FIGURE 4.5.6. The graph of the
function x (t ) of Example .
Ifweseparatethecasest < 2n andt 2n,wenndthattheposltlonfunctlonmay
bewrlttenln theform
tsln2t
x (t ) =
|n sln2t
lft < 2n,
lft 2n.
As lndlcated by the graph ofx(t ) shown ln Flg. 4. 5. 6, the mass oscl||ates wlth
clrcu|ar frequency . = 2 and wlth |lnear|y lncreaslng amp|ltude untl| the force
ls removed at tlme t = 2n. Thereafter, the mass contlnues to oscl||ate wlth the
samefrequency but wlth constantamp|ltudenJ. The force F(t) = cos2t wou|d
produce pure resonance lfcontlnued lndennlte|y, but we seethat ltseectceases
lmmedlate|yatthemomentltlstumedo.
308 Chapter 4 Lapl ace Transform Methods
cXump| C
Sol ution
R
FIGUR 4.5.7. The series ..
circuit of Example
I fwewere to attack Examp|e 4 wlth the methods ofChapter2, we wou|d
need to so|ve one prob|em for the lnterva| < 2 and then so|ve a new
prob|emwlthdlerentlnltla|condltlonsforthe lnterva| 2 Insuchasltuatlon
the Lap|ace transform method enoys the dlstlnct advantage of not requlrlng the
so|utlonofdlerentprob|emsondlfferentlnterva|s.
ConsldertheKLCclrcultshownlnFlg. 4. 5. 7, wlthK = i i !, L = iH,C = i
F,andabattery supp|ylngL =V.Inltla||ytherels nocurrentlntheclrcultand
no charge onthe capacltor. At tlme = the swltch ls c|osedand|eft c|osedfor
i second. At tlme = i lt lsopened and |eft open thereafter. Flnd the resu|tlng
currentlntheclrcult.
Wereca||fromSectlon2.7thebaslcserlesclrcultequatlon
a/

I
L- + K| + -, =e,; ,
a C
(5)
weuse|owercase|ettersforcurrent,charge,andvo|tageandreserveuppercase|et-
tersforthelrtransforms. Wlththeglvenclrculte|ements,Eq (5)ls
a/
a
+ i i c+ i ,=e, ; , ,6)
wheree,; = i - , I ) ] , correspondlng to the openlngandc|oslng ofthe
swltch.
InSectlon2.7ourstrategywastodlfferentlatebothsldesof Eq.(5),thenapp|y
there|atlon

a,
= -
a
toobtalnthesecond-orderequatlon
a
:
/ a/ i

_
L-+ K- + - =e, ;
a
:
a C
(7)
Herewe do notusethatmethod, becausee ,; = exceptat = i , whereas the
umpfrom e,;= when < i toe,; = when ~ i wou|dseemtorequlre
thate , i ; =0 Thuse ,;appearstohaveanlnnnltedlscontlnultyat= i Thls
phenomenonwl|| bedlscussedlnSectlon4. 6. Fornow, wewl|| slmp|y notethatlt
lsanoddsltuatlonandclrcumventltratherthanattempttodea|wlthlthere.
To avold the posslb|e prob|em at = i , we observe that the lnltla| va|ue
,,;=andEq. (7)yle|d,uponlntegratlon,
, , ; =

/ ,r; ar
We substltuteEq. (S)lnEq. (5)toobtaln
a/ i
,

L- +r/ + - / ,r; ar =e,;


a C
a
(S)
,;
Thls ls the integrodiferential equation ofa serles KLC clrcult, lt lnvo|ves both
thelntegra|andthederlvatlveof theunknownfunctlon/, ; TheLap|acetransforn
methodworks we||wlthsuchanequatlon.
4. 5 Peri odi c and Pi ecewise Conti nuous I nput Functi ons 309
Inthepresentexamp|e,Eq.,)ls
Because
a/

a
+ l l c/+ l ccc

/ ,:; a:c l - ,- l ) }


! ,s)
J

/ ,:; a: -
s
-
, l c)
byTheorem2ofSectlon4. 2ontransformsoflntegra|s, thetransformedequatlonls
!,s) c
s ! ,s)+ l i c! ,s)+ l ccc-, l - e
-
;
s s
Weso|vethls equatlonfor!,s)toobtaln
But
sowehave
c, l - e
-
;
! ,s)
s

+ l i cs+ l ccc

c l l
-
s

+ l l cs+ l ccc s+ l c s+ l cc
l l __ l l
!,s)
s+ l c
-
s+ l cc
- e
s+ l c
-
s+ l cc

We nowapp|yTheorem l wlth ),; e


-

- e
-

,thus the lnversetransfornls
/ , ) e
-
,

- e
-
,

- ,- l )
_
e
-
,

-
,
.
e
-
,

-
,

Afterweseparatethecases < land l , wenndthatthecurrentlntheclrcultls


glvenby
e - e
-

-

/ , )
e
-
,

.
e
,

-
,
.
e
-
,

+e
-

-
,

lf < l ,
lf l
The portlone
,

- e
-
,

ofthe so|utlon wou|ddescrlbethecurrentlftheswltch


were|eftc|osedfora||ratherthanbelngopenfor l
31 0 Chapter 4 Lapl ace Transform Methods
'
'
'
'
p
FIGUR 4.5.8. The graph of a
function with period ].
Transforms of Periodic Functions
Perlodlc forclng functlons ln practlca| mechanlca| ore|ectrlca| systems oen me
morecomp|lcatedthanpureslnesorcoslnes. Thenonconstantfunctlon),;denned
for 0ls saldtobeperiodic lftherelsanumberp> 0 suchthat
),+ p; = ),; , i i ;
fora|| 0. The|eastposltlveva|ueofp(lfany)forwhlchEq. , i i ; ho|dslsca||ed
theperiod of). Suchafunctlon ls shown lnFlg. 4. 5. S. Theorem2 slmp|lnesthe
computatlonoftheLap|acetransformofaperlodlcfunctlon.
THEOREM Z Transforms of Periodic Functions
Let),; beperlodlcwlthperlodpandplecewlsecontlnuousfor 0. Thenthe
transformi,s; = 1( ), ; ,exlstsfors > 0andls glvenby
i,s;= e
-st
,;a
I

P
I - e
-PS

Poof: ThedennltlonoftheLap|acetransformglves
!

(
n+l ) p
i,s; = e
-s
t ),; a =
_
e
-s
t ),; a

n=O
n
p
, i z;
Thesubstltutlon= :+plnthenthlntegra|fo||owlngthesummatlonslgnyle|ds

(n+l ) p

P
e
-s
t ),; a= e
-s
(
r
+np)
),:+p;a:= e
-
n
p
s
e
-sr
),:;a:

, ,
because),:+p; = ),:;byperlodlclty. Thus
Consequent|y,
i,s; =

e
-
n
p
s
P
e
-sr
),:;a:
n=O
,
= i +e
-
P
s
_
e
-2p
s
+
. . _
P
e
-s
r
),:;a:
i,s;=
i

P
e
-sr
),:;a:
i - e-
P
S
,
Weusethegeometrlcserles
1
2
3
= I + x + x + x +
. . .
,
I - x
wlth x = e
-
P
s
< i (fors > 0)to sum the serles lnthenna| step. Thus wehave
derlvedEq. ( I 2) . A
Theprlnclpa|advantageofTheorem2 lsthatltenab|esustonndtheLap|ace
transformofaperlodlcfunctlonwlthoutthenecessltyofanexp|lclteva|uatlonof
lmproperlntegra| .
cXump| e
f(t )
. .
.
- - -
a
2
a 3a 4a 5a 6a

, . . .
FIGURE 4.5.9. The graph of the
square-wave function of Example

cXump| e
g(t)
-
-

a
2
a 3a 4a 5a 6a
FIGUR 4.5.10. The graph of
the triangular-wave function of
Example
cXump| eb
4. 5 Peri odi c and Pi ecewi se Conti nuous I nput Functi ons 31 1
Flgre 4. 5. 9 shows me graph ofthe square-wave functlon ),; = (
_
i) [
t
jD of
perlod = 2a, ]x] denotes the greatest lntegernot exceedlng x. By Theoren
2theLap|acetransformof), ; ls
Therefore,
I
,
2
a
F(s) =
2
e
-st
),;a
I - e-
a
s

=
I

e
-s
t
a+
,
2
a
( -l
) e
-s
t
a
I - e-2
a
s

a
=
I

_
!
e
-s
t

_!
e
-s
t

2
a

I - e-2
a
s s

S
a
I - e
-
a
s
F(s ) =
s ( I +e-
a
s )
I
-
a
s
- e
e
a
s
/
2
- e
-
a
s
/
2
I as
= = tanh
s
(e
a
s
/2
+e-
a
s
/2
) S 2
( I Ja)
( I Jb)

Flgure4. 5. I 0showsthegraphofatrlangu|ar-wavefunctlon,,;ofperlod] =2a.


Becausethe derlvatlve , ,; ls the square wave functlon ofExanp|e 6, ltfo||ows
fromtheformu|aln( l 3b) andTheorem2 ofSectlon 4.2 thatthetransfornofthls
trlangu|ar-wavefunctlonls
F(s ) I as
G(s) = = tanh .
s s
:
2
... ~ .. . . . - . _ _ .. _ _ . _ - - -_._--- _ . .. _ . ...
( I 4)

Consldera masssprlngdashpot systemwlth m = I , c = 4, and / = 20 ln ap-


proprlateunlts. Supposethat the system ls lnltla||yat restatequl|lbrlun(x(0) =
x'(0) = 0) andthatthe mass ls acted onbeglnnlngat tlme = 0by theextemal
force),; whosegraphls shownlnFlg. 4. 5. I I . thesquarewavewlthanp|ltude20
andperlod2. Flndtheposltlonfunctlon), ;
Sol uti on Thelnltla|va|ueprob|emls
x+4x'
+20x= ), ; , x(0) =x'
(0) = 0
The transformedequatlonl s
S
2
X(S) + 4s X(s) + 20X(s)= F(s ) . ( I 5)
31 2 Chapter 4 Lapl ace Transform Methods
Jr)
l0
|
|
|
|
|
x lx 1x +x 5x ox
' ' ' '
' ' ' ' '
' ' ' '
' ' '
-l0 +
FIGURE 4.5. 11. The graph of
the external-force function of
Example
FromExamp|e:wlth.=-weseethatthetransformof](t ) ls
sothat
zc l - ,
i,s; = -
s l+,
z+
=

l - ,
, l
a
,
+,-
,
+
. . .
,
s
z+
=

l -
,
+
,-,
+

,,
s
20 40
~
I,s) = - +

_,-i
,

s s

SubstltutlonofEq. , l :)ln Eq. , i ,yle|ds


x,s)
_ i,s;
s

+ +s+zc
, l :)
z+
O

+
,

=
s ,s+z)

+l :}
+z

i ,

s ,s+z)

+l :}

, l )
FromthetransformlnEq. (S)ofSectlon4. J, weget
.

z+

=,-
sln 4t
,s + z)

+ l :

sobyTheoremzofSectlon4. 2wehave
g(t ) = .

zc

,-

sln 4rdr.
s ,s+z) +l :}

Uslngatabu|atedformu|afor],

slnbtdt , weget
g(t ) = l - ,

,cos 4t+sln 4t) = l - h(t ) ,


where
h(t ) =
,

,cos 4t+sln 4t) .


, l s)
, l )
Nowweapp|yTheorem ltonndthelnversetransformoftherlght-handtem
lnEq. , l ) Theresu|tls
~
x |t ) =g|t ) +z_,-i ,- n) g|t - -; . ,zc)

andwenotethatforanynxedva|ueofthesumlnEq. ,

+,ls nnlte. Moreover,


g(t- -;= l -
,-

[cos 4(t- -;+sln 4(t- -; ]


= l -
,-

,-
cos 4t+ sln 4t) .
4. 5 Peri odi c and Pi ecewise Conti nuous I nput Functi ons 31 3
Therefore,
,, n) = i- e
:-
/, ;
Hencel f< < , then
x , ; = i- / , ;
I f < < 2,then
x, ; = i - /,; }- z[ i- e
:
/, ; ] = -i +/,; - z/,;[ i- e]
If 2 < < 3, then
x ,;= i - /, ; } - z[ i- e
:
/,; ]+z[ i- e
-
/,;]
= i+/, ; - z/,;[ i- e
:
+e
-
]
Thegenera|expresslonforn < < (n+i ; -ls
x, ; = /,; +,
.
i ;
-
- z/ , ; [ i- e
:
+

+,
.
i ;
-
e
:-
]
i+,
.
i ;e
:

-
,

= /,;+,-i - z/,;
:
i + e
,zi ;
,zz;
whlch weobtalned wlth thealdofthefaml|larformu|aforthe sumofannltege-
ometrlc progresslon. A rearrangement ofEq. ,zz; nna||y glves, wlth the ald of
Eq., i ;,
e
:
- i
x , ; =
:
e
-
:

(cos :+sln :)+,


.
i ;
-
e
+i
z

,
.
i ;
-
e
:

- e
-
:

-
-

(cos:+
~ sln:)
e
:
+ i
:
forn < < (n+i ; -The nrsttermlnEq. ,z;lsthetranslentso|utlon
,z;
x,,; ,c :; e
-
:

(cos :+sln :) , i i i ;e
-
:

cos ,:- :::; ,z1;


The|asttwo terms lnEq. ,z;glve thesteadyperlodlcso|utlonx, Tolnvestlgate
lt,we wrltet = - n forln thelnterva|n < < (n+i ; -Then
x, ,;= ,
.
i ;
-

i -
:
ze
:

e
-
:

(cos :t +sln :t)


'
e+i
,z;
,
.
i ;
-
[ i- ,z z i ; e
:

cos ,:t - :::;]


Flgure: i zshowsthegraphofx, ,; Itsmostlnterestlngfeature ls the appear-
ance ofperlodlca||y damped oscl||atlons wlth a frequency
)
c/mes that ofthe
lmposed force ), ; InChapterS (FourlerSerles) we wl|| see whyaperlodlcex-
tema|force sometlmes excltes oscl||atlonsatahlgherfrequencythanthelmposed
frequency.
31 4 Chapter 4 Lapl ace Transform Methods
~j
~ j (
. ),c -


\
\
\
/

/
/
/
-
j
~ (
. ),c -
/

Z
FIGURE 4.5.12. The graph of the steady periodic solution for
Example 8; note the "periodically damped" oscillations with
frequency four times that of the imposed force.

|
:
h|
.......,..........
...............,.

1. .

3. .=
2 .

5. .
.

7. .=
.

9. .

.

2. .

4. .
.
.
6. .=

.
8. .
.


10. . =
.

.
.....,.............
....
11. if = if
12. = if .= if or if .
13. sin t if
14. cos m if if
15. sin t if
16. sin 2t if if or if

17. sin m if 2 = if Lr if
18. = cos m if 3 O if t or if t
19. = if = if
20. if t if t
21. if if if

22. *

i f 1 2; = i f I or if 2
23. Apply Theorem with ,= to verify that . .
24. Apply Theorem to verify that . cos . ..


25. Apply Theorem t o show that the Laplace transform of
the square-wave function of Fig. . is

=
.


| | |

| | | | |
a
a 1a +a :a ea
FIGURE 4.5.13. The graph of the
square-wave function of Problem
26. Apply Theorem to show that the Laplace transform of
the sawtooth function of Fig. . .is

.
..

!'
a

a )a +a :a ea
FIGURE 4.5.14. The graph of the
sawtooth function of Problem 26.
4. 5 Peri odi c and Pi ecewi se Conti nuous I nput Functi ons 31 5
27. Let .be the staircase function of Fig. . Show that
.= . where is the sawtooth function of
Fig. . . and hence deduce that

. . =
.


s

+
|
|
) "
|
|

"
|
|
"
a

a )a +a
FIGURE 4.5. 15. The graph of the
staircase function of Problem
28. Suppose that i s a periodic function of period .with
= if , .and = if ., .Find
.
29. Suppose that is the half-wave rectifcation of sin kt ,
shown in Fig. . Show that
.
. =
(S
2

/
_
r

! !
FIGURE 4.5.16. The half-wave rectifcation of
sin .
30. Let .= . . . where i s the
function of Problem and . Note that .=
.is the full-wave rectifcation of sin .shown
in Fig. . Hence deduce from Problem that
. .
. . =
S
2
.

coth
.


k k k
FIGURE 4.5.17. The full-wave rectifcation of
sin .
............ .,.
... ....,.........
....,.....,.........
....,
....= .= . =
.......,..,...
31. = .= . = = if , = U i f

32. = .= .= = if , U if

33. = .= = = sin t if ,,
= if
34. = .= = = if , = U if

35. = .= .= .= if ,, U if

............
RLC .......,
.
L- .

.=
. C
0
.....,.....

36. L = R = C =

= if ,
= if
37. L = R = C =

= if ,
= if
38. L = R = C =
= sin if
, = if t
39. L = R = C =

= if U _ I ;
= i f
40. L = R = C = .

= if ,
= if
........,.....,....
...... ......,..
.= . = .......,..
......,........
.....,..,... .....
.....,..........
.......
= ..

. ,
....... ....
............ ,..
..,...........,.....,.
......,......
................,,..
.....
41. = .= .= is a square-wave function with
amplitude .and period
42. = .= = is a square-wave function
with amplitude 1 0 and period
31 6 Chapter 4 Lapl ace Transform Methods

!1:
ulses and Delta Functions
x
s



Atea = l




+ s
FIGUR 4.6. 1. The graph of the
impulse function .


Consldera force ), ; thatacts on|ydurlng a very shorttlmelnterva| . /.
wlth ),; = 0 outslde thls lnterva| . A typlca| examp|e wou|d be the /-p/s/.e
)o.eofa bat strlklng a ba||the lmpact ls a|most lnstantaneous. A qulcksurge
ofvo|tage (resu|tlngfrom a|lghtnlng bo|t, forlnstance) lsan ana|ogous e|ectrlca|
phenomenon. In such a sltuatlon lt oen happens that the prlnclpa| eectofthe
forcedependson|yontheva|ueofthelntegra|
p=
-
),; a , i ;
anddoesnotdependotherwlseonpreclse|yhow),;varleswlthtlme Thenum-
berp lnEq., i ; lsca||edtheimpulse oftheforce),; overthelnterva|../}
Inthecaseofaforce ),; thatactsonapartlc|eofmass-l n|lnearmotlon,
lntegratlonofNewton` s|aw
yle|ds
a
),; =-c ,;= -c,; }
a

-
a
p= -c, ; } a -c,/;- -c,.;

a
,z;
Thus thelmpu|seoftheforcel sequa|tothechangelnmomentumofthepartlc|e.
Solfchangelnmomentumlstheon|yeectwlthwhlchweareconcemed,weneed
knowon|ythelmpu|seoftheforce, weneedknowneltherthepreclsefunctlon),;
noreventhepreclsetlmelnterva| durlngwhlchltacts. Thls lsfortunate, because
ln a sltuatlon such as that ofabattedba||, weareun|lke|y to have such detal|ed
lnformatlonaboutthelmpu|slveforcethatactsontheba||.
Ourstrategyforhand|lngsuchasltuatlonlstosetupareasonab|emathemat-
lca| mode| lnwhlchtheunknown force),; ls rep|aced wlthaslmp|eandexp|lclt
forcethathasthesamelmpu|se. Supposeforslmp|lcltythat), ; haslmpu|seiand
acts durlng some brlef tlme lnterva| beglnnlng at tlme = . 0. Then wecan
se|ectanxed numbere > 0 that approxlmates the|engthofthls tlmelnterva|and
rep|ace),; wlththespeclncfunctlon
lf. < .+e,
,;
otherwlse.
Thlslsafunctlonof .wlth.ande belngparametersthatspeclfythetlmelnterva|
.. .+e} . If/ .+e, thenwesee(Flg.: : i ; thatthelmpu|seofa

.
over../}
ls
p=

-
a

,
.
,;a=

.
!
a= i

e
Thusa

,
.
has a/lmpu|se,whateverthenumbere may be. Essentla||ythesame
computatlonglves
,:;
|aaet|ea

FIGUR 4.6.2. A diagram


illustrating how the delta function
"sifts out" the value ..
4. 6 I mpul ses and Del ta Functi ons 31 7
Becausethepreclsetlme lnterva| durlngwhlchtheforceactsseensunlnpor-
tant,ltlstemptlng tothlnkofan/s..eos/-p/sethatoccurspreclse|yatthe
lnstant = . We mlghttrytomode|suchanlnstantaneousunltlmpu|sebytaklng
the|lmltasc c,therebydennlng
(5)
where. 0. Ifwecou|da|sotakethe|lmltunderthelntegra|slgnlnEq. (4),then
ltwou|dfo||owthat
,:;
Butthe|lmltlnEq. ,;glves
lf=.,
(7)
lf =.
Obvlous|y, no actua| functlon can satlsfy both ,:; and (7)lfa functlon ls zero
exceptataslng|epolnt,thenlts lntegra|ls not ibutzero. Neverthe|ess,thesynbo|

,;lsveryusefu|. Howeverlnterpreted,ltlsca||edtheDirac delta function at.


aftertheBrltlshtheoretlca|physlclstP. A. M. Dlrac , i z-i s:;,wholntheem|y
i clntroducedafunctlona||eged|yenoylngthepropertleslnEqs. ,:;and(7)
Delta Functions as Operators
Thefo||owlng computatlonmotlvatesthe meanlngthatwe wl|| attach heretothe
symbo|

,; If,,; ls contlnuousfunctlon,thenthemeanva|uetheorenforlnte-
gra|slmp|lesthat

,, ; a= -,, ,
forsomepolntl n.,.+c] . I t fo||owsthat

i
|lm ,,;a

,;a= |lm ,,;

- a= |lm,)=,,.;

c
(S)
by contlnulty of, at = . If

,; ee a functlon ln the strlct senseofthe


dennltlon, and lf we cou|d lnterchange the |lmlt and the lntegra| ln Eq. (S), we
thereforecou|dconc|udethat

,, ;

,;a=,,.; ,;
WetakeEq. ,;astheae]//o, ; ofthe symbo|

, ; A|thoughweca|llt
thede|tafunctlon,ltlsnotagenulnefunctlon, lnstead,ltspeclnestheopem/o
whlchwhenapp|ledtoacontlnuousfunctlon,,;slsoutor se|ects the va|ue
,,.; of thls functlon at the polnt . 0. Thls ldea ls shown schenatlca||y ln
Flg. : : z Notethatwewl|| use the symbo|

,;on|y lnthecontextoflntegra|s
suchasthatlnEq. ,;,orwhenltwl|| appearsubsequent|ylnsuchanlntegra|.
31 8 Chapter 4 Lapl ace Transform Methods
Forlnstance,lfwetake,,; = elnEq. ,;, theresu|tls
Wethereforeae]etheLap|acetransformofthede|tafunctlontobe
Ifwewrlte
1,

,; ,= e
-

,. c;
, ; =

,; and ,- .;=

,; ,
then, i i ; wlth.= glves
1, ,; ,= i .
, i ;
, i i ;
, i z;
, i ;
Notethatlf , ; wereanactua|functlonsatlsfylngtheusua|condltlonsforexlstence
ofltsLap|acetransform,thenEq. , i ;wou|dcontradlctthecoro||arytoTheoremz
ofSectlon: i Buttherels noprob|emhere, , ; ls notafunctlon,andEq. , i ;ls
ourae]//oof1, , ; ,
Delta Function Inputs
Now, nna||y, supposethatweareglvenamechanlca| systemwhoseresponsex,;
totheextema|force),; lsdetermlnedbythedlfferentla|equatlon
~x+ sx+cx= ),; , i :;
Tolnvestlgatetheresponseofthls systemtoaunltlmpu|seatthelnstant = .,lt
seemsreasonab|etorep|ace),; wlth

,;andbeglnwlththeequatlon
~x

+ sx+cx=

,; , i ;
Butwhatls meantbytheso|utlonofsuchanequatlon!Wewl||ca||x,; aso|utlon
ofEq., i ;provldedthat
x,; = |lmx

,; ,

,
, i :;
wherex

,;ls aso|utlonof
~x

+ sx+cx=a

,; , i :;
Because
, i s;
ls an ordlnary functlon, Eq. , i :; makes sense. For slmp|lclty supposethe lnltla|
condltlons to bex,c; = x ,c; = c When wetransformEq. , i :; , wrltlng x

=
1,x

, , wegettheequatlon
cXump| e 1
4. 6 I mpul ses and Del ta Functi ons 31 9
Ifwetakethellmltlnthelastequatlonas--- c,andnotethat
I - e
-
.
llm = I
. s-
byl ' Hpltal ' srule,wegettheequatlon
,~s
:
+ss+c; x,s;= e

,
lf
x,s; = llm x
.
,x;
.
( I 9)
Note thatthl sls precl selythe same resultthatwe would obtaln lfwetransfomed
Eq. ( I 5)dlrectly,uslngthefactthat1,

, ; , = e

Onthlsbaslsltls reasonabletosolveadlerentlalequatlonlnvolvlngade|ta
functlon by employlng the Laplace transform method exactly as lf

,;were an
ordlnary functlon. It l s lmportant to verlfy that the solutlon so obtalned agrees
wlththeone dennedlnEq. ( I 6) , butthlsdependsonahlghlytechnlcalana|yslsof
the llmltlng procedures lnvolved, we conslder lt beyond the scope of the present
dlscusslon. Theformalmethodls valld lnalltheexamplesofthls sectlonandwlll
producecorrectresultslnthesubsequentproblemset.
... .. ................ .... . . ...
A massm = I lsattachedtoasprlngwlthconstant/= :, therelsnodashpot. The
massls releasedfromrestwlthx,c; = Atthe lnstant = z-the massls stmck
wlthahammer,provldlnganlmpulse = sDetermlnethemotlonofthemass.
Sol uti on AccordlngtoProblem i , weneedtosolvethelnltlalvalueproblem
x
+:x= s
:-
,; . x,c;= . x ,c;= c
WeapplytheLaplacetransformtoget
so
s
:
x,s;- s+:x,s;= se
:-

.
s se

:
-

x,s; =
s
:
+:
+
s
:
+:

Recalllngthetransformsofslneandcoslne,as wel| asthetheoremontranslatlons


onthet-axls(Theorem iofSectlon: ;, weseethatthelnversetransfomls
x,; = cosz+: ,- z-;slnz,- z-;
= cosz+:
:-
,;slnz
Becausecosz+:slnz= cos,z- :;wlth: = tan
|
,:}; c z:, sepa-
ratlonofthecases < z-andz-glves
, ;
cos z
x
.a,z- c z:;
lfz-,
lfz-
Theresultlngmotlonls shownlnFlg.: : Notethatthelmpulseat= z-results
lna vlslbledlscontlnulty lntheveloclty at = z-,asltlnstantaneous|ylncreases
theamplltudeoftheosclllatlonsofthemassfromto
320 Chapter 4 Lapl ace Transform Methods
x

o
-
o
0
t " /
ln +n
/
on
FIGURE 4.6.3. The motion of the mass of Example
Delta Functions and Step Functions
Itl susefultoregardthede|tafunctlon

,;asthederlvatlveoftheunltstepfunction

,; Tosee why thls lsreasonable,conslderthecontlnuousapproxlmatlon

, ,;
to

,;shownlnFlg.: : : Wereadllyverlfythat
Because
FIGURE 4.6.4. Approximation
of u
a
(t) by u
a
, . (t ) .
anlnterchangeoflimltsandderlvatlvesyields
cXump| CZ
andtherefore
a

,;=

,;= ,- .;
a
(20)
Wemayregardthlsasthe)orm./ae]//oof thederlvatlveoftheunltstepfunction,
although

,;l snotdlerentlablelntheordlnarysenseat=.
. . . .
WeretumtothericclrcultofExamp|eofSectlon: , wlth r = I I 0 !, i =
I H, c = 0. 00I F, and a battery supplylnge, = 90V. Supposethattheclrcultls
lnltlallypasslvenocurrentandnocharge. Attlme= 0theswltchlsclosedand
attlme= iltlsopenedandleftopen. Flndtheresultlngcurrent/,;lntheclrcult
Sol uti on InSectlon: weclrcumventedthedlscontlnultylnthevo|tagebyemp|oylngtheln-
tegrodlfferentlalformoftheclrcultequatlon.Nowthatdeltafunctlonsareaval|able,
wemaybeglnwlththeordlnaryclrcultequatlon
i/ + r/ +
_
/ =e ,;
c
Inthlsexamplewehave
e,;= 0- c ,- I ) =0- c ,,; .
cXump| C
4. 6 I mpul ses and Del ta Functi ons 321
so,
(t)= --+(t- | ,byEq.(20) . Hencewewant tosolvethelnltlalvalueproblem
+n+ +i +++= --+ ,- | , ,+,=+ ,+,=-+ ,

| ,
Thefactthat ,+,=-+comesfromsubstltutlonoft =+l ntheequatlon
|
L|
' (t)+ ,,+
c
, ,,=
,
,,
wlththenumerlcalvalues ,+,=, ,+,=+and,,+,=-+
Whenwetransformtheproblemln,zi ;,wegettheequatlon
s
-
t,s,- -++n+st,s,+| +++t ,s, = --+,

Hence
-+, | -
,
,
t,s,=
s-+n+s+| +++
Thls lspreclselythesametransformt,s,wefoundlnExampleofSectlon1 . so
lnverslonoft,s,yle|dsthesamesolutlon(t)recordedthere.
Conslderamassonasprlngwlthm =/= |and,+,= ,+,= +Ateachofthe
lnstants t = +-,z-,-, . . . , n, . . . , the mass ls strucka hammer b|ow wlth a
unltlmpulse. Determlnetheresultlngmotlon.
Sol uti on Weneedtosolvethelnltla|valueproblem


O
+=
_`.-
,, ,+,=+= ,+,

.
Because1, ,(t ) = ,

thetransformedequatlonl s
so
O
s
-
x,s, +x,s,= _
,

.
O
,

x,s, = _

.
s +
WecomputethelnverseLaplacetransformtermbyterm,theresultls
O
,,= _a (t - n)sln(t- n-, .

.
Becausesln(t - -; = , -| ,

slnt anda(t - -; = +fort < -,weseethatlt


- < t < (n +| , -,then
,,= sln t - sln t+sln t -

+(-i )sln t ,
thatls,
, , =

0
slnt lfn lseven,
lfn l sodd.
Hence, , ls thehalf-waverectlncatlonofslnt shownlnFlg. 1 : Thephyslca|
explanatlon ls thatthenrsthammerblow(attlmet = +,startsthemassmovlngto

therlght,ustas lt retums to the orlgln, the secondhammerblow stops ltdead, lt


R !n )n +n
FIGURE 4.6.5. The half-wave
rectifcation of sin t .
remalnsmotlonlessuntllthethlrdhammerblowstartsl t movlng agaln, ands oon.
Ofcourse, lfthehammerblowsarenotperfect|y synchronlzedthenthemotlonof
themasswlllbequltedlerent.
322 Chapter 4 Lapl ace Transform Methods
Systems Analysis and Duhamel' s Principle
Conslderaphyslcal systemlnwhlchtheoporesposex(t ) tothe/pfunc-
tlon ),;lsdescrlbedbythedlerentla| equatlon
.x

+/x+cx=), ; , ,zz;
wheretheconstantcoemclents., /,andcaredetermlnedbythephyslca|parameters
ofthesystemandarelndependentof),; Themass-sprlng-dashpotsystemandthe
serlesRLC clrcultarefamlllarexamp|esofthlsgeneralsltuatlon.
Forslmpllcltyweassumethatthesysteml slnltlallypasslve. x,c; =x ,c;
*
c Then thetransformofEq. ,zz;l s
so
Thefunctlon
.s
:
x,s; +/sx,s;+cx,s; =i,s; ,
i,s;
x,s; =
:
= u,s; i,s;
.s + /s+c
I
u,s; =
.s
:
+/s+c
,z;
,z1;
lscalledthetransfer function ofthesystem. Thusthetransformoftheresponseto
the lnput),;ls theproductofu,s; and the transform i,s;
Thefunctlon
a,;=1

,
, u,s; , ,z;
ls calledtheweight function ofthesystem. FromEq. ,z1;weseebyconvolutlon
that
x,; =

a,:; ),- :;a: ,z:;


Thls formula ls Duhamel's principle forthe system. What ls lmportant ls that
thewelghtfunctlona,; lsdetermlnedcompletelybytheparametersofthesystem.
Oncea,;hasbeendetermlned,thelntegralln,z:;glvestheresponseofthesystem
toanarbltrarylnputfunctlon), ;
I nprlnclplethat l s , vla the convolutlon lntegralDuhamel ' sprlnclp|ere-
ducestheproblemofnndlngasystem' soutputsforallposslblelnputstocalculatlon
oftheslnglelnverseLaplacetransformln,z;thatlsneededtonnd ltswelghtfunc-
tlon. Hence,acomputatlona|analogueforaphyslcalmasssprlng-<ashpotsystem
descrlbedby,zz;canbeconstructedlntheformofablackboxthatlshard-wlred
tocalculate(andthentabulateorgraph,forlnstance)theresponsex,;glvenby,z:;
automatlcally wheneveradeslredforcefunctlon),;lslnput. Inenglneerlngprac-
tlce,allmannerofphyslcalsystemsaremodeledlnthl smanner,sothelrbehavlors
canbestudledwlthoutneedforexpenslveortlme-consumlngexperlmentatlon.
cXump| e4
4. 6 I mpul ses and Del ta Functi ons 323
Conslderamasssprlngdashpotsystem(lnltlallypasslve) thatrespondstotheex-
temalforce),;lnaccordwlththeequatlonx+:x+ i cx= ),; Then
I I
u,s;


s

+:s+ I 0

,s+

+ i

sothewelghtfunctlon ls a,; = e

sln t . Then Duhame| ' sprlnclplelmp|lesthat


theresponsex ,;totheforce),;l s
Notethat
x,;=

(sln:; ),- :;a:

u,s; =
I
=
1, ,; ,

.s

+/s+c .s

+/s+c

Consequently, lt follows from Eq. ,z;that the welghtfunctlon ls slmply the re-
sponseofthesystemtothe deltafunctlonlnputo (t ) . Forthlsreason a,;ls some-
tlmescalledtheunit impulse response. Aresponsethatlsusuallyeaslertomeasure
ln practlce ls theresponse/,;to theunltstep functlon , ; , /,;ls theunit step
response. Because1, , ; , = i}s, weseefrom Eq. ,z;thatthetransformof/,;
ls
u,s;=
u,s;

s
Itfollowsfromtheformulafortransformsoflntegralsthat
/ ,;=

a,:;a:. sothat a,;=/ ,; ,z:;


Thusthewelghtfunctlon,orunltlmpulseresponse,lsthederlvatlveoftheunltstep
response. Substltutlonof,z:;lnDuhamel ' sprlnclpleglves
x,; =

/ ,; ),- :; a: ,zs;
fortheresponseof thesystemtothelnput), ;
ApPLICATIONS : To descrlbeatyplcal appllcatlonofEq. ,zs;, supposethatwe
areglvenacomplex serlesclrcultcontalnlng many lnductors, reslstors, andcapac-
ltors. Assumethat lts clrcultequatlon ls allnearequatlonoftheformln,zz;,but
wlth / ln place ofx What lfthe coemclents ., /, and c are unknown, perhaps
onlybecause they are too dlfncultto compute! We would stlll want to know the
current/ ,;correspondlngto any lnput ),; = e , ; Weconnectthe clrcultto a
llnearlylncreaslngvoltagee,;= ,sothat),;= e ,;= I = , ; , andmeasure
the response /,;wlth an ammeter. We then compute the derlvatlve / ,;. elther
numerlcallyorgraphlcally. ThenaccordlngtoEq. ,zs;, theoutputcurrent/,;cor-
respondlngtothelnputvoltagee,;wlllbeglven by
/,;=

/,:; e,- :;a:


(uslngthefactthat),;= e ,;
324 Chapter 4 Lapl ace Transform Methods
HISTORICAL REMARK: Inconcluslon,weremarkthataround i c, afterengl-
neersandphyslclsts hadbeenuslngdeltafunctlonswlde|yandnultfullyforabout
zcyearswlthoutrlgorous]ustlncatlon,theFrenchmathematlclanLaurentSchwartz
developedarlgorousmathematlcaltheoryof,eem//,ea)./osthatsupplledthe
mlsslngloglcalfoundatlonfordeltafunctlontechnlques. Everyplecewlsecontlnu-
ousordlnaryfunctlonlsagenerallzedfunctlon,butthedeltafunctlonlsanexample
ofagenerallzedfunctlonthatls notanordlnaryfunctlon.
.....,...1 .......
.,..........
1. ...= .= . =
2. ...= .= . =
3. .....= .= . =
4. . . .= .= . =
5. ...= .= . =
6. ..= cos .= . =
7. . .. .= .= . =
8. ...= .= . =
,,.... .,.,.........
.........,......
12.
9. ...= .= . =
10. ...= .= . =
11. ...= .= . =
12. ....= .= . =
13. This problem deals with a mass initially at rest at the
origin, that receives an impulse ,at time = (a) Find
the solution . of the problem
.= ,. .= . =
(b) Show that l i m .agrees with the solution of the

,
problem
.= , .= . =
(c) Show that = ,for ~ = ...
14. Verify that . .= .by solving the problem
.= . .=
to obtain .= . .
15. This problem deals with a mass on a spring (with con
stant .that receives an impulse .= at time =
Show that the initial value problems
and
...= .= . =
...= . .= . =
have the same solution. Thus th effect of . is, in
deed, to impart to the particle an initial momentum .

16. This is a generalization of Problem Show that the


problems
....= .= . =
and
....= + . .= . =
have the same solution for ~ Thus the efect of the
term . is to supply the initial condition . =
17. Consider an initially passive .circuit (no inductance)
with a battery supplying volts. (a) If the switch to
the battery is closed at time = .and opened at time
= ~ .(and lef open thereafer), show that the current
in the circuit satisfes the initial value problem

= . =
(b) Solve this problem if .= , =

F,
= V, .= (s), and = (s). Show that ~
if and that i if ~
18. Consider an initially passive .circuit (no resistance)
with a battery supplying volts. (a) If the switch is
closed at time = and opened at time = ~ show
that the current in the circuit satisfes the initial value prob
lem

= .
= =
(b) .= H, =

F, = V, and .= (s),
show that
'
(
Sin
l =
0
if
if t ~
Thus the current oscillates through fve cycles and then
stops abruptly when the switch is opened (Fig. .
i(/)
|
-l
V V V V
FIGURE 4.6.6. The current function of
Problem
19. Consider the .circuit of Problem 1 8(b), except suppose
that the switch is alterately closed and opened at times
t (a) Show that satisfes the
initial value problem
O

_
( -I )
n
8 t -
.
, =
n=O

(b) Solve this initial value problem to show that
.+ sin if
. .+
- < t < .

Thus a resonance phenomenon occurs (see Fig. .
i(/)
l 0
-l 0
FIGURE 4.6.7. The current function of
Problem
20. Repeat Problem except suppose that the switch i s al
terately closed and opened at times t
. Now show that if
then
. .+
- < t <

=

i n if .is even;
if .is odd.
Thus the current i n alterate cycles of length frst ex
ecutes a sine oscillation during one cycle, then i s dormant
during the next cycle, and so on (see Fig. .
4. 6 I mpul ses and Del ta Functi ons 325
i(/)
FIGURE 4.6.8. The current function of
Problem
21. Consider an ..circuit in series with a battery, with
.H, .= , -3 F, and eo = V. (a)
Suppose that the switch is alternately closed and opened
at times t . Show that i (t) satisfes
the initial value problem
O

.
, + + _( -l t8 t -
n=O

=
(b) Solve this problem to show that if
then
. .+
- < t < -- .. --

e3
mr+
3
1
-
= e-30
t
sin
e3
1
-
Construct a fgure showing the graph of this current func
tion.
22. Consider a mass m = on a spring with constant k
initially at rest, but struck with a hammer at each of the in
stants t .. . . . Suppose that each hammer blow
imparts an impulse of + Show that the position function
(t) of the mass satisfes the initial value problem
O

/
+ X =
_
8 (t - . . ()

,
e
Solve thi s problem to show that if .< t < .+ l ) r,
then (t .+ sin t . Thus resonance occurs because
the mass is struck each time it passes through the origin
moving to the right-in contrast with Example in which
the mass was struck each time it retured to the origin. Fi
nally, construct a fgure showing the graph of this position
function.
Linear S
y
stells
of Differential
Equations
First-Order ystems and Aplicatios
326

ntheprecedlngchapterswehavedlscussedmethodsforsolvlnganordlnarydlf-
ferentlalequatlonthatlnvolvesonlyone dependentvarlable. Many appllcatlons,
however, requlrethe use oftwoormoredependentvarlables,eachafunctlonofa
slngle lndependentvarlable (typlcally tlme). Sucha problem leads naturally to +
syse-ofslmultaneousordlnarydlerentlalequatlons. Wewlllusually denotethe
lndependentvarlablebyandthedependentvarlables(theunknownfunctlonsof;
byx[ , x , x
)
, orbyx, y, ,, . . . . Prlmeswllllndlcatederlvatlveswlthrespectto

We wlllrestrlctourattentlonto systemsl nwhlchthenumberofequatlonsls
thesameasthenumberofdependentvarlables(unknownfunctlons). Forlnstance,
a system oftwo nrst-order equatlons ln the dependent varlables x and y has the
generalform
), , x. y. x .y;=0,
,, . x. y. x .y;=0,
( I )
wherethefunctlons)and,areglven. Asolution ofthlssysteml sapalrx, ; . y,;
offunctlonsofthatsatlsfybothequatlonsldentlcallyoversomelntervalofvalues
of
Foranexampleofasecond-ordersystem,conslderapartlcleofmass-that
moveslnspaceunderthelnuenceofaforceneldF thatdependsontlme ,thepo-
sltlon,x ,; .y,; ., ,; ofthepartlcle,andltsveloclty,x , ; . y ,; ., ,; Applylng
Newton' slaw-a= F componentwlse,wegetthesystem
-x

= i,, .x. y.,.x .y ., ; ,


-y

= i
:
, .x. y. ,. x .y ., ; .
-,= i

, .x. y. ,, x .y ., ;
,z;
Exampl e 1

AAA
- -

.
La||||t|ames|t|eas
FIGURE 5. 1. 1. The mass
and-spring system of Example I .
,

,
_

_
.
, x)

FIGURE 5. 1.2. The "free


body diagrams" for the
system of Example 1 .

_
c a|/m|a
|tesawatet
|
Exampl e Z
_
ca|/m|a
FIGURE 5. 1.3. The two brine
tanks of Example
5. 1 Fi rst-Order Systems and Appl i cati ons 327
ofthreesecond-orderequatlonswlthlndependentvarlabletanddependentvarlables
x, y, z; the three rlght-hand-slde functlons i, , F
:
, F

are the components ofthe


vector-valuedfunctlonF.
Initial Applications
Examples I throughfurtherlllustratehowsystemsofdlerentla|equatlonsarise
naturallylnsclentlncproblems.
.. .... .. .. . . ....... .... ..... ......... .
Conslder the system oftwo masses and two springs shown ln Flg. 5. l . I , wltha
glven extemal force ](t ) actlng on the rlght-hand mass m
:
. We denote by x(t)
thedlsplacement(totherlght) ofthemass-, from lts statlcequlllbrlumposltlon
(when thesystem ls motlonlessandlnequlllbrlumand ](t ) = 0)andby ,(t)the
dlsplacement of the mass m
:
from lts statlc posltlon. Thus the two sprlngs are
neltherstretchednorcompressedwhenxand yarezero.
IntheconnguratlonlnFlg. 5. I . I , thenrstsprlng ls stretched x unlts andthe
secondby y- xunlts. We applyNewton' slaw ofmotlon to the two 'freebody
dlagramsshownlnFlg. 5 . I . 2, wetherebyobtainthesystem
-, x=-/, x+/
:
,y- x; .
-_y =-/
:
,y- x; +](t )
,;
ofdlfferentlalequatlonsthattheposltlonfunctlonsx,; andy,;mustsatlsfy. For
lnstance, lf-, = 2, -_ = I , /, = 4, /
:
= 2, and ](t) =40sln lnapproprlate
physlcalunlts,then the systemln (3)reducesto
zx

=-:x+zy.
y

=zx- zy+40 sln 3t .


(4)

Conslder two brlne tanks connected as shownlnFlg. 5. I . 3. Tank I contalnsx,;


poundsofsaltln I 00galofbrlneandtank2contalnsy,;poundsofsaltln200ga|
ofbrlne. The brlne ln each tank ls kept unlform by stlrrlng, and brlne ls pumped
fromeachtanktotheotherattherateslndicatedlnFlg.5. I . 3. Inaddltlon,freshwa-
terowslnto tank I at20galJmln,andthebrlnelntank2owsoutat20galJmln
(so thetotal volume ofbrlnelnthetwotanksremalnsconstant) . Thesaltconcen-
tratlonslnthetwotanksarexJI 00poundspergallonandyJ200poundspergallon,
respectlvely. Whenwecomputetheratesofchangeoftheamountofsaltlnthetwo
tanks, we thereforeget the systemofdlerentlalequatlonsthatx,;andy,;must
satlsfy.
thatl s,
, x y 3 I
x = -c + i c = --x + -y
I 00 200 I 0 20

, x y y 3 3
y =30 -- I 0
.
-- 20
.
-= -x- -y

I 00 200 200 I 0 20

zcx=-:x+y.
zcy=:x- y
(5)

328 Chapter 5 Li near Systems of Di fferenti al Equati ons

-a
Inductor
Re
s
istor
Capacitor
cXump| e


.
.
.
.

FIGURE 5. 1.5. Voltage drops


across common circuit elements.
L :aeat|es

|
R
|
( ccve|ts :ceams
C. c.ccstataes
FIGURE 5. 1.4. The electrical network of Example
Conslder the e|ectrlcal network shownl nFlg. i :,where I,, ; denotes the cur-
rent ln the lndlcateddlrectlonthroughthe lnductoriand j,; denotesthecurrent
throughthereslstorr
:
Thecurrentthroughthereslstorr, ls I =( - jlnthedl-
rectlonlndlcated.Wereca||Klrchho' svo|tage|awtotheeectthatthe(a|gebralc)
sum ofthe vo|tage dropsaroundanyc|osed|oopofsuch anetworkls zero. Asln
Sectlon z :, the vo|tage drops acrossthe three types ofclrculte|ements are those
shownlnFlg. i Weapp|yKirchho' s|awtothe|eft-hand|oopofthenetwork
toobtaln
a(
z-+c,/- j; - i cc=c.
a
,:;
because the vo|tage drop from the negatlve to the posltlve po|e ofthe battery ls
-i ccTherlght-hand|oopyle|dstheequatlon
i z _
:
+zj+c,j () =c. ,:;
where _
:
,;ls thechargeonthecapacltor. Becausea_
:
,a = jdlerentlatlon
ofeach sldeofEq. ,:;yle|ds
ai
:
ai,
i z j+: -- c-=c
a a
(S)
AfterdlvldlngEqs. ,:;and(S)bythefactors zand -z, respectlve|y, wegetthe
system
ai,
+zi
(
- zj=c.
a
a( aj
z-- -- j= c
a a
ofdlfferentla|equatlonsthatthecurrents(,;andj, ; mustsatlsfy.
(9)

cXump| e4
5. 1 Fi rst-Order Systems and Appl i cati ons 329
First-Order Systems
Ceas|ae:+sys:emeia|iie:ea:|+iea+:|eas:a+:.+a|eseiveaie::aea|,aes:e:a:
ae:|v+:|ves ei:ae aeeaaea: v+:|+|ies :a+: +e+:, +sesi|.|:iaa.:|easei +aa
iewe:e:ae:ae:|v+:|vesei:aeaeeaaea:v+:|+|ies te:|as:+a.e, |a:ae.+seei+
sys:emei:wese.eaae:ae:ea+:|eas, ea:+ssam:|ea|s:a+:|:.+a|e:|::ea|a:ae
ie:m
, i o)
i:|sei|e:a:+.:|.+i+aa:aee:e:|.+i|me::+a.e:a+:+aysa.aa|,ae:e:ae:sys:em
.+a|e::+asie:mea|a:e+aea|v+iea:sys:emei)soaeea+:|eas
1e aes.:||eaewsa.a+::+asie:m+:|ea|s+..emi|saea,e.eas|ae:a:s::ae
sys:em.eas| s:|a,ei:aes|a,ie:ae:ae:ea+:|ea

),
,

-
-
,

x = , x, x , . , x
we|a::eaa.e:aeaeeaaea:v+:|+|iesx, ,x
:
, ,x
-
aeaaea+sieiies
x, =x, x
:
=x . x

=x
. x
-
=x

-
-
,

, l l )
, i z)
Ne:e:a+:x =x =x
:
,x=x=x

,+aaseea uea.e:aesa|s:|:a:|eaei, l z;|a


, l l ) y|eias:aesys:em
, i )
,
x
-

=x
-
.
x=), , x, ,x
:
, . . .x
-
;
ei)soaeea+:|eas v|aea:iy,:a|ssys:em|sea|v+iea::e:aee:|,|a+i:a
e:ae:ea+:|ea|a, l l ),|a:aesease:a+:x,;|s+seia:|eaei, i l ) |i+aaeaiy|i
:aeiaa.:|easx,, ; , x
:
, ; , ,x
-
,;aeaaea|a, l z)s+:|siy:aesys:emeiea+:|eas
|a, l ;
1ae:a|:ae:ae:ea+:|ea
x

+x

+zx- x=s|az
|sei:aeie:m|a, l l ) w|:a
), ,x.x .x

;=x- zx- x

+s|az
uea.e:aesa|s:|:a:|eas
x, = x,
I / H /
x
:
=x =x
,

=x =x
:
y|eia:aesys:em
,
x
,
= x
:
,
,
x
:
=x

,
x=x, - zx
:
- x

+ s|az
ei:a:eea:s:e:ae:ea+:|eas
330 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| e
i:m+y+e+::a+::aea:s:e:ae:sys:eme|:+|aea |a s+mie:eiie:s i|::ie
+av+a:+,e|e.+asewe.eaiaase:aeme:aeaseiCa+:e:z:eseive:aee:|,|a+i,i|a
e+:):a|:ae:ae:ea+:|ea sa:saese:a+:wewe:e.eai:ea:eaw|:a:aeaeai|ae+:
ea+:|ea
x

=x

+ ,x;

,
:ewa|.aaeaeeiea:e+:i|e:me:aeas.+a|e+i|ea1ae.e::eseaa|a,a:s:e:ae:
sys:em|s
, l :)
+aawew|ii see |a se.:|ea: ::a+::ae:ees|s:ene.:|ve aame:|.+i:e.aa|aesie:
+:es|m+:|a,:aeseia:|eaeiessea:|+iiy+aya:s:e:ae:sys:em se|a:a|s.+se:ae
::+asie:m+:|ea:e+a:s:e:ae:sys:em/s+av+a:+,eeas t:em+:+.:|.+iv|ee|a:,
i+:,esys:emseia|,ae:e:ae:a|iie:ea:|+iea+:|eas:y|.+iiy+:eseiveaaame:|.+iiy
w|:a:ae+|aei:ae.ema:e:,+aa:aea:s:s:e|s:e::+asie:msa.a+sys:em|a:e+
a:s:e:ae:sys:emie:wa|.a+s:+aa+:a.ema:e::e,:+m|s+v+|i+|ie
1aesys:em
zx=-:x+ zy,
y= zx - zy+ :c s|a
,:)
eise.eaae:ae:ea+:|easw+sae:|vea|as+miel 1:+asie:m:a|ssys:em|a:e+a
ea|v+iea:a:s:e:ae:sys:em
Sol uti on He:|v+:ea|y:aeea+:|eas|a, l z) , weaeaae
x, = x, y, =y,
1aea:aesys:em| a,:)y|eias:aesys:em
,
x
,
= x, ,
zx=-:x, +zy, ,
y
,
=y, ,
, ,
y,= y =y
,

y=zx, - z
y,+:c s|a
, l )
eiiea:a:s:e:ae:ea+:|eas|a:aeaeeaaea:v+:|+|iesx, ,x,,y, ,+aay,
Simple Two-Dimensional Systems
1aei|ae+:se.eaae:ae:a|iie:ea:|+iea+:|ea
x

+px+,x=c , l :)
,w|:a.eas:+a:.eeia.|ea:s+aa|aaeeaaea:v+:|+|ie;::+asie:msv|+:aesa|s:|:a
:|easx=y,x=y|a:e:ae:wea|meas|ea+ii|ae+:sys:em
,
x = y,
,
y = -,x- py
, l )
cXump| e
X
FIGUR 5. 1.6. Direction feld
and solution curves for the system
A
/
= = !X of Example 6.
5. 1 Fi rst-Order Systems and Appl i cati ons 331
Ceave:seiy,we.aaseive:a|ssys:em|a, l )|yseiv|a,:aeiam|i|a:s|a,ieeaa:|ea
|a, l :)
1eseive:ae:wea|meas|eaaisys:em
we|e,|aw|:a:|eeise:va:|ea:aa:
x

= -zy,
'

l
y

x,
x= -zy= -z |x,= -x
1||s,|ves:|es|a,iese.eaae:ae:eaa:|eax+x =cw|:a,eae:aiseia:|ea
x , ) = n.es+ ss|a= c.es,- a)
wae:en= c.es aaaas= cs|a a 1aea
y, ) = -x , ) = - , -n s|a +s .es )
= cs| a,- a)
, l s)
1|e|aea:|:y.es

J +s|a

J = l:ae:eie:e|mi|es:aa:,ie:ea.avaiaeei , :aee|a:
,x, ) , y, ) ) i|esea:|eeii|se
w|:asem| asescaaac}zt|,a:e l :saewsseve:aisa.aeii|ses|a:aexyiaae.

Aseia:|ea,x, ) , y, ) ) eia:wea|meas|eaaisys:em
x= ), , x, y) ,
y= y, ,x, y)
mayie:e,a:aeaasaa:ame::|za:|eaeiasolution curve e:trajectory ei:|esys
:em|a:aexyiaae. 1|as:|e::a] e.:e:|esei:|esys:em|a, l s)a:e:aeeii|sesei
r|,. l : 1|e.ae|.eeiaa|a|:|aie|a:,x,c),y,c))ae:e:m|aeswa|.aeaeei:aese
::a] e.:e:|esaa::|.aia:seia:|eaa:ame::|zes.
1|e |.:a:e saew|a, a sys:em s ::a] e.:e:|es |a :aexyiaae-|:s se.aiiea
p/.sep/.epom/-ia|is:e :eveai :e.|seiy aew :ae e|a: ,x,) ,y, ) ) meves
aiea,|:s::a]e.:e:yii:|eiaa.:|eas)aaayaeae:|aveive:ae|aaeeaaea:va:|a|ie
, :|ea adirection held-saew|a,:y|.ai a::ews :e:esea:|a,ve.:e:s w|:|.em
eaea:s,:ee::|eaai:e):|eae:|va:|vesx = ),x, y) aaay = y,x,y)-.aa|e
ie::ea se.aase:aemev|a,e|a: ,x, ) , y, ) ) aas veie.|:y ve.:e: ,x ,) ,y ,) ) ,
:a|s a|:e.:|ea aeia |aa|.a:es :aee|a: sa|:e.:|eaeime:|ea aiea, |:s ::a]e.:e:y
re:|as:aa.e,:aea|:e.:|eaaeiaie::ea|ar|,. l :|aa|.a:es:aa:ea.asa.ae|a:
meves.eaa:e:.ie.iw|sea:eaaa|:seii|:|.ai::a] e.:e:y.Aaa|:|eaai|aie:ma:|ea.aa
ies|ewa|a:aesea:a:e,:aaseix , ) aaay,)asiaa.:|easei

332 Chapter 5 Li near Systems of Di fferenti al Equati ons


+
J
Z
l
cXump| e
COlI|luCO
cXump| e
0M
- l
-
l
- J
x
FIGURE 5. 1.S. Direction feld
and solution curves for the system
.= y, y' = .+ y of Example
w|:a|a|:|+iv+iaesx,c)=z, y,c)=c, :ae,eae:+iseia:|ea|as+mie:y|eias
x,c)=~=z, y,c)=- s=c
1ae:esai:|a,+::|.ai+:seia:|ea|s,|veaiy
x , ) =z .es , y, ) =s|a
1ae,:+|sei:ae:weiaa.:|eas+:esaewa| at|, l wesee:a+:x, ) |a|:|+iiy
ae.:e+ses wa|ie y, ) |a.:e+ses

i:ieiiews:a+:, +s |a.:e+ses, :ae seia:|eae|a:


,x , ) , y, ; ::+ve:ses:|e::+]e.:e:yx
:
+y

= i|a:ae.eaa:e:.ie.iw|sea|:e.:|ea,
+s|aa|.+:ea|y:aea|:e.:|eaaeiave.:e:s|at|, l :
+ -
J
Z
x = Z cos /
0m
- l
-Z
- J
}
= s n/
-+
l
FIGURE 5. 1.7. . and y-solution curves for the initial
value problem .= y' =

..= =
... .. ......... ........ . .. . .. ..,. . ... . .. ... ..
1eaaa+,eae:+iseia:|eaei:aesys:em
x = y,
y=zx+y,
we|e,|aw|:a:aee|se:v+:|ea:a+:
x=y =zx+y=x+zx
1a|s,|ves:|es|a,iei|ae+:se.eaae:ae:ea+:|ea
x- x- zx=c
w|:|.a+:+.:e:|s:|.ea+:|ea

:
- z= ,+ i , ,- z)=c
+aa,eae:+iseia:|ea
x , ) =~e
-
+ se
:

1aea
y, ) =x ,)=-~e

+zse
:

, l )
,zo)
,zl )
1y|.+ia+sei+ae::+]e.:e:|esei:|esys:em| a, l )+:+me::|zea|ys ,zc)+aa
,zl ) +:e s|ewa |at|, l s 1aese::+] e.:e:|esm+y:esem|ieaye:|ei+ssa+:|a,
.emmea+sym:e:es, |a:r:e|iemzsaews:a+::ae|:+.:a+iie:m|ssemewa+:me:e
.emi|.+:ea
cXump| eb
X
5. 1 Fi rst-Order Systems and Appl i cati ons 333
1eseive:ae|a|:|+iv+iae:e|iem
x =-y,
y = , l cl )x- ,c z) y,
x,c)= c, y,c) = -i ,
we|e,|aw|:a:|ee|se:v+:|ea:a+:
x=-y= - , l cl )x- ,c z) y[ = , -l cl )x- ,c z)x
1||s,|ves:aes|a,iei|ae+:se.eaae:ae:ea+:|ea
x+,o z)x+, l cl )x=c
w|:a.a+:+.:e:|s:|.ea+:|ea

:
+,c z)+l cl =,+c i )
:
+l = c,
.|+:+.:e:|s:|.:ee:s-c l/ . +aa,eae:+iseia:|ea
x, ) =e
-
,

,~.es +ss|a )
1aeax ,c)=~=c,se
x , ) = se

,

s|a ,
,zz)
FIGUR 5. 1.9. Direction feld y, ) = -x,) = se

,

s|a- se
-
, ,
.es
and solution curve for the system
. = -y, y' = . of
t|a+iiy,y,c)= -s=-l , se:aeaes|:easeia:|eaei:aesys:em|a,zz)|s
Example
0.
0. +
0.0M

-0. +
-0.
," ,

- l . l
0 l0 l l0 l 10
FIGUR 5. 1. 10. . and
y-solution curves for the initial
value problem of Example
x, ) =e
-
,

s|a ,
,z)
y, ) = e
-
, ,
,s|a - l c .es )
1aeseea+:|eas+:+me::|ze:aes|:+i::+] e.:e:y| at|, l , :|e ::+]e.:e:y+
:e+.|es:aee:|,|a+s +~ t|,a:e l l csaews:aex +aayseia:|ea.a:ves
,|vea|a,z)
waeawes:aayi|ae+:sys:emsas|a,:aee|,eav+iaeme:aeaeise.:|ea:, e
w|iiie+nway :aesae:a.|+iiys|m|i+:sys:ems|as+mies::a:ea,asa+ve:ae
m+:ieaiya|ne:ea:::+] e.:e:|ess|ewa|at|,s l :, l s, +aa l
Linear Systems
ia+aa|:|ea:e:+.:|.+i+av+a:+,esie:aame:|.+i.ema:+:|ea,:ae,eae:+i:aee:yei
sys:ems+aasys:em+:|.seia:|ea:e.aa|aes+:eme:ee+s|iy+aame:e.ea.|seiyae
s.:||eaie:a:s:e:ae:sys:ems:a+aie:a|,|e:e:ae:sys:emste:|as:+a.e,.eas|ae:
+//eca:s:e:ae:sys:emei:aeie:m
x
=;, , ,;x, +;
:
,;x,+

+;, -
x
-
+), ,) ,
x =
;
:
,;x, +;
::
,;x
:
+

+;:-
x
-
+
!
,) ,
,z:)
x
=
;- ,,;x, +;-, ,;x
:
+

+;--
x
-
+)
-
,;
334 Chapter 5 Li near Systems of Di fferenti al Equati ons
we s+y:a+::a|s sys:em|shomogeneous |i:ae iaa.:|eas
), , . . )-
a:e +ii
|aea:|.+iiyze:e,e:ae:w|se, |:|snonhomogeneous. 1aas:aei|ae+:sys:em|a,)|s
aeme,eaeeas,wae:e+s:aei|ae+:sys:em|a, l )|saeaaeme,eaeeas 1aesys:em
|a, l :)|saeai|ae+:|e.+ase:ae:|,a:a+aas|aeei:aese.eaaea+:|ea|sae:+i|ae+:
iaa.:|eaei:aeaeeaaea:v+:|+|iesx, +aax
:

Asolution ei:aesys:em|a,z:) |s +a:aieeiiaa.:|easx,, ; . x


:
,; . .
x
-
,;:a+:,easeme|a:e:v+i)|aea:|.+iiys+:|siye+.aei:aeea+:|eas|a,z:)we|ii
see:a+::ae,eae:+i:aee:yei+sys:emeii|ae+:a:s:e:ae:ea+:|eassa+:esm+ay
s|m|i+:|:|esw|:a:ae,eae:+i:aee:yei+s|a,ie:ae:ae:i|ae+:a|ne:ea:|+iea+:|ea
1aee:em i ,:evea|a:aeAeaa|s)|s+a+ie,eas:e1aee:emzeise.:|eaz z i:
:eiisas:a+:|i:ae.eeia.|ea:iaa.:|eas;
,
+aa |a,z:)+:e.ea:|aaeas, :aea:ae
sys:em|+s+aa|aeseia:|eas+:|siy|a,,|vea|a|:|+i.eaa|:|eas
THEOREM 1 Existence and Uni queness for Li near Systems
saese:|+::|eiaa.:|eas;, , , ;
:
, , ;--
+aameiaa.:|eas]. , . )-
+:e.eat|aaeasea:|eeea|ate:v+iI.ea:a|a|a,:|ee|a:O. 1|ea, ,|vea:ae
aamie:s/, , /
:
, ,/, ,:|esys:em|a,z:) |+s +aa|aeseia:|eaea:aeea:|:e
|a:e:v+iI:|+:s+:|saes:|e|a|:|ai.eaa|:|eas
,z)
1aas |a|:|+i .eaa|:|eas +:eaeeaea:e ae:e:m|ae+ seia:|eaei+sys:emei
i|ae+:a:s:e:ae:ea+:|eas, +aa we:ae:eie:eese.:+,eae:+iseia:|eaeisa.a+
sys:em:e|aveive+:||::+:y.eas:+a:ste:|as:+a.e,wes+w|as+mie:a+::ae
se.eaae:ae:i|ae+:sys:em
zx= - :x+ zy,
y= zx - zy+ :cs|a .
wa|.aaes.:||es:aees|:|eaiaa.:|easx,; +aay,; eis+miel , |sea|v+iea::e
:aesys:emei)ca:s:e:ae:i|ae+:ea+:|eas|a , l ) uea.eiea:|a|:|+i.eaa|:|eas
weaia|eaeeaea:eae:e:m|ae:aesa|seaea:me:|easei:ae:wem+sses|as+mie
l 1y|.+i|a|:|+iv+iaesweaia|e:ae|a|:|+ies|:|easx,c)+aay,c)+aa:ae|a|:|+i
veie.|:|esx ,c)+aay ,o) Oa:|ee:ae:a+aa,weieaaa:a+::ae+meaa:sx,;+aa
y, ; eis+i:|a:|e:we:+aiseis+miez+:eaes.:||ea|y:aesys:em
z0x= - :x + y,
zcy= :x- y
eica:s:e:ae:i|ae+:ea+:|eas uea.e:ae:we|a|:|+iv+iaesx,o)+aay,c)saeaia
sau.e:eae:e:m|ae:aeseia:|eaC|vea+a|,ae:e:ae:sys:em,weei:eamas:::+as
ie:m|:|a:e+aea|v+iea:a:s:e:ae:sys:em:ea|s.eve:aewm+ay|a|:|+i.eaa|:|eas
+:eaeeaea:eae:e:m|ae+aa|aeseia:|ea 1aee:em l :eiisas:a+::aeaam|e:ei
sa.a.eaa|:|eas|s :e.| seiy:ae s+me+s :aeaam|e:eiea+:|eas|a :aeea|v+iea:
a:s:e:ae:sys:em
Problems
............., ..
,........,........
..,....
1. .+ .+ .=

2. .
+ . . + .= cos 3t
3.

.. +

.= 0
4.

.+ . .= In
5. .

= . + cos .
6. . .+ .= 0, + .. = 0
..

.
7. .=
.

Y
=
.

8. .+ . + .. = 0, + .= cos
9. .= .- Y + 2z, = .+ Y
- . = .
10. .= ( 1 - . = ( 1 .
......,.6, 7, .........
...............
..........,....,......
..,...,....,.....
...........,......
.....
11. .= = .
12. .= = .
13. .= 2 = ..= 1 , = 0
14. .= = ..= 3, = .
15. .= = ..
16. .= . = .
17. .= = . .= 1 , = 2
18. .= = . . = 2, =
19. . = = .+ ..= 0, = 3
20. .= = .+
21. (a) Calculate .,

+ ,

to show that the trajecto


ries of the system . = = .of Problem 1 1 are
circles. (b) Calculate . to show that the
trajectories of the system .= = .of Problem 1 2
are hyperbolas.
22. (a) Beginning with the general solution of the system
.= = .of Problem 1 3, calculate .

to
show that the trajectories are circles. (b) Show similarly
that the trajectories of the system .= = ..
of Problem 1 5 are ellipses with equations of the form
.

23. First solve Eqs. (20) and (2 1 ) for and

in terms of
. and the constants .and . Then substitute the
results in

= 1 to show that the trajectories of


the system .= = .in Example sati sfy an
equation of the form
..

y = (constant).
Then show that = 0 yields the straight lines = .
and = .that are visible in Fi g. 5 . 1 . S.
5. 1 Fi rst-Order Systems and Appl i cati ons 335
24. Derive the equations
.= .

+ .

+ .

.= .

. .

+ .

for the displacements (from equilibrium) of the two


masses shown in Fig. 5. 1 . 1 1 .
FIGURE 5. 1. 11. The system of
Problem 24.
25. Two particles each of mass are attached to a string under
(constant) tension . as indicated in Fig. 5. 1 . 1 2. Assume
that the particles oscillate vertically (that is, parallel to the
y-axis) with amplitudes so small that the sines of the an
gles shown are accurately approximated by their tangents.
Show that the displacements YI and
Y
2
sati sfy the equa
tions
26.
.
= -2YI
+

= YI
- 2
Y

where .= L/ ..
:

FIGURE 5. 1. 12. The mechanical system of


Problem 25.
Three 1 00-gal fermentation vats are connected as indi
cated in Fig. 5 . 1 . 1 3, and the mixtures in each tank are
kept uniform by stirring. Denote by . the amount (in
pounds) of alcohol in tank { at time = 1 , 2, 3). Sup
pose that the mi xture circulates between the tanks at the
rate of 1 0 gal/min. Derive the equations
. = . +

.= . .

FIGURE 5. 1. 13. The fermentation tanks


of Problem 26.
336 Chapter 5 Li near Systems of Di fferenti al Equati ons
27. Set up a system of frst-order diferential equations for
the indicated currents

and ] in the electrical circuit of


Fig. . which shows an inductor, two resistors, and
a generator which supplies an alterating voltage drop of
= sin V in the direction of the current

FIGURE 5. 1. 14. The electrical circuit of Problem


28. Repeat Problem except with the generator replaced
with a battery supplying an emf of V and with the
inductor replaced with a I -millifarad (m) capacitor.
29. A particle of mass moves in the plane with coordinates
(x(t), y(t under the infuence of a force that is directed
toward the origin and has magnitude k/(x
2
y
2
)-an
inverse-square central force feld. Show that
"
kx
"
ky
mx = _ and = _,
where r = ,x
2
y
2
.
30. Suppose that a proj ectile of mass moves in a vertical
plane in the atmosphere near the surface of the earth un
der the infuence of two forces : a downward gravitational
force of magnitude .and a resi stive force F _ that is
directed opposite to the velocity vector v and has magni
tude kv
2
(where v = v is the speed of the projectile;
see Fig. Show that the equations of motion of the
projectile are
.= kv.' , = -kvy' .
where v = ,(x' )
2
(y
,
)
2
.
:
0
N

FIGURE 5. 1. 15. The trajectory of the


projectile of Problem
31. Suppose that a particle with mass and electrical charge
,moves in the xy-plane under the infuence of the mag
netic feld B = ..(thus a uniform feld parallel to the
z-axis), so the force on the particle is F = ,v 7 if its
velocity is v. Show that the equations of motion of the
particle are
. , . = , ..
. 1 | COl OH Gravitation and Ktpler' s Laws of Planetar Motion
A:eaaa:ae:anei:ael :a.ea:a:y,ea+aaeskeie:+a+iyzea+i|ie:|meeii+ae
:+:ye|se:v+:|eas|y:ae+s::eaeme:1y.aes:+aekeie:.ea.iaaea:a+::aeme:|ea
ei:aei+ae:s+:eaaa:|esaa|saes.:||ea|y:aeieiiew|a,:a:ee:ees|:|eas,ae
mewa+sKepler's laws of planetary motion:
1. 1aee:||:eie+.ai+ae:|s+aeii|sew|:a:aesaa+:eaeie.as
Z. 1ae:+a|asve.:e:i:em:aesaa: ee+.ai+ae:sweesea:+:e++: +.eas:+a:
:+:e
J. 1aesa+:eei:aei+ae: se:|eaei:eveia:|ea|s:ee::|ea+i:e:ae.a|eai
:aem+]e:sem|+s|sei|:seii|:|.+ie:||:
iaa|si/./p/.M./e-./.., l :s)is++.New:eaaeaa.ea:ae|ave:sesa+::
i+wei,:+v|:+:|eai:emkeie: si+ws ia:a|s+i|.+:|eaweie+ayea,|a:aeee
s|:ea|:e.:|ea):a:ea,a+ae:|v+:|eaeikeie: sa:s::wei+wsi:emNew:ea si+ei
,:+v|:+:|ea
Assame:a+::aesaa|sie.+:ea+::aee:|,|a|a:aei+aeeime:|eaei+i+ae:,
+aaw:|:e:aees|:|eave.:e:ei:aei+ae:|a:aeie:m
r, ) = ,x,; ,y, ; = xi +y] . , l )
w|e:ei = , l , c) +aa] = ,c, l ) aeae:e:aeaa|:ve.:e:s| a:aees|:|vex- +aa,
a|:e.:|eas 1aea:ae|ave:sesa+:ei+wei,:+v|:+:|ea|mi|es,r:e|iemz):a+::a:
:
FIGUR 5. 1. 16. The radial and
transverse unit vectors u, and uq .
0(0))
FIGUR 5. 1. 17. Area swept out
by the radius vector.
5. 1 Fi rst-Order Systems and Appl i cati ons 337
a..eie:a:|eave.:e:r,;ei:aeiaae:|s,|vea|y

/r
r = --,
'
,z)
wae:e = , x

+y

|s :aea| s:aa.ei:em:ae saa :e :aeiaae: ii:|eeia:.e


e:a|aa:esei:aeiaae:a::|me a:e, , ; . -,) ) , :aea:ae:aa|aiaaa::aasve:seaa|:
ve.:e:ssaewa|ar|, l l : a:e,|veaiy
a,=i .esJ +]s|aJ aaa a, =-i s|aJ +].es- ,)
1|e:aa|aiaa|:ve.:e:a, ,waeaie.a:eaa::aeiaae: ses|:|ea)aiwayse|a:sa|
:e.:iyawayi:em:aee:|,|a,sea, = r}, aaa:ae::aasve:seaa|:ve.:e:a, |se|
:a|aeai:ema,|yac.eaa:e:.ie.iw|se:e:a:|ea
STEP 1 : D|ne:ea:|a:e:aeeaa:|eas|a,).emeaea:w|se:esaew:aa:
,:)
STEP 2: Use:aeeaa:|eas|a,:):ea|ne:ea:|a:e:|eiaae: ses|:|eave.:e:r*
a,aaa:ae:e|ys|ew:aa:|:sveie.|:yve.:e:|s,|vea|y
ar a a-
v = - = a, + -a,
a a a
STEP 3: D|ne:ea:|a:e a,a|a :e s|ew :|a::aeiaae: sa..eie:a:|ea ve.:e:

av}a|s,|vea|y

a-

a +

a-

a,
a
:
a
,
a a
,)
,:)
STEP 4. 1ae:aa|aiaaa::aasve:se.emeaea:sea:ae:|,a:aaaas|aes|as ,z)
aaa,:) mas:a,:ee aa:|a,:ae::aasve:se.emeaea:s-:aa:|s, :ae.eeia.|ea:s
ei a,-we ,e:
se|:ieiiews:aa:

a-

=c.
a a

a-
=/
a

,)
,s)
wae:e/|sa.eas:aa:se.aase:aeeia:.ee:a|aa:ea:eaeiemea:-ie:.ema:a:|ea
ei:|ea:ea~() |ar|, l l -|s,|vea|yan =

a-, ,s)|mi|es:aa::ae
ae:|va:|ve~ ( ) |s.os..w||.||sas:a:emea:eikeie: sse.eaaiaw
STEP 5. aa:e :aa|ai .emeaea:s|a,z) aaa,:)aaa:aeaase:|e:esai:|a,s)
:esaew:aa::|eiaae: s:aa|ai.ee:a|aa:eiaa.:|ea , ) sa:|saes:aese.eaae:ae:
a|ne:ea:|aieaa:|ea
/
-

,)
338 Chapter 5 Li near Systems of Di fferenti al Equati ons
:
FIGURE 5. 1. 1S. The elliptical
orbit
L
r =
e cos(O a)
with perihelion distance
rl Lj( l e) and aphelion
distance r2 = Lj ( l - e) .

FIGURE 5. 1. 19. The shape of
the orbit of Halley' s comet.
STEP 6. Ai:aea,a:aea|ne:ea:|+iea+:|ea|a,)|saeai|ae+:, |:.+a|e::+as
ie:mea:e +i|ae+:ea+:|ea|yme+asei+s|miesa|s:|:a:|ea te::a|s a:ese,
+ssame:a+::aee:||:.+a|ew:|::ea|a:|eei+:.ee:a|a+:eie:m= ,-) , +aaa:s:
ase:ae.|+|a:aie+aa,s):esaew:a+:|i= i};, :aea
a a;
a
= -/
a-

D|iie:ea:|+:e+,+|a:eaeaa.ei:em ,;:a+::aeiaa.:|ea; ,-) = l} ,-) s+:|saes


:|ese.eaae:ae:ea+:|ea
STEP 7: saew:a+::ae,eae:+iseia:|eaei, l c)|s
/
; ,-) = ~sm-+s.es -+
/
:

STEP 8: t|a+iiy,aeaa.ei:em, l l ):|+: ,-) = i}; ,-) | s ,|vea|y


L
,-) =
i+e.es ,-- a)
, l o)
, l l )
, l z)
w|:ae = c/
:
}/. c.es a = ~, cs|aa = s, +aaL = /
:
}/ 1aeei+:.ee:a|a+:e
,:+a ei , l z) |s +.ea|.se.:|eaeie..ea::|.|:ye-+aeii|se|ic _ e < l , +
+:+|ei+|ie = l , +aa+aye:|ei+|ie > l -w|:|ie.as+::aee:|,|a ri+ae:+:y
e:||:s+:e|eaaaea+aa:|e:eie:e+:eeii|sesw|:|e..ea::|.|:ye < l As|aa|.+:ea
|ar|, l l s,:aem+]e:+s|sei:aeeii|sei|es+iea,:ae:+a|+ii|aeJ =a
STEP rie: seme:y|.+i eii|:|.+i e:||:s+s aes.:||ea|y, l z) w|:aa|ne:ea:
e..ea::|.|:|es,s|zes, +aae:|ea:+:|eas ia:e.:+a,ai+:.ee:a|a+:esyea.+aw:|:e
x , ) = , ) .es , y, ) = , ) s|a , c _ _ z-
:eie:+aeii|:|.+ie:||:w|:|e..ea::|.|:ye,sem|i+:as:e.:amL ,t|, l l s), +aa
:e:+:|ea+a,iea 1ae e..ea::|.|:yei:ae e+::a se:||:|s e c cl :, se.iese:e
ze:e:a+: :|e e:i|:ieeisae+:iy.|:.ai+:,:aea,aw|:|:aesaaen.ea:e:), +aa :a:
e..ea::|.|:|esei:|ee:ae:i+ae:+:ye:||:s:+a,ei:emc cc:sie:Veaas+aac o
ie:H+:s:ec zc:ie:He:.a:y+aac z:s:ie:ria:esa:m+ay.eme:sa+vea|,aiy
e..ea::|.e:||:s,i|ieu+iiey s.eme:w|:|e c ,t|, l l )
The Method of Elimination
1aemes:eiemea:+:y+:e+.a:ei|ae+:sys:emseia|ne:ea:|+iea+:|eas|aveives
:aeei|m|a+:|eaeiaeeaaea:v+:|+iies|y+:e:|+:eiy.em||a|a,+|:s eiea+
:|eas 1aee|] e.:ei:a|s:e.eaa:e|s:eei|m|a+:eaeeaaea:v+:|+|ies|asa..ess|ea
aa:|i:ae:e:em+|aseaiy+s|a,ieea+:|ea.ea:+|a|a,eaiyeaeaeeaaea:v+:|+|ie
1a|s:em+|a|a,ea+:|eaw|iiasa+iiy|e+i|ae+:ea+:|eaeia|,ae:ae:+aa.+ai:e
aea:iy|eseivea|y:aeme:aeaseiCa+:e:z Ai:e:|:sseia:|eaa+s|eeaieaaa,:a:
e:ae:aeeaaea:v+:|+|ies.+a|eieaaa|a:an,as|a,e|:ae::aee:|,|a+ia|iie:ea:|+i
ea+:|ease::aese:a+:a+ve+e+:ea|a:aeei|m|a+:|ea:e.ess
Exampl e 1
5. 2 The Method of El i mi nati on 339
1|emethod of elmination ie: i|aea:a|ne:ea:|ai sys:ems |s s|m|ia::e:ae
seia:|eaeiai|aea:sys:emeiai,e|:a|.eaa:|eas|ya:e.esseiei|m|aa:|a,:ae
aaiaewaseaea:a:|meaa:|ieaiyas|a,ieeaa:|eaw|:aas|a,ieaaiaewa:ema|as.
i:|smes:.eavea|ea:|a:ae.aseeimaaa,ea|iysmaiisys:ems :aese.ea:a|a|a,
ae me:e:|aa :we e::|:eeeaa:|eas. te: sa.a sys:ems :aeme:aeaeiei|m|aa
:|ea:ev|aesas|mieaaa.ea.:e:ea:ea.a:aa::ea|:esi|::ie:ei|m|aa:y:aee:y
e:ie:mai ma.a|ae:y. sa:ie: ia:,e: sys:emseia|ne:ea:|aieaa:|eas,as weii as
ie::|ee:e:|.aia|s.ass|ea,:|ema::|sme:aeaseiia:e:se.:|eas|a:a|s.aa:e:+::
:eie:a|ie.
t|aa:aea::|.aia:seia:|eaei:aesys:em
x= :x- y, y = :x- y
:aa:sa:|saes:|e|a|:|ai.eaa|:|easx,c)= z. y,c)= -i
, l )
Sol uti on iiweseive:aese.eaaeaa:|ea|a, i ; ie:x, we,e:
se:|a:
,
,
7
x=
,
y +
,
y

,
,

7
x =
,
y +
,
y

,z
,)
we:aeasa|s:|:a:e:aesees:ess|easie:xaaax|a:|ea:s:eaa:|eaei:aesys::m
|a, i ; , :||sy|eias
wa|.|wes|mi|iy:e
,

7
1

. 7

,
y +
,
y =
,
y +
-
y - y,
y

+y
- l cy= c
1a|sse.eaae:ae:i|aea:eaa:|eaaas.aa:a.:e:|s:|.eaa:|ea

:
+- ic= ,- z; ,+) = c.
se|:s,eae:aiseia:|ea|s
Nes:,sa|s:|:a:|eaei,1,|a,z;,|ves
:|a:| s,
1|ass. ,1,aaa,).eas:|:a:e:ae,eae:aiseia:|eaei:|esys:em|a, i ;
1|e,|vea|a|:|ai.eaa|:|eas|miy:aa:
x ,c)= .,+.
:
= z
aaa:aa:
y,c)= .,+ .
:
= -i ,
(4)
,)
340 Chapter 5 Li near Systems of Di fferenti al Equati ons
FIGUR 5.2. 1. Direction feld
and solution curves for the system
.= ..- 3y, y' = .- 7y of
Example
:aeseeaa:|eas a:e:eaa|iyseiveaie:., = z aaa.
:
= - uea.e:aeaes|::a
seia:|ea|s
x ,;= e
:

- e
-
, y,; =ze
:

- e
-

t|,a:e z l saews:a|saaae:|e::y|.aiseia:|ea.a:vesa:ame::|zea|y:aeea+
:|easx , ; = .,
e
:

+ .
:
e-
,y,; = .,
e
:

+.
:
e-
w|:|a|ne:ea:vaiaesei:a:
a:||::a:y.eas:aa:s., aaa.
:
wesee:weiam|i|esei.a:ves:esem|i|a,aye:|ei+s
saa:|a,:aesamea|:ei,e|i|ae)asym:e:es.
Remark: 1|e,eae:aiseia:|eaaeaaea|ys. ,:)aaa,)may|e:e,+:a:a
as:|ea|:e:ve.:e:,x,; ,y, ; ke.aii|a,:|e.emeaea:w|seaaa|:|eaeive.:e:s
,aaamai:|i|.a:|eaeive.:e:s|ys.aia:s) , we.aaw:|:e:ae,eae:ai seia:|ea|a(4)
aaa,)|a:aeie:m
1||ses:ess|ea:esea:s:ae,eae:aiseia:|eaei:aesys:em|a, l ) asai|aea:.em||
aa:|eaei:ae:wea::|.aia:seia:|eas

Polynomial Diferential Operators


iasamie l weasea aa aa|e.:e.eaa:e:e ei|m|aa:e eaeei:|e|aaeeaaea:
va:|a|ies |y es:ess|a, |: |a :e:ms ei:aee:ae:. weaew aes.:||e a sys:em+:|.
ei|m|aa:|ea:e.eaa:e. Oe:a:e: ae:a:|ea|s mes:.eavea|ea:ie::aesea:es:s
ke.aiii:emse.:|eaz:aa:apolynomial diferential operator |seaeei:aeie:m
,:)
wae:enaeae:esa|iie:ea:|a:|eaw|:a:ese.::e:ae|aaeeaaea:va:|a|ie
i ii| aaai
:
a:e:wesa.aee:a:e:s,:|ea:|e|::eaa.:i|i
:
|saeaaea:a|s
way
te:|as:aa.e,|ii, =n+.aaai
:
= n+/.:aea
i
.
i
:
x}= ,n+.; , n+/;x}= n, nx+/x;+., nx+/x;
= n
:
+,.+/; n+./}x
,)
1a|s|iias::a:es:aeia.::aa::weeiyaem|aiee:a:e:sw|:a.eas:+a:.eeu.|ea:s.+a
|emai:|i|eaas|i :aeywe:ee:a|aa:yeiyaem|ais|a:aeva:|a|ien se.aase:ae
mai:|i|.a:|eaeisa.|eiyaem|ais|s.emma:a:|ve,|:ieiiews:aa:
,s)
|i:aeae.essa:yae:|va:|veseix,;es|s:.sy.ea::as:,:||s:ee::yei.emma:a:|v|:y
,eae:aiiyia|isie:eiyaem|aiee:a:e:sw|:ava:|a|ie.eeia.|ea:s-seer:e|iemszl
aaazz
5. 2 The Method of El i mi nati on 341
Aaysys:emei:wei|aea:a|ne:ea:|+ieaa:|easw|:a.eas:+a:.eeia.|ea:s.+a
|ew:|::ea|a:aeie:m
i, x+i
:
y=
)
,,; .
i

x+i
-
y= , ; .
,)
wae:ei
, ,i
:
,i

.aaai
-
a:eeiyaem|aia|ne:ea:|+iee:a:e:s,e:a+seia|iie:ea:
e:ae:s) as |a ,:), aaa
)
,,; +aa ,;a:e ,|veaiaa.:|eas te:|as:+a.e, :ae
sys:em|a, l ) ,samiel ) .aa|ew:|::ea|a:aeie:m
, b- :)x+ y= c,
-:x + ,b+:;y= o,
w|:ai, = b :, i
:
= , i

= -:, aaai
-
= b+
1eei|m|a+:e:aeaeeaaea:v+:|+|iexi:em:aesys:em| a,), weee:+:e|ta
i

ea:aea:s:eaa:|eaaaaw|:ai,ea:aese.eaa1aaswee|:a|a:aesys:em
i

i, x+i

i
:
y= i,,,; .
i, i

x+i
,
i
-
y= i, ,;
, l l )
sa|::+.:|eaei:aea:s:i:em:aese.eaaei:aeseea+:|easy|eias:aes|a,ieea+:|ea
, l z)
|a:aes|a,ieaeeaaea:va:|+|iey Ai:e:seiv|a,ie:y= y,;e.+asa|s:|:a:e:ae
:esai:|a:ee|:ae:ei:aee:|,|aaieaa:|eas|a,)aaa:aeaseiveie:xx,;
Ai:e:aa:|veiy,we.eaiaei|m|a+:e|ai||em+aae::aeaeeaaea:va:|+||eyi:em
:aee:|,|aaisys:em|a ,) iise, wewea|a,e::aeea+:|ea
, l )
wa|.a.aaaew|eseiveaie:x= x, ;
Ne:e:aa::aesameee:a:e:i, i
-
i
:
i

aea:sea:aeiei:aaaas|ae| a|e:a
, l z)aaa, l ) 1a| s|s:aeoperational determinant
.
ei:aesys:em|a,) iaae:e:m|aaa:ae:a:|eas , l z)+aa, l ).+a|e:e:|::ea+s
)
,,; i
:
,; i
-

. s
i:|s|me::+a::eae:e:aa::aeae:e:m|aaa:sea:ae:|,a:aaaas|ae|a . s +:eev+i
aa:ea|ymeaasei:aeee:a:e:see:a:|a,ea:aeiaa.:|eas 1aeea+:|eas|a . s
a:es::ea,iy:em|a| s.ea:eiC:ame:s:aieie::aeseia:|eaei:wei|ae+:ea+:|eas|a
:we,ai,e|:+|.)va:|a|ies+aaa:e:ae:e|yeasy:e:emem|e: iaaeea,yea.+aseive
asys:emei:wei|aea:a|iie:ea:|aiea+:|ease|:ae:|y.+::y|a,ea::aesys:em+:|.
ei|m|aa:|ea:e.eaa:eaes.:||eaae:ee:|ya|:e.:iyemiey|a,:aeae:e:m|a+a:ae
:a:|ea|a . s |:ae::e.ess|s ese.|aiiy s|mie|i:ae sys:em |saeme,eaeeas
, ),,;= caaa,;= c), |e.aase|a:a|s.ase:ae:|,a:aaaas|aesei :aeea+:|eas
|a, l z) , , l ), aaa, l )a:eze:e
342 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| eZ r|aaa,eae:aiseia:|eaei:aesys:em
,n :;x + y= c,
-:x+,n+:;y= c
, i ;
Sol ution 1|eee:a:|eaaiae:em|aaa:ei:||ssys:em|s
,n- :; ,n+:;

,-:;= n
:
+n- i c
uea.es. , i ;aaa, i z;a:e
x

+x- i cx= c,
y

+y- i cy=c
1|e.aa:a.:e:|s:|.eaa:|eaeiea.a|s

:
+- l c= ,- z) ,+;=c,
se:ae|:,sea:a:e),eae:aiseia:|easa:e
x ,; = ., e
:

+.
:
e
-
.
y,; = /, e
:

+/
:
e
-

, i :;
, i :;
A::||se|a:weaea::e|ave)oea:||::a:y.eas:aa:s..
:
./, aaa/

sa:
|:ieiiewsi:em1|ee:eml|ase.:|ea i :aa::ae,eae:aiseia:|eaeiasys:emei:we
a:s:e:ae:eaa:|eas|aveiveseaiy:wea:||::a:y.eas:aa:s. 1a|saa:ea:a|u.a|:y
aemaaasa:eseia:|ea.
1aeesiaaa:|ea|ss|mie 1ae:emas:|esemea|aaea:eia:|easamea,ea:
iea:.eas:aa:s.we.aaa|s.eve::aem|ysa|s:|:a:|a,:aeseia:|eas|a, i :;|a:ee|:ae:
ei:|ee:|,|aaieaa:|eas|a, l c) . Oasa|s:|:a:|ea|a:aea:s:eaa:|ea,we,e:
c = x- :x+ y
= ,z., e
:

- .
:
e
-
;- :,., e
:

+.
:
e
-
;+ ,/, e
:

+/
:
e
-
; ,
:aa:| s,
c = , -z., +/, ; e
:

+, -.
:
+/
:
;e
-

sa:e
:
aaae
-
a:eiaea:iy|aaeeaaea:iaa.:|eas, se|:ieiiews:aa:., = /, aaa
.
:
= /
:
1|e:eie:e,:aeaes|:ea,eae:aiseia:|ea|s,|vea|y
Ne:e:|a::||s:esai:|s|aa..e:aw|:|:|e,eae:aiseia:|ea,s. ,:;aaa,,:aa:we
e|:a|aea|yaa|ne:ea:me:|ea|asamiel .
As|iias::a:ea|ysamiez.:aeei|m|aa:|ea:e.eaa:easea:eseiveai|aea:
sys:emi:eaea:iy w|ii|a::eaa.eaaam|e:ei|a:e:aeeaaea:.eas:aa:s :aa:may
aea::e|ea:||::a:y,|a:a.:aaiiya:eae:|aaeeaaea:.1aees::a.eas:aa:smas:
:|ea|eei|m|aa:ea|ysa|s:|:a:|eaei:ae:eesea,eae:aiseia:|ea|a:eeaee:me:e
ei:aee:|,|aaia|ne:ea:|aieaa:|eas. 1aea:e:|a:eaam|e:eia:||::a:y.eas:aa:s
|aa,eae:aiseia:|eaeiai|aea:sys:em|sae:em|aea|y:aeieiiew|a,:ees|:|ea
5. 2 The Method of El i mi nati on 343
ii:aeee:a:|eaaiae:e:m|aaa:|a, l ) |so/ae/..//y,eo.:aea:ae
aam|e:ei|aaeeaaea:a:||::a:y.eas:aa:s|a a,eae:aiseia:|eaei:ae
sys:em|a ,)|s eaai:e:|ee:ae:ei|:s ee:a:|eaaiae:e:m|aaa:-:aa:
| s, |:sae,:eeasaeiyaem|ai|an
,te:a:eeiei:|| sia.:, seea,es l ::-l cei L ia.e soa/.u ee/./
i,./os,NewYe:i Deve:, l :). ) 1|as:|e,eae:aiseia:|eaei:aesys:em|a
, l c)eisamiez|aveives:wea:||::a:y.eas:aa:s, |e.aase|:see:a:|eaaiae:e:m|
aaa:n
:
+ n- l c|seie:ae:z
ii:|eee:a:|eaaiae:e:m|aaa:|s|aea:|.aiiyze:e,:aea:aesys:em|ssa|a:e|e
degenerate. Aae,eae:a:esys:emmay aavee|:ae:aeseia:|eae:|aaa|:eiymaay
|aaeeaaea:seia:|eas te:|as:aa.e,:aeeaa:|eas
nx- ny=c,
znx- zny= l
w|:|ee:a:|eaaiae:e:m|aaa:ze:ea:ee|v|easiy|a.eas|s:ea:aaa:aasaaveaeseia
:|eas Oa:|ee:ae:aaaa,:aeeaa:|eas
nx+ ny= ,
znx+zny=z
w|:aee:a:|eaaiae:e:m|aaa:ze:ea:ee|v|easiy:eaaaaaa:, we.aasa|s:|:a:e.v
,.ea:|aaeasiy a|ne:ea:|a|ie) iaa.:|ea ie: x , ) aaa :aea |a:e,:a:e :e e|:a|a v,;
kea,|iyseai|a,,eve:yae,eae:a:esys:em|s ea|vaiea::ee|:ae:aa|a.eas|s:ea:
sys:eme:a:eaaaaaa:sys:em
Ai:|ea,|:aeaie:emea:|eaea:e.eaa:esaaa:esai:sa:eaes.:||eaie::ae.ase
eiasys:emei:weeaa:|eas, :aey.aa|e,eae:ai|zea:eaa|iy:esys:emsei:a:eee:
me:eeaa:|eas te::aesys:em
i,

x+i
:
y+ i

:
= ),,; ,
i
:
x+i
::
y+i
: :
=
J
,; ,
i

x +i
:
y+i
:
= ,;
, l s)
ei:|:eei|aea:eaa:|eas, :aeaeeaaea:va:|a|iex , ) sa:|saes:aes|a,iei|aea:eaa
:|ea
i

i
:
i

i
:
i
::
i
:
x =
i

i
:
i

),,; i
:
i

J
,; i
::
i
:
,; i
:
i

, l )
w|:|aaaie,easeaa:|easie:y= y,;aaa,=, , ; te:mes:sys:emseime:e:aaa
:|:eeeaa:|eas, |eweve:,:ae me:|eaeiee:a:|eaaiae:e:m|aaa:s|s:ee:ea|eas:e
|e:a.:|.ai
Mechanical Vibrations
Ame.|aa|.aisys:em:y|.aiiyv||:a:ese:es.|iia:ese:|ea|.aiiy|aeaee:me:ese
.|a.ways 1|eme:aeasei:a|sse.:|eaei:ea.aa|eai|ea:eaaaiyze:aeaa:a:ai
meaeseies.|iia:|eaeia,|veasys:emsamie|iias::a:es:a|sa:ea.a
344 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| e iasamieleise.:|ea l , weae:|vea:aeeaa:|eas
, b

+)x+ , -l ) y=c,
-zx + , b

+ z)y=c
,z;
ie::aea|sia.emea:sei:ae:wemasses|ar|, z z ue:e], ) = c|e.aas::
assame:aa::|e:e|saees:enaiie:.e r|aa:ae,eae:ai seia:|eaei:aesys:em|a
,zc)
Sol uti on 1aeee:a:|eaaiae:e:m|aaa:ei:aesys:em|a,zc)|s

() ()
La|||ot|ames|t|eas
FIGURE 5.2.2. The mass
and-spring system of Example
, b

+) ,b

+z) - , -l ) , -z)=b
.
+b

+:= , b

+l ) ,b

+4)
uea.e:aeeaa:|easie:x , ) aaay, ) a:e
, b

+l ) ,b

+:)x=c,
, b

+l ) ,b

+:)y=c

,zl )
1|e.|a:a.:e:|s:|.eaa:|ea,

+ l ) ,

+:) = c|as:ee:s/ , -/ , z/ ,aaa-z/

se
:ae,eae:aiseia:|easei:aeeaa:|eas|a,zl ) a:e
x , ) =a, .es +a

s|a + .es z+/

s|a z ,
y, ) =., .es+c s|a+d .esz+a

s|az
,zz;
se.aase:aeee:a:|eaaiae:e:m|aaa:|seie:ae::,:ae,eae:aiseia:|easaeaia
.ea:a|aiea: ,:a:|e: :|aae|,a:) a:||::a:y .eas:aa:s waea wesa|s:|:a:ex,)+aa
y,)i:em,zz)|a:aea:s:eaa:|ea|a,zc),we,e:
:aas
c = x+x- y
= ,-a, .es- a

s|a- :/,.esz- :/

s|az)
+ ,a, .es+a

s|a+/, .esz+/
:
s|az)
- ,., s|a+cs|a+d.esz+d
z
s|az) ,
= ,za, - cj ) .es+,za

- .

)s|a
+,-/, - dj ).es z+, -/

- d
z
)s|a z
se.aase.es , .esz, s|a , aaas|aza:ei|aea:iy|aaeeaaea:,|:ieiiews:aa::a:|:
.eeu.|ea:s|a:aeias:eaa:|eaa:eze:e1|as
1ae:eie:e
x , ) = .,.es + a

s|a +/,.es z+/

s|a z,
y, ) =za

.es+za

s| a - .esz- /
:
s| az
| s:aeaes|:ea,eae:aiseia:|eaei:aesys:em|a,zc)
,z);

1
-1
0

==
-

===
la +a
FIGUR 5.2.3. The two masses
move in the same direction, each
with frequency u =
1

"z
= ees

-I
-l
-1

0 a la 1a +a
FIGUR 5.2.4. The two masses
move in opposite directions with
frequency =
5. 2 The Method of El i mi nati on 345
1|eeaa:|eas|a,z)aes.:||efree oscillations ei:aemassaaas:|a,sys
:emeit|,. . z. z-me:|ea sa|]e.::eno es:e:aai ie:.es. tea:|a|:|ai .eaa|:|eas
,:y|.aiiy,|a|:|aia|sia.emea:saaaveie.|:|es)weaia|e:ea|:ea:eae:e:m|ae:ae
vaiaesei., ..
:
,/, ,aaa/
:
1aees:ess|ea
,x,; ,y, ; =.,,.es,z.es ) +.
:
,s|a,zs|a)
+ /, ,.es z ,- .es z)+ /

,s|a z, - s|a z)
,z:)
:|ea :esea:s :|e ,eae:ai seia:|eaei:ae sys:em |a ,zc) as ai|aea:.em||aa:|ea
eiiea:a::|.aia:seia:|eas. He:eeve:, :ae a:s::weei:aesea::|.aia:seia:|eas
:e:esea:|ys|.aiiys|m|ia:es.|iia:|easei:aemasses, asae:aeia::e::we.
aaa
iaaeea,we.aa,|y:aeasaai::|,eaeme::|.ma.a|aa:|eas)w:|::
.,.es+ .
:
s|a= n.es ,- a) ,
z.,.es+z.
:
s|a=zn.es ,- a)
/,.esz+/
:
s| az= s.es ,z- f) ,
-/,.esz- /
:
s| az=-s.es ,z- f)
w|:| n = ,.j+ ., :aa a = .,}., , s = ,/+ /, aaa:aa f = /
:
}/, 1aea
.,z:):aies:|eie:m
,z)
w|e:e:|ea::|.aia:seia:|eas
,x,,) ,y,,; = ,.es ,- a) ,z.es ,- a) ,z)
aaa
,x
:
,; , y
:
,; = ,.es ,z- f) , - .es ,z- f; ,z)
aes.:||e:|e:wenatural modes of oscillation ei:aemassaaas:|a,sys:em.Me::
eve:,:aeyesa|||:|:s:we,.|:.aia:)natural frequencies o
.
= iaaau =z.
1aei|aea:.em||aa:|ea|a. ,z):e:esea:saaa:||::a:yi:eees.|iia:|eaet
:|emassaaas:|a,sys:emasasae:es|:|eaei|:s:weaa:a:aimeaeseies.|ii+
:|ea,w|:a:ae.eas:aa:sn,a, s,aaafae:e:m|aea|y:ae|a|:|ai.eaa|:|eas. t|,a:e
. z. ,wae:ea = c)|iias::a:es:|eaa:a:aimeae ,x, ,y, ;ei. ,z:),|awa|.a:a:
:wemassesmeve|asya.a:eay|a:aesamea|:e.:|eaw|:a:aesamei:eaea.yeies
.|iia:|eao. = i . |a:w|:a:aeami|:aaeei-
:
:w|.e:aa:ei-,,|e.aase
y.
=zx, )

t|,a:e . z. :,wae:e f = c) |iias::a:es :ae aa:a:ai meae ,x


:
,y
:
; ei. ,z), |a
w||.|:ae masses meve |a sya.a:eay |aees|:ea|:e.:|eas, w|:a :ae samei:e
aea.yo
:
=zaaaw|:aeaaiami|:aaeseies.|iia:|ea,|e.aasey
:
=x
:
;
............... ..
.... ............,..
.................6, ...
,.....,...........
.....,...........
1. x
'
= -x =
2. x
'
= x = .
346 Chapter 5 Li near Systems of Di fferenti al Equati ons
3. .= . = ...= 0, =
4. .= . = . .= 1 , = -1
5. .= . . = .
6. .= . = . .(0) = =
7. .= .. = .
8. .= .= .

9. .= . .. = . cos
... = .. . = ..= 1 , = -
11. .= . . . = .
12. .= . = .
13. .= . = .
14. .= ..sin t , = ..
15. . .= 0, . = 0
16. . ..= sin , . = 0
17. . . .= 0,
. . ..= 0
18. .= .,, = . ,

= . ,
19. .= .. = ... ,, ,

= ..
20. . = , = .,, ,

= ......
Solve the characteristic equation by inspection. )
21. Suppose that .

= ..

and .

= .

where the coeffcients are all constants, and


that .is a twice diferentiable function. Verify that
.

.= .

. .
22. Suppose that . .= ...and that .

.= ...
Show that ..

...

. Thus linear operators with


..coeffcients generally do not commute.
..................
. ...,...,...
..................
....
23. ...=

... =

24. ...=
.. . =

25. .

... . = 0
...= 0
..26 ....29, .....,...
..................
............,,...........
.,h .....,...........
....
26. .

..

=
.

..

= 0
27. .

. .

=
.

..

= 0
28. .

...

=
.

. .

. = 0
29. .

. .

=
.

..

0
30. Suppose that the salt concentration in each of the two brine
tanks of Example of Section 5 . 1 initially = 0) is 0. 5
Ib/gal. Then solve the system in Eq. there to fnd the
amounts .and of salt in the two tanks at time
31. Suppose that the electrical network of Example of Sec
tion is initially open-no currents are fowing. As
sume that it is closed at time = 0; solve the system in
Eq. there to fnd

and 1
2
(t) .
32. Repeat Problem except use the electrical network of
Problem of Section 5. 1 .
33. Repeat Problem 3 1 , except use the electrical network of
Problem of Section 5. 1 . Assume that 11 (0) = and
,= 0, so that at time = 0 there is no charge on the
capacitor.
34. Three 1 00-gal brine tanks are connected as indicated in
Fig. of Section 5. 1 . Assume that the frst tank ini
tially contains 1 00 lb of salt, whereas the other two are
flled with fresh water. Find the amounts of salt in each of
the three tanks at time .....Examine the equa
tions to be derived in Problem of Section 5. 1 . )
35. From Problem 3 1 of Section recall the equations of
motion
.= ,. = ,..
for a particle of mass and electrical charge ,under the
infuence of the uniform magnetic feld B = .. Sup
pose that the initial conditions are .= = U,
. = 0, and = .where .= ,.Show
that the trajectory of the particle is a circle of radius
36. If, in addition to the magnetic feld B = .. the charged
particle of Problem moves with velocity vunder the in
fuence of a uniform electric feld .= Ei, then the force
acting on it is F = , .v 7 B) . Assume that the particle
starts from rest at the origin. Show that its trajectory is the
cycloid
.= . cos . = .. sin .
where .= E]..and .= ,.The graph of such a
cycloid is shown in Fig.
FIGUR 5.2.5. The cycloidal path of
the particle of Problem
37. In the mass-and-spring system of Example suppose in
stead that =

= 0. 5, .= and .

= (a)
Find the general solution of the equations of motion of the
system. In particular, show that its natural frequencies are
WI = and W2 =

(b) Describe the natural modes


of oscillation of the system.
38. Consider the system of two masses and three springs
shown in Fig. Derive the equations of motion
.

..

.+ .

. .

+ .


La| | | o;| ames|t|eas
FIGURE 5.2.6. The mechanical system of
Problem
..39 ....46, ..........
...38 .............,...
.... .......,...........,..
................. ..
,.....,...........
........,.......5. 2. 3 ...5. 2. 4).
39. .

. .

40.

. .

41.

. .

. .

42.

. .

43.

. .

44.

..

45.

..

..

.
46.

...

.
47. (a) For the system shown in Fig. derive the equa
tions of motion
...+ .
.. . .
. .
.Assume that . . Show that the natural fre
quencies of oscillation of the system are
WI
_
W2 , .. and W

,.
}atrices and Lineayems
. 3 Matri ces and Li near Systems 347


FIGURE 5.2.7. The mechanical system of
Problem .
48. Suppose that the trajectory . of a particle mov
ing in the plane sati sfes the initial value problem
. + .
.
.. .
Solve this problem. You should obtain
.cos cos
.. sin
Verify that these equations describe the .,.traced
by a point . fxed on the circumference of a circle
of radius that rolls around inside a circle of radius
..If begins at A.when then the pa
rameter represents the angle A OC shown in Fig.
Y
A |, 0)

FIGURE 5.2.8. The hypocycloid of


Problem .

i:aea,a :a: s|mi: :i|m|aa:|ea ::.aa|a:s ei s:.:|ea z sau.: te: :a:


seia:|eaeismaiii|a:a:sys::ms.ea:a|a|a,eaiy :wee::a::::a+:|easw|:a
.eas:aa:.eeu.|ea:s,:ae,:ae:ai:e:::|:seii|a:a:sys::ms-asw:iiasseia:|ea
me:aeassa|:a|ieie:ia:,::sys::ms-a::mes::as|iyaaa.ea.|s:iya:s.:||:aas|a,
:aeiaa,aa,eaaaae:a:|eaeive.:e:saaama::|.:s

te::eaay::i:::a.:aaa::v|:,
:a|sse.:|ea|:,|asw|:aa.emi:::aaas:ii.ea:a|a:aa..eaa:ei:a:m+::|sae:+:|ea
aaa:::m|aeie,y:aa:|sae:a:a

s:.|ai::.aa|a:seii|a:a:ai,:|:a-s:.|a.aiiy,
:aeseasse.|a::aw|:a :|,:avaiaes aaa:|,:av:.:e:s-a:: |a::eaa.:aas a::a:a|a
sa|sea:a:s:.:|easei:a|s.aa:e:
348 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| e 1
Revew of Matrix Notation and Terminolog
Aam n matrix n| sa:e.:aa,aia:a::ayeimnaam|e:s,e:elements) a::aa,ea|a
m ,ae:|zea:ai)rows aaan ,ve::|.ai)columns:
c, , c
:
c,

c,
,
c,
-
c
:
c
::
c
:
c
:,
c
:-
c

c
:
c

c
,
c
-
n = , i ;
c , c
:
c

.,
,
c -
c
-
, c
-:
c
-
c
-,
c
--
wew|iie:a|aa:|iyaeae:ema::|.es|yboldface .a|:aiie::e:s seme:|mesweas:
:|ea||:ev|a:|ean = .,
,
]ie::|ema::|sw|:a:|eeiemea:c
,
|a:a:/ :a:ewaaa
]:|.eiama,as|a, i ; weaeae:e:|ezero matrix, ea.aea::yeiwa|.a|sze:e,
|y
(2)
A.:aaiiyie:ea.aa|:eies|:|ve|a:e,e:sm aaan :ae:: |s aam n z::ema::|s,
|a::|es|a,iesym|ei0 w|iisaia.eie:aii:aeseze:ema::|.es
1we m n ma::|.es n = .,
,
] aaau = /,
,
] a:: sa|a:e|e equal |i
.e::eseaa|a,eiemea:sa:eeaai ,:aa:| s, |ic
,
=/,
,
ie:i / m aaai ] n.
weadd naaau|yaaa|a,.e::eseaa|a,ea::|es
n+u= c
,
]+/,
,
= c
,
+/,
,
] ,;
1aas:|eeiemea:|a:ew/aaa.eiama]eiC=n+u|s.
,
=c
,
+/
,
1emai:||y
:aema::|sn|y:aeaam|e:.,wes|miymai:|iyea.aei|:sei:mea:s|y.
ii
:aea
aaa

z -
'
n =
: 7

.n=n.=.c
,
}
u
.

-i i c
'

7
_g

z
'

-i i
'

-i i
n+u=
: 7
+
7
_
g
=
i i

'

i s c

:C=:

g
-7
=
c -:z

(4)
'

5. 3 Matri ces and Li near Systems 349


weaeae:e,-i ;n|y-naaaaeaaesubtraction eima::|.esasiaiiaws
n- u=n+,-u; ,)
1|ema::|s ee:a:|eas]as:aeaaea |ave :|e ieiiaw|a, :ee::|es, ea.a ai
w||.||saaaie,eas:eaiam|i|a:ai,e|:a|.:ee::yei:|e:eaiaam|e:sys:em
n + = + n = n. n - n = .
n + u = u + n
n+,u+C) =,n+u;+C
.,n+u;=.n+.u.
,.+a;n=.n+an
,.amma:a:|v|:y),
,assa.|a:|v|:y),
,a|s::||a:|v|:y)
,:)
,)
,s)
,)
a.|ei:|ese:ee::|es|s:eaa|iyve:|aea|yeiemea:w|seai|.a:|eaaia.a::e
seaa|a,:ee::yei:|e:eaiaam|e:s re:esamie,.
,
+/
,
=/
,
+.
,
ie:aii/
aaa]|e.aaseaaa|:|eaei:eaiaam|e:s|s.emma:a:|veCeaseaea:iy,
n+u= .
,
+/
,
]= /
,
+.
,
]=u+n
1|etranspose n'ei:|e- ma::|sn=.
,
}|s:|e -,aa:e ) ma::|s
w|ese]:|.eiama|s:|e]:|:ewein,aaa.easeaea:iy,w|ase/ :|:aw|s:ae/ :a
.eiamaein; 1|as n' = .
,
] , ai:|ea,|:||s |s ae: ae:a:|aaaiiye:ie.:, yaa
mas::emem|e::|a:n' w|iiaa: |ave:ae same s|ae asnaaiessn|s asquare
ma::|s-:|a:|s, aaiess-=
Aa- I ma::|s-eae |av|a,eaiy a s|a,ie.eiama-| s .aiieaacolumn
vector, e:s|miyavector. weeueaaeae:e.eiamave.:e:s|yboldface iawe:.ase
ie::e:s, as|a
s|m|ia:iy, arow vector |s a I ma::|s-eae |av|a, eaiy a s|a,ie:aw, sa.a
asc = I 7 0 - ] re:aes:|e:|.aaa:ye,:a||.ai:easaas, wew|ii
i:eaea:iyw:|:ea.eiamave.:e:as:|e::aaseseeia:ewve.:e:,ie:esamie,:ae
:we:e.ea|a,.eiamave.:e:smay|ew:|::ea|a:|eie:ms
b = -7
seme:|mes|:|s.eavea|ea::aaes.:||eaa- ma::|s|a:e:msaie|:|e:|:s
-:ewve.:e:se:|:s.eiamave.:e:s1|as|iwew:|:e
. . .
b ] n ,
|:|saaae:s:eea:|a:a I , a2, . ,aaaam a:e:|emwve.:e:sei:|ema::|snaaa:|a:
bl , b2 , ,aaabn a:e:|e.o/-ve.:e:sei:|ema::|su
350 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| eZ
Matrix Multiplication
1|e:ee::|esi|s:ea|as ,:):|:ea,|,)a:ea|:eaa:a:aiaaaese.:ea1aea::
sa::|ses|a:|e:eaimeima::|sa:|:ame:|..emew|:|mai:|i|.a:|aa weaeaa:a::
:|escalar product aia:ewve.:a:a aaaa.eiamave.:e:b, ea.a|av|a,:aeam:
aam|e: eieiemea:sii

/
]
T
,
.
:|eaa b |saeaaeaasiaiiews

a b = c,
/
,
=c,
/
, +.
,
/
,+'

+ .
,
/

.
,
, i o,
esa.:iyas|a:|es.aia:e:ao:eaa.:ai:wave.:a:s-aiam|i|a::a|.i:amei:m:a
:a:y.ai.aias
1|e:eaa.:Auei:wama::|.es|saeaaeaeaiy|i:|eaam|e:ei.aiamasa|
A| seaai:e:aeaam|e:ei:ewseiu iiA|saam ma::|saaau|sa 7
ma::|s, :|ea :|e|: product Au |s :aem ma::|s C = .,] , wae:: .,, . :a:
s.aia::eaa.:ai :|e/ :|:awve.:e:a,eiAaaa:|e]:|.eiamave.:e:bj ai u 1aa
, i i ,
i a:e:msei:|e|aa|v|aaaiea::|eseiA = c,] aaau =
/
,] , , l l , .aao:
:e.as:|a:|eie:m
]
.
,
= c ,
/
,,
,
, i z,
ra:a:esesei|aaa.ema:a:|aa, :|eaeaa|:|ea|as , l l ) aaa, l z) |seasy:a
:emem|e:|yv|saai|z|a,:|e|.:a:e
c, , c, , c,

/
, ,
/
, ,
/
-
c,

c,, c,

/
,
/
,,

/
,-
a

c
-
, c
-, c
-

,
/

-
1
bj
w||.|s|ews:aa: aaeie:ms:|eae::aaa.:ei:|e:ew ve.:e:a, w|:a :ae.aiama
ve.:e:b
,
:aa|:a|a:|eeiemea:.
,
|a:|e/ :|:ewaaa:|e]:|.aiamaaiAui:m+y
|ei:a:||a|aiea:|a,:|e:awseiAaawa:ae.aiamasaiu1a|saisa:em.aa
as:|a::aeaam|e:ei.eiamasaiAmas:|eeaai:a:|eaam|e:ai:awsaiu
C|e.|yea:aaae:s:aaa|a,ai:ae aeaa|:|aa eima::|smai:|i|.a:|ea|yve:|iy|a
:|a:|i
z
A=
-l
l
aaa u= _
:|ea
Au=

s|m|ia:iy,ve:|iy:|a:

-
aaa:aa:
5. 3 Matri ces and Li near Systems

l: l s
: 0

x +

y
= :x+ y
:x y

l

'
l
l l

351

i:.aa|es|ewa|ya|:e.:,:|ea,|iea,:|y).ema:a:|aa|aseaaa|:saeaa|:|aa
:|a:ma::|smai:|i|.a:|ea|sasse.|a:|veaaa|saisea|s::||a:|vew|:|:ese.::ama::|s
aaa|:|ea,:aa:|s,
A,uC)= ,Au) C , l )
aaa
A,u + C)= Au + AC, , l :)
:ev|aea:aa::|ema::|.esa:e aisa.| s|zes:aa::ae|aa|.a:eamai:|i|.a:|aasaaa
aaa|:|easa:eess||ie
sa:ma::|smai:|i|.a:|ea|saa:.emma:a:|ve 1|a:|s, |iAaaaua:e|a:a
ma::|.es,sa:|a:|e:|:|e:aaa.:sAuaaauAa:eaeaaeaaaaaave:|esam:
a|meas|eas- ) , :|ea,|a,eae:ai,
Au = uA

He:eave:,|:.aa|aea:|a:
Au=0 evea:|ea,| A = 0 aaa u= 0.
, l )
, l :)
samies|iias::a:|a,:|e|eaemeaa|a, l )aaa, l :)may|eiaaaa|a:|e:a|iems,
ai:|ea,|yea.aaeas|iy.eas::a.:yea:ewaesamiesas|a,ma::|.esw|:asmaii
|a:e,:aieiemea:s
Inverse Matrices
Asaa:ema::|s|ssa|a:e|aveorder 1|eidentity ma::|saia:ae:|s:a:
saa:ema::|s
0 0 0 0
0 l 0 0 0
0 0 l 0 0
=
0 0 0 l 0
, l )
0 0 0 0
352 Chapter 5 Li near Systems of Differenti al Equati ons
forwhlcheachentryontheprincipal diagonal lsianda||o-dlagona|entrlesac
zero. Itlsqulteeasytoverlfythat
n = n = n , i s;
forevery square matrlxnofthesameorderas
lfnls asquare matrlx, thenaninverse ofnls asquare matrlxuofthesame
orderasnsuchthat /o/
nu= and un=
lt ls not dlfncu|tto show that lfthe matrlx nhas an lnverse, then thls lnverse ls
unlque. Consequent|y, wemayspeakof/elnverseofn,andwewl|| denoteltby
n
-
,
Thus
, l )
glven the exlstence of n
-
,
lt ls c|ear that some square matrlces do not have
lnversesconslderany squarezeromatrlx. ltls a|soeasytoshowthatlfn
-
,
exlsts,
then (n
-
,
)
-
,
exlstsand,n
,
;

,
=n
ln|lneara|gebraltlsprovedthatn
-
,
exlstslfandon|ythedetermlnantdet(A)
ofthesquarematrlxnlsnonzero. Ifso,thematrlxnl ssaldtobenonsingular; lf
det(A) =0,thennlsca||edasingular matrlx.
Determinants
Weassumethatthestudenthascomputed2 2and determlnants lnear|ler
courses. lfn = .
,
] ls a 2 2 matrlx, then lts determinant det(A) =

ls
dennedas
Determlnantsofhlgherordermaybedennedbylnductlon,asfo||ows. Ifn= .
,
]
ls an matrlx, |etn
,
denote the ,- i ; ,- i ; matrlxobtalnedfromnby
de|etlnglts / throwand lts ]thco|umn. The exp.s/oofthedetermlnant

a|ong
lts / throw ls glvenby
n

=,
.
i ;

,
.
,
n
,
,.
andltsexpanslona|onglts]thco|umnlsglvenby
n

=,-i ;

,
.
,
n
,

(/ nxed) , (20a)
(]nxed) . (20b)
ltls shownln|lneara|gebrathatwhlcheverrowweuse lnEq. (20a)andwhlchever
co|umnweuseln Eq. (20b), theresu|tsare the sameln a|| zcases. Hence

ls
we||dennedbytheseformu|as.
cXump| e
cXump| e4
5. 3 Matri ces and Li near Systems 353
If
. `
-2
I -2
2 I ,
5
thentheexpanslonof

a|onglts secondrowls
n

= -4

-
,

+2

.
,
-
,

I

-
,
= -4 I I +2

I I - I
.
I I = -
Andtheexpanslonof

a|ongltsthlrdco|umnl s

n = -2

.

.
I
.

2
= -2 I6- I
.
I I +5 2 = -
Ca|cu|atorsandcomputersareconvenlentfortheca|cu|atlonofhlgher-dlnen
slona| determlnants and lnverse matrlces, but determlnants and lnverses of2 2
matrlcesareeasytocomputebyhand. Forlnstance,lfthe2 2matrlx
hasnonzerodetermlnant

=ad - /c= 0,thenltslnversematrlxls


n
-
.

I
d -/

c a

(2l )
Notethatthematrlxontherlght-handsldeofEq. (2I ) l s obtalnedfromnbylnter-
changlngthedlagona|e|ementsandchanglngtheslgnsoftheo-dlagona|e|ements.
lf


then

= 6

7 - 5 S = 2. HenceEq.(2I ) glves
n
-
,
=
!

7 -S

-4

.
2
-5 6 -
Youshou|dpausetoverlfythat

354 Chapter 5 Li near Systems of Differenti al Equati ons


cXump| C
Matrix-Valued Functions
Amatrix-valued function, orslmp|ymatrix fnnction, lsamatrlxsuchas
x,,;
x
:
,;
(22a) s,;=
x
-
,;
or
c, , ,; c; c,
-
,;
n, ; =
c
:
,,; c
::
,; c
:-
,;
(22b)
c
-
, ,; c
-:
,; c
--
,;
ln whlch each entry ls a functlon of We say that the matrlx functlon n,;ls
continuous (ordiferentiable) atapolnt(oronanlnterva|)lfeachofltse|ements
hasthesameproperty. Thederivative ofadlerentlab|ematrlxfunctlonlsdenned
bye|ementwlsedlfferentlatlon,thatls,
lf
then
and
n ,;=
.
=

c[

and .

,;=
co
I
s

'

sH
Thedlerentlatlonru|es
. .

. .=
.

.
. .
..=..

(23)

(24)
(25)
fo||ow readl|ybye|ementwlseapp|lcatlonofthe ana|ogousdlerentlatlonru|esof
e|ementary ca|cu|us for rea|-va|uedfunctlons. If c lsa (constant) rea| numberand
Cls aconstantmatrlx, then
.
(c.) = c,


.= C,

.
and .= -C

(26)
Because ofthe noncommutatlvlty ofmatrlx mu|tlp|lcatlon, lt ls lmportant not to
reversethe orderofthefactors ln Eqs. (25)and(26).
cXump| e
5. 3 Matri ces and Li near Systems 355
First-Order Linear Systems
Thenotatlon andtermlno|ogy ofmatrlces andvectors may seemrather e|aborate
when nrstencountered,butltls readl|yasslml|atedwlthpractlce. Ourmalnusefor
matrlx notatlon wl|| be the slmp|lncatlon of computatlons wlth systems ofdler-
entla|equatlons, especla||ythosecomputatlonsthatwou|dbeburdensomelnsca|ar
notatlon.
We dlscussherethegenera|systemofnrst-order|lnearequatlons
x,
=
;, .
,;x, +p;x
:
+

+;, -
,;x
-
+), , ; .
x=;
:.
,;x, +;
::
,;x
:
+

+;:-
,;x
-
+J,; .
x=
;
, ,;x, +;
:
,;x
:
+

+;-
,;x
-
+,; ,
,z:;
Ifwelntroducethe.oe./e-.o
andtheco|umnvectors
l
,;=[ ;
,
,;]
s = x ] and r,;= ],; ].
thenthesystemln,z:;takestheformofaslng|ematrlxequatlon
as
- =r,;s+r,;
a
,zs;
We wl|| see that the genera| theory ofthe|lnear system ln ,z:;c|ose|ypara||e|s
thatofa slng|enth-orderequatlon. Thematrlx notatlon used lnEq. ,zs;noton|y
emphaslzesthls ana|ogy, buta|sosavesagreatdea|ofspace.
A solution ofEq. ,zs; on the open lnterva| i ls a co|umn vector functlon
s,; = x,;] such thatthe componentfunctlons ofssatlsfy the systemln ,z:;
ldentlca||y on i lfthe functlons
;
,
,; and ],; are a|| contlnuous on i, then
Theorem I ofSectlon 5. I guarantees the exlstenceon iofa unlque so|utlons,;
satlsfylngpreasslgnedlnltla|condltlonss,.;=b.
Thenrst-ordersystem
x =1x, - x
:
.
x=:x, - :x
:
canbewrltten astheslng|ematrlx equatlon
Toverlfythatthevectorfunctlons
356 Chapter 5 Li near Systems of Di fferenti al Equati ons
are bothso|utlonsofthe matrlxdlfferentla|equatlonwlthcoemclentmatrlxP, we
needon|yca|cu|ate
: -c

,
:c

, ,
rs, =
:

zc

=
:c

=s
,
and

Tolnvestlgatethegenera| natureoftheso|utlonsofEq. ,zs), weconslderhrs|


theassociated homogeneous equation
as
- =r, ;s.
a
,z,
whlch has theformshown lnEq. ,zs) , but wlthr,; = 0. Weexpectl|to have
so|utlonss, .s

. .s
-
that are lndependentln some approprlate sense, and scca
thateveryso|utlonofEq.,z)lsa|lnearcomblnatlonofthesepartlcu|arso|ct|oas.
Glvenso|utlonss, .s

. .s
-
ofEq.,z), |etuswrlte

Thusx
,
,;denotesthe / thcomponentofthe vectors, ; . so the secondscbscr|pt
referstothevectorfunctlons
,
,; .whereasthenrstsubscrlptreferstoaconponea|
otthlsfunctlon. Theorem l lsana|ogoustoTheorem l ofSectlonz z
THEOREM 1 Pri nci pl e O Superposition
Letx
,
,X

.. . . , X, beso|utlonsofthehomogeneous|lnearequatlonln,z)on
theopeninterva|. If., ..

. ,.
-
areconstants, thenthe|lnearcomblnatlon
, i
l sa|soaso|utlonofEq. (29)on.
Proof: We know thats = r, s, for each / , i ;, so ltfo||ows
lmmedlate|ythat
s= ., s+.

s,+

+..s
= ., r,; s,+.

r,; s

+

+.
-
r,s
-
= r,; ,., s, +.

+

+.
-
s
-
;
Thatls, s= r,;s.asdeslred.Theremarkab|eslmp|lcltyofthlsproofdemonstrates
c|ear|yoneadvantageofmatrlxnotatlon.
cXump| C
COl/|lUCO
5. 3 Matri ces and Li near Systems 357
lfs,ands,arethetwoso|utlonsof
as_
4 -
a
6 -7
s
dlscussedlnExamp|e6, thenthe|lnearcomblnatlon
lsa|soaso|utlon. lnsca|arformwlths= x, x,}. thls glves theso|utlon
x,,;=., e
:

+ .
:
e
-
.
x, ,;=z.,
e
:

+.
:
e
-
.
whlch ls equlva|entto the genera| so|utlonwefoundbythe method ofe|lmlnatlon
lnExamp|ezofSectlon5. 2.
Independence and General Solutions
Llnearlndependencelsdennedlnthesamewayfor vector-va|uedfunctlonsasfor
rea|-va|uedfunctlons(Sectlonz z; Thevector-va|uedfunctlonss, .s,. ,s
-
are
linearly dependent on thelnterva| iprovldedthatthereexlstconstants., ..,. .
.
-
o.//zerosuchthat
., s,,;+.,s, ,;+

+.
-
s
-
,; =0 ,z;
for a|| ln i Otherwlse, theyarelinearly independent. Equlva|ent|y, they are
|lnear|y lndependentprovlded that no one ofthem ls a |lnear comblnatlon ofthe
others. Forlnstance,thetwoso|utlonss
,ands,
ofExamp|e6are|lnear|ylndepen
dentbecause,c|ear|y,neltherlsasca|armu|tlp|eoftheother.
Just aslnthe case ofaslng|enth-orderequatlon,therels aWronsklandeter-
mlnantthat te||sus whetherornot n glvenso|utlonsofthehomogeneousequatlon
ln,z;are|lear|ydependent. lfs, .s, . .s
-
are suchso|utlons, thenthelrWron
skian lsthen ndetermlnant
u,;=
x, , ,; x ,;
x,, ,; x,, ,;
x
- ,,; x
-, ,;
,;
uslngthenotatlonln,;forthecomponentsoftheso|utlons. We maywrlteelther
u,;oru,s,

s, . ,s
-
; Notethatulsthedetermlnantof thenatrlxthathasas
lts .o/-vectorstheso|utlonsss, . ,s
-
Theoremzlsana|ogoustoTheoren
ofSectlonz z Moreover, ltsprooflsessentla||ythesame,wlththedennltlonof
u,s, .s, . ,s
-
; ln Eq. ,;substltutedforthe dennltlonoftheWronsklanofn
so|utlonsofaslng|enth-orderequatlon(seeProb|ems42through44).
358 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| e
THEOREM Z Wronsklans of Sol utions
SupposethatXj , X , , X areso|utlonsofthehomogeneous|lneequatlon
x
'
= onanopenlnterva| i Supposea|sothatls contlnuousoniLet
Then.

IfX , X , , X, are |lne|ydependenton i ,then u = 0 ateverypolnt


of I .
IfX , X , , X are|lnear|ylndependenton i, then u = 0ateachpolnt
of I.
Thusthereareon|ytwoposslbl|ltlesforso|utlonsofhomogeneoussystems. El-
therw = cate.epolntofi , oru= 0atopolntofi
. . ... .. ............. ........
Itlsreadl|yverlned(aslnExamp|e:;that
areso|utlonsoftheequatlon

0
TheWronskanoftheseso|utlonsls
w =
ze
ze

ze
0
e .
e

-z
3
-i
z z z
z 0 -z = -i :e
.

.
i -i l
(34)
whlchls neverzero. Hence Theoremzlmp|lesthattheso|utlonsXj , X , andx

are
|lnear|ylndependent(onanyopenlnterva| ).

Theorem3 ls ana|ogous toTheorem1ofSectlon z z ltsaysthatageneral


solution ofthe/o-o,eeos system= ls a|lnearcomblnatlon
(35)
ofanyglven|lnear|ylndependentso|utlonsXj , X_ , , Xg
THEOREM 3 General Sol utions of Homogeneous Systems
LetX , X , ,Xbe|lnear|ylndependentso|utlonsofthehomogeneous|lne
_
equatlon= on an open lnterva| i , where ls contlnuous. If
ls anyso|utlonwhatsoever oftheequatlon = x on I, thenthere exlst
numbersc , c , c suchthat
(35)
fora||lni
5. 3 Matri ces and Li near Systems 359
Proof: Letabeanxedpolntofi Weshow nrstthatthereexlstnumbersc, ,
c
:
, ,c, suchthattheso|utlon
-
hasthesamelnltla|va|uesatt =a asdoestheglvenso|utlonx(t ) , thatls, suchthat
:
LetX(t) bethe matrlx wlthco|umnvectorsx, , x
:
, . . . , x, ,and|etcbethe
co|umnvectorwlthcomponentsc, ,c
:
, ,c, . ThenEq.: may bewrlttenlnthe
form
X(a) c=x(a) . s
TheWronsklandetermlnant w(a) = X(a) lsnonzero because the so|utlons x, ,
x
:
, . . . , x, are |lnear|ylndependent. Hencethe matrlx X(a)has an lnversematrlx
X(a)
,
. Thereforethevectorc =X(a)
,
x(a) satlsnesEq. s , asdeslred.
Flna||y, notethatthe glven so|utlon x(t) andthe so|utlony(t) ofEq. -
wlththeva|uesofc determlnedbytheequatlonc = X(a)
,
x(a)havethesame
lnltla|va|ues(att =a) . ltfo||owsfromtheexlstence-unlquenesstheoremofSec-
tlon | thatx(t)=y(t)fora||t ln i Thls estab|lshesEq. , A
Remark: Every systemx' =P(t )xwlth contlnuouscoemclentma-
trlx does have a set of |lnear|y lndependent so|utlons x, , x
:
, . . . , x, as ln the
hypothesesofTheorem ltsufnces to chooseforx (t) theunlqueso|utlon such
that
0
0
0
x (a) = 0
|
posltlon]
0
0
thatls, theco|umnvectorwltha||e|ementszeroexceptfora | lnrow]. (lnother
words, x (a)ls mere|ythe]thco|umnoftheldentltymatrlx. )Then
sotheso|utlonsx, ,x
:
, . . . ,x,
are|lnear|ylndependentbyTheorem2. Howactua||y
to nnd these so|utlonsexp|lclt|y ls anothermatteronethatweaddresslnSectlon
+ (forthecaseofconstantcoefnclentmatrlces).

Initial Value Problems and Elementary Row Operations


Thegenera|so|utlonlnEq. ofthehomogeneous|lnearsystemx' =P(t)xcan
bewrlttenlntheform
x(t)=X(t)c, -
360 Chapter 5 Li near Systems of Di fferenti al Equati ons
where
x,;= s, ,; s
:
,; ,+a
lsthe matrlxwhose.o/-.e.osarethe|lnear|ylndependentso|utlonss, ,
s
:
, ,s
-
.andc = ., .
:

.
-
ls the vectorofcoemclentslnthe|lnem
comblnatlon
Supposenowthatwewlshtoso|vethe///./../ep//e-
as
- = rs, s,.; = e,
a
,
,+|
where the lnltla| vector e =

ls glven. Then, accordlng to


Eq. ,-, ltsumcestoso|vethesystem
x,.;c=e ,+:
to nndthecoefnclents., ,.
:
. ,.
-
lnEq. ,
Wethereforerevlewbrleythee|ementarytechnlqueofoea./otoso|ve
an ./,e/./.|lnearsystem
c, , x, +c,
:
x
:
+

+c, -
x
-
=/, ,
c
:
x, +c
::
x
:
+

+c
:-
x
-
=/
:
,
,+
wlth nonslngu|arcoefnclent matrlx n = .[
,
[ ,constantvectore = /, andun-
knownsx, .x
:
. ,x
-
Thebaslc ldea ls to transform the systemln ,+, lnto the
slmp|erppe/.,/.)o
, , x, +,
:
x
:
+

+, -
x
-
=|, .

::
x
:
+

+
:-
x
-
=
|
:
.
,++
ln whlchon|ytheunknownsx
,
,x
,
., , ,x
-
appearexp|lclt|y ln the]thequatlon
(] = | , : ,; Thetransformedsystemls theneasl|yso|vedbytheprocessof
/../s/s//o Flrstthe|astequatlonln,++ lsso|vedforx
-
,thenthenext-to-last
lsso|vedforx
-
-
, .andsoforth,untl|thenrstequatlonlsnna||yso|vedforx,
The transformatlon ofthe system ln ,+, to upper trlangu|ar form ls most
easl|ydescrlbedlntermsofe|ementaryrowoperatlonsonthe .,-eea.oe./e
-./x
c, , c,
:
c,
-
/,
l
n e ] =
c
:
c
::
c
:-
/
:
l
,+ l
|
l
l
l
c
- , c
-:
c
--
/
-
that ls obtalnedbyadolnlngthevectoretothematrlxnasanaddltlona|co|umn.
Theadmlsslb|eelementary row operations areofthefo||owlngthreetypes.
cXump| eb
5. 3 Matri ces and Li near Systems 361
1. Mu|tlp|yany(slng|e)rowofthematrlxbyanonzeroconstant.
Z. Interchangeanytworowsofthematrlx.
J. Subtractaconstantmu|tlp|eofonerowfromanyotherrow.
The goa| lstouseasequenceofsuchoperatlons(onebyone,lntum)totrans-
form n b ] lnto an upper trlangu|armatrlx, onethathas on|y zerosbeneathlts
prlnclpa| dlagona| . Thls upper trlangu|ar augmented coefnclent matrlx then cor-
responds to an upper trlangu|ar system as ln ,++ The process of transfomlng
n b ]ls carrledoutoneco|umnatatlme,from|efttorlght, aslnthenextexam-
p|e.
- --_. _-------- ------_ ..-.. -. _-- ---- ....... . .. .......
Usetheso|utlonvectors glvenlnExamp|e: toso|vethelnltla|va|ueprob|em

-z 0

,+-
Sol ution Itfo||owsfromTheorem

thatthe|lnearcomblnatlon
lsagenera|so|utlonofthe

|lnearsystemln,+- Insca|arform,thlsglvesthe
genera|so|utlon
x,,;=z.,
e

+z.
:
e

+z.

.
x
:
,;=z., e

- z.

.
x

,;= ., e

- .
:
e

+ .

Weseekthepartlcu|arso|utlonsatlsfylngthelnltla|condltlons
x,,a =0, x
:
,;=z. x

,;=-
Whenwesubstltutetheseva|ueslnthethreeprecedlngsca|arequatlons,wegetthe
a|gebralc|lnearsystem
z., +z.
:
+z.

=0,
z., - z.

=z.
., - .
:
+ .

=-
wlth augmentedcoefnclentmatrlx
z z z 0
z 0 -z z
| -| | -
Mu|tlp|lcatlonofeachofthenrsttworowsby glves
| | 0
0 -| .
-i i -
362 Chapter 5 Li near Systems of Di fferenti al Equati ons
then subtractlon ofthe nrstrow bothfromthe second row andfrom theta|rdrow
glvesthe matrlx
| | |
+ - | -:
+ ~: 0

The nrstco|umnofthls matrlxnowhasthedeslredform


Nowwemu|tlp|ythesecondrowby- | , thenaddtwlcetheresu|ttotaeta|rd
row. Therebywegettheuppertrlangu|araugmentedcoefnclentmatr|x
|
+
thatcorrespondstothetransformedsystem
cj
( c
( ., = 0,
.
( :c, =-|
+., = +
Wenna||yso|velntumfor., = | c
=-,and., =: Thusthedeslredpart|cular
so|utlon lsglvenby
Nonhomogeneous Solutions
We nna||ytumourattentlontoa o/o-o,eeos|lnearsystemoftheforn
dx
dt
=P(t)x(f(t )

,47)
The fo||owlngtheorem ls ana|ogous toTheorem otSectlon : : and |s proved |a
preclse|ythe sameway, substltutlngtheprecedlngtheorems ln thls sectlonforthe
ana|ogous theorems of Sectlon : : In brlef, Theorem + means that the general
so|utlonotEq. +: hastheform
x(t)=x (t)( x, (t ) , ,4S)
wherex, (t)lsaslng|epartlcu|arso|utlonof Eq.+: andthecomplementary fonc-
tion x (t)lsagenera|so|utlonoftheassoclatedhomogeneousequatlonx' =P(t )x.
THEOREM 4 Solutions of Nonhomogeneous Systems
Letx be partlcu|arso|ution ofthenonhomogeneous |lnearequatlon ln +:
onanopenlnterva| I on whlch thefunctlonsP(t) andf(t ) arecontlnuous. Let
Xj , X , , X be |lnear|ylndependentso|utlonsofthe assoclatedhomogeneous
equatlononI . Ifx(t)lsanyso|utlonwhatsoeverofEq.+:, onI, thenthereexlst
numbers.j , c , , .

suchthat
(49)
fora||t lnI .
cXump| e
5. 3 Matri ces and Li near Systems 363
Thusnndlngagenera| so|utlonofanonhomogeneous|lnearsystemlnvo|ves
twoseparatesteps.
1. Flndlngthegenera|so|utlon. oftheassoclatedhomogeneoussysten,
Z. Flndlngaslng|epartlcu|arso|utlon., of thenonhomogeneoussysten
Thesum. = .

., wl||thenbeagenera|so|utlonof thenonhomogeneous
system.
Thenonhomogeneous|lnearsystem
x = . zx
:



x = . ., zx


,
x= .,
.

ls oftheformln.

wlth
-z a
_
-z ,


_



_
=


InExamp|e

wesawthatagenera| so|utlonoftheassoclatedhomogeneous|lnear
system
lsglvenby
.. _

-z a
.
=
'

.
andwecanverlfybysubstltutlonthatthefunctlon
(found uslng a computer a|gebra system, or perhaps by a hunan belng uslng d
methoddlscussedln Sectlon s ls apartlcu|arso|utlon oftheorlglna| nonhomo-
geneous system. Consequent|y, Theorem . lmp|les that a genera| so|utlon otthe
nonhomogeneoussystemls glvenby
that l s, by
. = z., e

z.
:
e


z.



., = z., e
- z.

,
x

,;= ., e
.
:
e

364 Chapter 5 Li near Systems of Di fferenti al Equati ons


Problems
1. Let
and B =
_
-4
}

Find (a) A 3B; (b) A

B; (c) AB; (d) BA.


2. Verify that (a) A(BC) = (AB)C and that (b) A(B+C) =
AB AC, where A and B are the matrices given in Prob
lem and
C =

. }
3. Find AB and BA given
A =

0
-4

4. Let A and B be the matrices given in Problem and let


and y = s.
cos t
Find Ay and Bx. Are the products Ax and By defned?
Explain your answer.
5. Let
A =

-5
Find (a) 4B; (b) 3A - 5B; (c) AB; (d) BA;
(e) A - .
6. Let
Al =

B =

}
(a) Show that A
l
B = A
2
B and note that Al i

Thus
the cancellation law does not hold for matrices; that is, if
A
l
B = A
2
B and B i 0, it does not follow that Al =

(b) Let A = Al -A
2
and use part (a) to show that AB = .
Thus the product of two nonzero matrices may be the zero
matrix.
7. Compute the determinants of the matrices A and B in
Problem Are your results consistent with the theorem
to the efect that
det(AB) = det(A)
.
det(B)
for any two square matrices A and B of the same order?
8. Suppose that A and B are the matrices of Problem 5. Ver
ify that det(AB) = det(BA) .
..9 ...,.,....., ..
.(AB)' = A'B AB'.
9. A(t ) =

10. A(t) =

...... .........
x' = -( ) x .
11. . =

= .
12. . = . = .
13. .= .4 = 5.

14. .= . cos t , = .

sin t
15. .= = . = .
16. .= . = .= 5

17. .= .

. = .

/
=

18. .= .

= .

= .

19. .= .

.= .

.= .

.= 4xI
20. .= .

.= .


. = . .

. = . .

.......,......
.....................
...........,.........
.......
21. x' =

_ _
.}
x; . =

[
.

22. x' =

} 4
x; .=

23. x' =
_
_
}
x; .=

}
24. x' =

}
x; . =

_ [.

}
27.
-4


-4
0 2

0
l

-6
x; Xl = e
O
'
0
-
l
O
t

,
X
2
= ' X3 =

Xi =
..31 ....40, ...,........
.......................
31. The system of Problem 22: Xl = X
2
= 5
32. The system of Problem 23: Xl = 5, X
2
() = -3
33. The system of Problem 24: Xl = X
2
= -7
34. The system of Problem 25: Xl = 8, X
2
() =
35. The system of Problem 26: Xl = X
2
() =
X3 () = 4
36. The system of Problem 27: Xl = X
2
() = 2,
X3 () =
37. The system of Problem 29: Xl = X
2
() = 2,
X () = 3
38. The system of Problem 29: Xl (O) = 5, X
2
() = -7,
X3 () =
b. | COl O
5. 3 Matri ces and Li near Systems 365
39. The system of Problem 30: Xl = X_ () X3 ()
X () =
40. The system of Problem 30: Xl = X_ () 3,
X3 () = 4, X () = 7
41. (a) Show that the vector functions
Xl =

,
[
and X
2
=

,[
are linearly independent on the real line. (b) Why does it
follow from Theorem 2 that there is .continuous matrix
-such that Xl and X
2
are both solutions of x
'
.
42. Suppose that one of the vector functions
Xl =

Xl l
[
X
2
l
is a constant multiple of the other on the open interval
Show that their Wronskian = [X
i
j must vanish
identically on This proves part (a) of Theorem 2 in the
case n = 2.
43. Suppose that the vectors Xl and X
2
(t) of Problem 42 are
solutions of the equation x
'
= -. where the 2 X 2 ma
trix -is continuous on the open interval Show that if
there exists a point .of at which their Wronskian .
is zero, then there exist numbers Cl and C
2
not both zero
such that Cl Xl .+
2
x_ (.) = . Then conclude from the
uniqueness of solutions of the equation x
'
= -.that
for all in that is, that Xl and X
2
are linearly dependent.
This proves part (b) of Theorem 2 in the case . 2.
44. Generalize Problems 42 and 43 to prove Theorem 2 for .
an arbitrary positive integer.
45. Let Xl X
2
. . . , X
n
be vector functions whose i th
components (for some fxed X
i
i X
i 2
. . . , X
in
are
linearly independent real-valued functions. Conclude that
the vector functions are themselves linearly independent.
| | Z: Z : Z | Z : : Z | :
:
| | Z Z Z
| Z *Z
L
| | | Z | b
Llnear systems wlth morethantwoorthree equatlons aremostfrequent|y so|ved
wlththealdofca|cu|atorsorcomputers. Forlnstance,reca||thatlnExamp|es we
neededto so|vethe|lnearsystem
| |
Z
b
| Z
| *_

| L
FIGURE 5.3. 1. TI-86 solution
of the system AC = B in I ).
Zcj ( Zc
z
( Zc
)
=c,
Zcj - Zc
)
=Z,
cj - c
z
( c
)
=-
|
that canbewrltten l nthe torm AC = u wlth 3 coemclent matrlx A, rlght-
hand sldethe | co|umnvectoru= c Z - andunknown co|umnvector
C = cj c
z
c
)
Flgure | showsaTI ca|cu|atorso|utlonforC =A
j
u,
wlth theresu|tthatcj = Z, c
z
= -, andc
)
= | OncethematrlcesAanduhave
beenentered,thesameresu|tcanbefounduslngtheM.p/ecommand
366 Chapter 5 Li near Systems of Di fferenti al Equati ons
C multiply ( inverse ( A) , B ) i
theM./e-./..command
C Inverse [ A] . B
ortheMATLAB commands
C inv ( A) *B
Useyourownca|cu|atororaval|ab|ecomputera|gebrasystemt oso|ve'automatl-
ca||yProb|ems | through40lnthls sectlon.
Wenowlntroduceapowerfu|a|tematlvetothemethodofe|lmlnatlonforconstruct-
lng the genera| so|utlon ofa/o-o,eeosnrst-order|lnearsystem wlth .os.
coefnclents,
x = c, , x, +c
:
x
:
+

+c, -
x
-
,
x
= c
:
x, +c
::
x
:
+

+c
:-
x
-
.
, | ,
ByTheorem ofSectlon , weknowthatltsufncestonnd|lnear|ylndependent
so|utlonvectorss, ,s
:
, ,s
-
,the|lnearcomblnatlon
,:,
wltharbltrarycoemclentswl||thenbeagenera|so|utlonofthesystemln, | ,
To search fortheneeded|lnear|ylndependentso|utlonvectors, weproceed
by ana|ogy wlth the characterlstlc root methodforso|vlng a slng|ehomogeneous
equatlonwlthconstantcoemclents(Sectlon: Itlsreasonab|etoantlclpateso|u-
tlonvectorsoftheform
x, c,
e

:,
x
:
:
:
e

:
:
s,;=
x

= ve

(3)
x
- c
-
e

:
-
where, :, ,:
:
, ,:
-
areapproprlatesca|arconstants. Forlfwesubstltute
,/= | : , ;
ln , | , then each term ln the resu|tlng equatlons wl|| have the factor e

, so we
can cance| lt throughout. Thls wl|| |eave us wlth |lnear equatlons whlchfor
approprlateva|uesofwecanhopetoso|veforva|uesofthe coefnclents :, ,:
:
,
,:
-
ln Eq. ,, sothats,;= ve

ls, lndeed,aso|utlonofthesystemln, | ,
Tolnvestlgatethlsposslbl|lty, ltls moreefnclentto wrltethesystemln , i ; ln
thematrlxform
s= Ax (4)
5. 4 The Ei genval ue Method for Homogeneous Systems 367
where
_
= .
,
] Whenwesubstltutethetrla| so|utlonx = ve

wlthderlvatlve
x
'
= ve

ln Eq. ,+, theresu|tls


We cance|thenonzerosca|arfactore

toget
_
v = v ,,
Thlsmeansthatx =ve

wl||beanontrlvla|so|utlonofEq. ,+, provldedthatvlsa


o,eovectorandlsaconstantsuchthatEq., ho|ds,thatls, the-.opa.
_
v/s.s../.-//p/eo)/e.e.ovThequestlonnowlsthl s. Howdowenndv
and:
To answerthlsquestlon,werewrlteEq., l ntheform
(
_
-

)v= 0. -
Glven,thlslsasystemofhomogeneous|lnearequatlonslntheunknowns:, ,:
:
.
, :
-
By astandardtheoremof|lneara|gebra, lthasanontrlvla| so|utlonlfand
on|ylfthedetermlnantoflts coefnclentmatrlxvanlshes,thatls, lfandon|y lf

_
-

=det(A-

)=a ,:
Inltsslmp|estformu|atlon,theeigenvalue method forso|vlngthesystemx =
_
x
conslstsofnndlngsothat,: ho|dsandnextso|vlngEq. - wlththlsva|ueof
toobtaln::
:
. , :
-
Thenx =ve

wl||beaso|utlonvector. Thenameofthe
methodcomesfromthefo||owlngdennltlon.
DEFI NITI ON Eigenval ues and Eigenvectors
Thenumber(eltherzeroornonzero)lsca||edaneigenvalue ofthematrlx
Aprovldedthat

_
-

= 0 ,:
Aneigenvector assoclatedwlththeelgenva|uelsao,eovectorvsuchthat
Av= v,sothat
(
_
- )v=0. -
Notethatlfvlsanelgenvectorassoclatedwlththeelgenva|ue,thensolsany
nonzeroconstantsca|armu|tlp|ecvofvthlsfo||owsuponmu|tlp|lcatlonofeach
sldelnE-, by.= a
Theprenxe/,elsaGerman wordwlththeapproxlmatetrans|atlon./..
e/s/.lnthlscontext,theterms./...e/s/.../eand./.m.e/s/..e.oare
lncommonuse. Forthlsreason,theequatlon
._ - .
. :
., -
.
:.
.
::
- .
:-

_
-

= = a ,s
.
- , .
-:
.
--
-
368 Chapter 5 Li near Systems of Di fferenti al Equati ons
ls ca||ed the characteristic equation ofthe matrlx n, l tsroots are the elgenval -
ues ofn Upon expandlng the determlnant ln ,s weevldent|y getan nth-degree
po|ynomla|oftheform
-
By the tundamenta| theoremofa|gebra, thlsequat|onhas rootsposs|b|y sone
arecomp|ex,posslb|ysomearerepeatedandthusanmatr| xhaselgenval-
ues (countlngrepetltlons,lfany) A|though weassumethatthe e|ements otnare
rea| numbers, we a||owthe posslbl|ltyofcomp|exelgenva|uesand comp|ex-valued
elgenvectors
OurdlscusslonofEqs ,+ through : provldesaproofofthetol |ow|ngthe-
orem, whlch ls the basls for the elgenva|ue method otso|v|ng a nrst-order ||near
systemwlthconstantcoemclents
THEOREM 1 Eigenval ue Sol utions of x' " Ax
Let be an elgenva|ue ofthe constant] coefnclent matrlx nofthe nrst-order
|lnearsystem

as
- = ns
a
Ifvlsanelgenvectorasociatedwlth, then
s,;=ve

'
ls anontrlvla|so|utlonofthesystem.
The Eigenvalue Method
Inout|lne,thlsmethodforso|vlngthehomogeneousconstant-coemclentsys-
tems=nsproceedsasfo||ows.
1. Wenrstso|vethecharacterlstlcequatlonln,s fortheelgenva|uesj , ,, .

.ofthematrlxn
Z. Next we attempt to nnd //e./y /aepeaeelgenvectors Y( , Y , . . . , Y,
assoclatedwlththeseelgenva|ues
3. Step : ls not a|ways posslb|e, but when ltls, weget |lnear|y lndependeat
so|utlons
( l 0)
Inthls casethegenera|so|utlonofs=nsls a|lnearcomblnatlon
oftheseso|utlons.
We wl||dlscussseparate|ythevarlouscasesthatcanoccur, dependlngonwhether
the elgenva|ues are dlstlnct or repeated, rea| or comp|ex The case of repeated
elgenva|uesmu|tlp|erootsofthecharacterlstlcequatlonwl||bedeferredtoSec-
tlon -
cXump| C 1
5. 4 The Ei genval ue Method for Homogeneous Systems 369
Distinct Real Eigenvalues
Ifthe elgenva|ues , .
:
. .
-
are rea| and dlstlnct, then wesubstltute each ot
them ln turn lnEq. (6) and so|ve tor the assoclated elgenvectors Vj , V, . . . , V
In thls case lt can be proved that the partlcu|ar so|utlon vectors glven ln , i are
a|ways |lnear|y lndependent. (For lnstance, see Sectlon 6. 2otEdwards and Pen-
ney, i/e-e.i/e.~/,e/.(Eng|ewoodC|ls, NJ. PrentlceHa||, i ss ln
any partlcu|arexamp|e such |lnear lndependence can a|ways be verlhed by uslng
the Wronsklan determlnant ofSectlon The to||owlng examp|e l||ustrates the
procedure.
Flndagenera|so|utlonofthesystem
x :x,
(
zx
:
.
x x, - x
:

, i i
Sol uti on Thematrlxformofthesystemln, i i ls
s ,

, x. ( i 2
Thecharacterlstlcequatlonofthecoemclentmatrlx ls
2
.

:
- - i ,( z , .
sowehavethedlstlnctrea|elgenva|ues
, -2and
:

ForthecoemclentmatrlxAlnEq.( l 2)theelgenvectorequatlon,ntv= 0
takestheform
, i
fortheassoclatedelgenvectorV . /
CASE 1 : , -2. Substltutlonofthenrstelgenva|ue, -2l nEq., i ylelds
thesystem
thatls, thetwosca|arequatlons
6.( z/.
.( /a
( i .
Incontrastwlth thenonslngu|ar(a|gebralc)|lnearsystemswhoseso|utlonswedls-
cussed ln Sectlon . the homogeneous |lnear system ln ( i . ls s/,/.:a-
twosca|arequatlonsobvlous|yare equlva|ent (each belng amultlp|eottheother).
Therefore, Eq. .haslnnnlte|ymanynonzero so|utlonswecan choose.arbl-
trarl|y(butnonzero)andthenso|vefor/
370 Chapter 5 Li near Systems of Di fferenti al Equati ons
Substltutlonofanelgenva|ue lnthe elgenvectorequatlon (A- v= 0
a|waysyle|dsaslngu|arhomogeneous|lnearsystem,andamong ltslnnnltyofso|u-
tlons wegenera||yseekaslmp|eso|utlonwlthsma||lntegerva|ues(lfposslb|e).
Looklngatthesecondequatlonln( l 4), thecholce.= | yle|ds/= -. andthus
ls anelgenvectorassoclatedwlth , = -z(as ls anynonzeroconstantmu|tlp|eot
l
Remark: I flnsteadof theslmp|estcholce.= | /= -, wehadmade
anothercholce.=.=0, /= -., we wou|dhaveobtalnedtheelgenvector
Vj =

-
|= .|
Becausethlslsaconstantmu|tlp|eofourprevlousresu|t,anycholcewemake|eads
to(aconstantmu|tlp|eofthesameso|utlon
CASE 2:
:
= 5. Substltutlonofthesecondelgenva|ue= 5l n, i yle|dsthe
palr
-.+z/= 0,
.- :/=0
( l 5)
ofequlva|entsca|arequatlons. Wlth/= | lnthenrstequatlonweget.= z,so
lsanelgenvectorassoclated wlth
:
= 5. A dlerentcholce. = z.,/ = . =0
wou|dmere|yglveaconstant]mu|tlp|eofv
:
.
Thesetwoelgenva|uesandassoclatedelgenvectorsyle|dthetwoso|utlons
Theyare|lnear|ylndependentbecausethelrWronsklan

e
:
ze

:

=:e
- e e
ls nonzero. Henceagenera|so|utlonofthesystemln, | | , ls
lnsca|arform,
x,,= .je
-
:

+z.
:
e

.
x
:
,= -., e
-
:

+ .
:
e

+
1
l
-l
-1
-+ -~

FIGUR 5.4. 1. Direction feld


and solution curves for the linear
system x = ..+ .

x = . .

of Example
,.c.-,
l
\gm)

l
FIGUR 5.4.2. The three brine
tanks of Example
cXump| eZ
5. 4 The Ei genval ue Method for Homogeneous Systems 371
Flgure : i shows some typlca| so|utlon curves ofthe system l n | | We see
twofaml|lesofhyperbo|as sharlng thesamepalrofasymptotes. the|lnex, = 2x
:
obtalnedfrom the genera| so|utlon wlth ., = .andthe|lnex
:
= -3x, obtalned
wlth .
:
= Glvenlnltla| va|uesx,,; = /, , x
:
,; = /
:
,ltls apparenttromthe
ngurethat

If ,/,

/
:
;|lesto the rlght ofthe |lnex
:
= -x, .thenx,,;andx
:
,;both
tendto+>as +>,
If,/, . /
:
;|lestothe|eftofthe|lnex
:
= -3x, g thenx,,;andx
:
,;bothtend
to->as- +>
z
Remark: AslnExamp|ei , ltlsconvenlentwhendlscusslnga|lnearsystem
x
'
= Ax touse vectorss, .s
:
, ,s
-
todenotedlfferentvector-va|uedsolutlonsot
the system, whereas thes../.sx, ,x
:
, ,x
-
denotethecomponentsotaslng|e
vector-va|uedso|utlonx. z
Compartmental Analysis
Frequent|yacomp|exprocessorsystemcanbebrokendownlnto slmp|ersubsys-
tems or compartments that can be ana|yzed separate|y. The who|e system can
then be mode|edby descrlblngthe lnteractlonsbetweenthevarlouscompartments.
Thusachemlca|p|antmayconslstofasuccesslonofseparatestages(orevenphys-
lca|compartments)ln whlchvarlous reactants andproducts comblneoraremlxed.
Itmayhappenthataslng|edlerentla|equatlondescrlbeseachcompartmentotthe
system, andthenthe who|e physlca| systemlsmode|ed byasystemofdlerentlal
equatlons.
As a slmp|e examp|e ofa three-stage system, Flg. : z shows three brlne
tanks contalnlng r, , r
:
, and r

ga||ons ofbrlne, respectlve|y. Fresh waterHows


lntotank i , whl|emlxedbrlneHowsfromtank i lntotank2,fromtank2lntotank
3, and out oftank 3. Let x,;denote the amount (ln pounds) of sa|tln tank / at
tlmefor / = i , z, and 3 . IfeachHowratelsga||onspermlnute,thenaslmp|e
accountlngofsa|tconcentratlons, as lnExamp|e 2 ofSectlon i , yle|dsthe hrst-
ordersystem
where
x = -/, x, ,
x= /, x, - /
:
x
:
,
x
= /
:
x
:
- /

/ =
r

/ = I , 2, 3.
| -
| :
Ifr, = z. r
:
= :, r

= c, = i (ga|Jmln),andthelnltla|amountsofsa|tln
the threebrlnetanks,lnpounds,are
nndtheamountofsa|tln eachtankattlme
372 Chapter 5 Li near Systems of Di fferenti al Equati ons
Sol uti on Substltutlng the glven numerlca| va|ues ln , i 6;and , i : , wegetthe lnltla| va|ue
prob|em
-0. 5 0. 0 0. 0
s ,;= c -0. 25 0. 0 s.
0. 0 0. z -0. z
forthevectors,;= x,,; x
:
,; x

,;[ Theslmp|eformof thematrlx


-0 5 - 0. 0
A - = c - z-
c z
|eadsreadl|ytothecharacterlstlcequatlon
c c
c c
-c z -
A- = ( -0. 5- ; ,-0. 25- ; ,-c z- ;= a
( l S)
, | -
ThusthecoemclentmatrlxAln, | S;hasthedlstlnctelgenva|ues, = -c ,
:
-
- z. and

= -c z
CASE 1: , = -0. 5. Substltutlng= -0. 5ln, i . wegettheequatlon
c 0. 0 0. 0 c
A+ (0. 5)

]v = 0. 5 0. 25 0. 0 b = c
0. 0 c z 0. 3 c c
fortheassoclatedelgenvectorv = . b c ]

. The|asttworows, afterdlvlslon
by0. 25 and0. 05, respectlve|y,yle|dthesca|arequatlons
2a + b = 0,
5b+ 6c= a
Thesecondequatlonl ssatlsnedbyb = -6andc = 5, andthenthenrstequatlon
glves= 3. Thustheelgenvector
,
= 3 -6 5[
ls assoclated wlththeelgenva|ue, = -0. 5.
CASE 2:
:
= -0. 25. Substltutlng= -c zln, i , wegettheequatlon

- z
A+ (0. 25)
.
]v = . 5
+
+
0. 25
fortheassoclatedelgenvectorv = . b c ]

. Eachofthenrsttworowslmp|les
that= 0, anddlvlslonofthethlrdrowbyc glvestheequatlon
5b + c = c.
whlchlssatlsnedbyb = | c = -5. Thustheelgenvector

:
= c l s [
ls assoclatedwlththeelgenva|ue
:
= -0. 25.
l 0
FIGUR 5.4.3. The salt content
functions of Example
5. 4 The Ei genval ue Method for Homogeneous Systems 373
CASE J.

= - z Substltutlng= - zln, i , wegettheequatlon


A+, z;

]= o -o o

-
z
for the elgenvector The nrst and thlrd rows lmp|y that . = , and / .
respectlve|y,butthea||-zerothlrdco|umn|eaves.arbltrary(butnonzero) Thus

= i
ls anelgenvectorassoclatedwlth

= - z
Thegenera|so|utlon
thereforetakestheform
s,; .
.
,

e-
,

+.
:

_
e
-
:

+.

-
:

Theresu|tlngsca|arequatlonsare
x, ,;= .,
e
-


,
x
:
,;= -:.,e
-


+ .
:
e
-

:
,
x

,;= ., e
-


- .
:
e
-

:

+ .

e
-

When welmposethelnltla|condltlonsx,,; = i , x
:
,; = x

,; = ,wegetthe
equatlons
., = i .
-:., + .
:
= ,
., - .
:
+ .

=
thatarereadl|yso|ved(lntum) tor., = , .
:
= , and.

= i z Thus,nna||y,
theamountsofsa|tattlmelnthethreebrlnetanksareglvenby
x, ,;= i e
-


.
x
:
,;= -e
-


+ e
-

.
x

,;= ze
-


- i oe
-

:

+ i ze
-

Flgure : showsthegraphs ofx,,; ,x


:
,; ,andx

, ; Aswewou|dexpect, tank
i lsrapld|yHushedbythelncomlngfreshwater,andx,,;-- as-- +. The
amountsx
:
,;andx

,;ofsa|tlntankszandpeaklntumandthenapproachzero
asthewho|ethree-tanksystemls purgedotsa|tas-- +c. z
374 Chapter 5 Li near Systems of Di fferenti al Equati ons
Complex Eigenvalues
Evenlfsomeoftheelgenva|uesarecomp|ex,so|ongastheyaredlstlnctthemethod
descrlbedprevlous|ystl||yle|ds|lnear|ylndependentso|utlons. Theon|y comp|l-
catlon ls that the elgenvectors assoclated wlth comp|ex elgenva|ues areordlnarl|y
comp|exva|ued,sowewl||havecomp|ex-va|uedso|utlons.
Toobtalnrea|-va|uedso|utlons,wenotethatbecauseweare assumlngthat
thematrlxA has on|yrea| entrlesthecoefnclentsln the characterlstlcequatlonln
(S)wl||a||berea| . Consequent|yany comp|exelgenva|uesmustappearln comp|ex
conugate palrs. Supposethenthat = p +q/andz = - q/are such apalrot
elgenva|ues. Ifvls anelgenvectorassoclatedwlth,sothat
(A- ) v= 0,
thentaklngcomp|exconugateslnthlsequatlonyle|ds
(A- z)V= 0
slnceA = AandI = (thesematrlcesbelngrea|) andtheconugateofacomp|ex
product ls theproductoftheconugates ofthefactors. Thus theconugateV otV
ls anelgenvectorassoclatedwlthz. Ofcoursetheconugateofa vector ls denned
componentwlse, lf
(20)
thenV = a- b/ . Thecomp|ex-va|uedso|utlonassoclatedwlthandvls then
thatl s,
x(t) = ve

=ve


= (a+e/ ; e

(cosqt+/slnqt ) ,
x(t) = e

(acosqt - bslnqt ) +/ e

(bcosqt +asln qt ) . (2l )


Because the rea| and lmaglnary partsofa comp|ex-va|ued so|utlon are a|so solu-
tlons, wethusgetthetwoe./../easo|utlons
x
(
(t) =Rex(t ) ] =e
o
' (acosqt - b sln qt ) ,
x
:
(t) =Imx(t ) ] =e
o
' (b cos qt +a sln qt )
(22)
assoclatedwlththecomp|ex conugateelgenva|ues q/ . Itls easytocheckthat
the same two rea|-va|ued so|utlons resu|tfrom taklng rea| and lmaglnary parts ot
ve' Rather than memorlzlng the formu|as ln (22), lt ls preferab|e ln a speclnc
examp|etoproceedasfo||ows.

Flrstnndexp|lclt|yaslng|ecomp|ex-va|uedso|utlonx(t)assoclated wlththe
comp|exelgenva|ue,

Then nnd therea|andlmaglnarypartsx


(
(t)andx
:
(t)togettwolndependent
rea|-va|uedso|utlonscorrespondlngtothetwocomp|exconugateelgenva|ues
andz.
cXump| e
5. 4 The Ei genval ue Method for Homogeneous Systems 375
Flndagenera|so|utlonofthesystem
dx,
= 4x, - 3x
:
,
dt
dx
:
= 3x, + 4x
:
.
dt
(23)
Sol ution Thecoefnclentmatrlx
FIGUR 5.4.4. Direction feld
and solution curves for the linear
system x; = 4x, - 3X
2
,
X = 3x, + 4X
2
of Example 3.

4 -3
A =
3 4
hascharacterlstlcequatlon
andhencehasthecomp|exconugateelgenva|ues=4- 3/ andz=4+3/ .
Substltutlng=4- 3/ lntheelgenvectorequatlon- v=0,wegetthe
equatlon
A- (4- 3/ )

} v=

for an assoclated elgenva|uev = a bDlvlslonofeachrow by 3 yle|ds the


two sca|arequatlons
/a - b = 0,
a + /b = 0,
eachofwhlchl ssatlsned bya = i and b = / . Thusv = i / lsacomp|ex
elgenvectorassoclatedwlththecomp|exelgenva|ue=4- 3/ .
Thecorrespondlngcomp|ex-va|uedso|utlonx(t) = ve

ofx' =l s then

@
i

( 3
.
3 )
@
cos3t - /sln3t
|
x t

e cos t - SH t e
3 +
.
3
.
cos t SH t
Therea|andlmaglnarypartsof x(t)aretherea|-va|uedso|utlons
( )
-

cos3t
x, t

e .
3 SH t

- sln3t
and x
:
(t) = e
cos 3t

Area|-va|uedgenera| so|utlonofx' =Axls then glven by

c, cos3t - c
:
sln3t

x(t)= c, x,(t)+c
:
x
:
(t) = e .
3t + 3t

c, SH c
:
cos
Flna||y,agenera|so|utlonofthesystemln(23)lnsca|arformls
x, (t) = c

(c, cos3t - c
:
sln3t) ,
x
:
(t) = c

(c, sln3t +c
:
cos3t ) .
Flgure5. 4. 4shows sometyplca| so|utloncurvesofthesysteml n(23) . Each
appearstosplra|counterc|ockwlseasltemanatesfromtheorlglnlnthex, x
:
-p|ane.
Actua||y,becauseofthefactorc

lnthegenera|so|utlon,weseethat
376 Chapter 5 Li near Systems of Di fferenti al Equati ons
FIGURE 5.4.5. The three brine
tanks of Example 4.
cXump| C4
A|ong each so|utlon curve, the polnt (XI (t ) , X2 (t ) ) approaches the orlgln as
t - -, whereas
The abso|uteva|uesofXI (t) and X2 (t) both lncrease wlthoutbound as t --
+> z
Flgure : shows a c|osed system ofthreebrlne tanks wlth vo|umes VI ,
V2 , and v

Thedlfferencebetweenthls systemandthe'opensystemofFlg. : z
ls thatnowthelnHowtotank ls theoutHowfromtank3. Wlththe same notatlon
aslnExamp|ez.theapproprlatemodlncatlonotEq , i :ls
dXI

= /, x,
dt
dX2
dt
whereki =JVi asln, i :
,z:
Flnd theamountsXI ( t) , X2 (t) , andX3 (t) ofsa|tattlme t lnthethreebrlne tanks ot
Flg. : lfVI =ga|, V2 =zga|,V3 =ga|, and= i ga|Jmln
Sol ution Wlththeglvennumerlca|va|ues,,z:takestheform
- = c z : x
dx

- z z
dt
: - z
,z
wlthx = XI X2 X3 asusua|. Whenweexpandthedetermlnantofthematrlx
- z- c z
A-

= c z - :
c : z-
a|onglts nrstrow, we nndthatthecharacterlstlcequatlonofls
, -c z- , -c :- , - z- + , z , z , :
=

3
- , s

:
, z

= -,+ :
2
+ , z
2
]=c
,z:
ThusA hasthezeroelgenva|ue

= andthecomp|exconugateelgenva|ues,
z= - :+,c z /
CASE 1 : ,=+ Substltutlonot=clnEq. ,z:glvestheelgenvectorequatlon
- z z .
(A- c

= z - : / =
: - z c
. 4 The Ei genval ue Method for Homogeneous Systems 377
forv = a b c . Thehrstrowglvesa = c andthesecondrow glves a = 2b,
sovo = 2 l 2 ]
T
ls anelgenvectorassoclatedw| ththeelgenval ceA
O
= O. The
correspondlngsolutlonxo (t ) = voeAOI otEq. (25) lstheconstantsolutlon

(t ) =

(27)
CASE 2: = -0. 4 - (0. 2) i .
theelgenvectorequatlon
Substltutlonot= -0. 4 - (0. 2) i lnEq.(26) glves
0. 2 + (0. 2) i
[A - ( -0. 4 - (0. 2) i ) I] v = 0. 2
0. 0
0. 0
(0. 2) i
0. 4
0. 2
0. 0 /
0. 2 + (0. 2) i c
The secondeqcatlon (0. 2)a + (0. 2) i / = 0 is sati shedbya = and/ = i . Then
thehrstequatlon
[ 0. 2 + (0. 2) i ]a + (0. 2) c = 0
glvesc = l - i . Thusv =
;
i ( l - i ) lsacomplexelgenvectorassocl-
atedwlththecomplex elgenvalue A = -0. 4 - (0. 2) i .
Thecorrespondlngcomplex- valuedsol utlonx( t ) = veAl ot(25) ls
x( t ) = l
= l
; - i

T
e( -0
.
4-0
. 2i
) 1
l - i e( -0. 4) ' (cos O. 2t - i si n O. 2t )
= e( -0
.
4)
1
sln0. 2t + i cos0. 2t .
0. 2t - i sln0. 2t
cos0. 2t - sln0. 2t - i cos0. 2t + i sln0. 2t
Therea|andlmaglnarypartsotx( t ) arethereal -valuedsol ctlons
A[ (t ) = e( -0. 4)1 sln0. 2t ,
cos O. 2t
cos0. 2t - sla0. 2t
-
X2 ( t ) = e( -0. 4)1 cos O. 2t .
cos0. 2t + sln0. 2t
Thegeneral so|utlon
hasscalarcomponents
XI (t ) = 2co + e( -0. 4) 1 Cj cos0. 2t - c_sln0. 2t ) ,
(28)
X
2
( t ) = Cg + e( -0. 4) 1 C sln0. 2t + C_ cos0. 2t ) , (29)
X
3
( t ) = 2co + e( -
O
. 4) I [ ( Cj - C
2
) cos0. 2t + ( C + C
2
) s|n0. 2t ]
378 Chapter 5 Li near Systems of Di fferenti al Equati ons
glvlngtheamountsofsa|tlnthethreetanksattlme
Observethat
(30)
Ofcoursethetota|amountofsa|tlnthec|osedsystemlsconstant, theconstantC
o
ln(30)lsone-nfththetotal amountofsalt. Becauseotthetactorsofe

a
+
ln(29),
weseethat
|lm x, ,= 2c, , |lm x
:
,= c, , and |lm x

,= 2c,.
l = l = l =
l0
FIGUR 5.4.6. The salt content
functions of Example .
Thusas +>thesa|tlnthesystemapproachesase.a,s.edlstrlbutlonwlth
40ofthe sa|t ln eachofthe two 50-ga||ontanks and20 ln the25-ga||ontank.
So whateverthelnltla|dlstrlbutlonofsa|tamongthethreetanks,the|lmltlngdlstrl-
butlon ls one ofunlformconcentratlonthroughoutthesystem. Flgure 5. 4. 6shows
thegraphs ofthethreeso|utlonfunctlonswlth c, = l 0, c, = 30, andC
:
= -I 0, la
whlchcase
..1 ....16, .,,..........
..................
............,....,.....
...,...,......,...
.............,....
........
1. ...

...

2. ...

...

3. ....

.. .

. .

4. ...

.. .

5. .. .

.. .

6. . .

. . .

. .


7. . . ..

.. .

... .

.. .

9. .. .

... .

. .

10. .. .

...

11. . .

...

..

.
12. .. .

...

... .

.. .

14. .. ..

....

... .

....

16. ...

. . .

..1 7 ....25, .......,.


..........,........,,..
....................
.. .. .

..

. . .

....

..

.....

...

...

.....

....

...

..

x,(0) = 50 and x
:
(0)=x

(0)=c
20. .. .

...

...

21. .. .

.. .

... .

..

22. . . .

. . ..

...

23. .. .

..
.

...

24. ...

... .

.....

25. . . .

.. .

...

26. Find the particular solution of the system


..
.
..

. ..

.
that satisfes the initial conditions ..


....... ....

.. ........
.5.4. 7 ...,.. ..,....
......27 ...28 .
.......

..........

..
.......... ....


5. 4 The Ei genval ue Method for Homogeneous Systems 379
..................2. ...
..............,.......


!tesawatet
!|ewtate
FIGURE 5.4.7. The two brine tanks of
Problems and
27. . = (gal), .

= (gal)
28. .= (gal ), .

= .(gal)
...........

..........
.5.4. 8 ...,.., ..,....
...= .........29 ...30, .
.....

........= ....=
....

= ...............
.,.......


FIGURE 5.4.8. The two brine tanks of
Problems and
29. . = (gal), .

= (gal)
30. .= (gal), .

= .(gal)
.31 ....34 .....,........
.5.4. 2. .........1; ....
...1 ....2, ...2 ....3, .......
3; ..........,......
.....= .,.

= ....

= ..
................. .

..
.

.................
.....................
...3 ...................
.,... .

....


31. = .,= . = .

= .

=
32. = .,= . .= .

= .

=
33. = .,= . .= .

= .

=
34. = .,= ..= .

= .

=
.35 ....37 ............
.5.4. 5, .........,....
........1 ...2, ...2 .
...3, ... ...3 ...1, .....
....,...........= .,,...
.

= ....

= .........
....... .

....

.......
..................
......... +0) .......
................,... .


....


35. = .,= . = .

= .

= .
36. = .,= .= .

= .

=
37. = .,= .= .

= .

=
....A ....38 ....40, ..
..........,.......
..........x' = Ax.





.

39. A =






40. A =


41. The coefcient matrix A of the .Z .system
.= .. .

+ .

.= . + ..

+ .

.= . + .

..

+ .

.= . .

+ .

..

has eigenvalues Al =

= and

= Find the particular solution of this system that


satisfes the initial conditions
..42 ....50, ......,...
...................
....5. 4 .,,.......
.........x' = Ax ....,
.A.
. .

42. A = .
.

43. A =
.
380 Chapter 5 Li near Systems of Di fferenti al Equati ons
.
44.

48.
.
.

. .
[ . .
]
45.
.
49.
.


46.





. .



50.


47.
.

.
. 4| COl O
. > . - .
| Z- . Z ! ! *H
| | . . . !
| Z . !
| . Z Z ! !
e 9V! H
e 9VC H

| | . . I . !
| . I . Z . !
| I . . I . 67 ! !
Nh'T > L| 1 M ` > L` L
Q@ .

FIGUR 5.4.9. .
calculation of the eigenvalues and
eigenvectors of the matrix
Mostcomputatlona| systemsofferthecapabl|ltytonndelgenva|uesandelgenvec-
torsreadl|y. Forlnstance,Flg. : showsagraphlngca|cu|atorcomputatlonofthe
elgenva|uesandelgenvectorsofthematrlx
A = c -c z c c
-c c c c c
c c c z -c z
ofExamp|ez We see each elgenvectordlsp|ayed as a co|umn vector beneath lts
elgenva|ue. Notethatwlthresu|tspresentedlndeclma|form, ltls uptoustoguess
(andverlfybymatrlxmu|tlp|lcatlon)thattheexactelgenvectorassoclatedwlththe
thlrd elgenva|ue = 1 ls = | -z Once the matrlx A has been
entered,theM.p/ecommand
eigenvects ( A) ;
theM./e-./..command
Eigensystem[ A]
or theMA1LAucommand
[ V, D] eig ( A)
(whereD wl||beadlagona|matrlxdlsp|aylngtheelgenva|uesofA andtheco|umn
vectors ofV arethe correspondlngelgenvectors) produce slml|arresu|ts. You can
usethesecommandstonndthe elgenva|ues andelgenvectorsneededforanyofthe
prob|emsln thlssectlon.
Foramoresubstantla|lnvestlgatlon, chooseaposltlvelnteger < i c(=.
forlnstance)and|et,
, .,
:
. .,
-
denotethenrstnonzerodlgltsln your student
ID number. Now conslderan open system ofbrlne tanks as ln Flg. : z. except
wlthratherthanthreesuccesslvetanks havlngvo|umes r = i c, ,/ = i . z. .
;ln ga||ons. Ifeach How ratels
= | cga||onspermlnute,thenthe sa|t amounts
x, , . x
:
, . .x
-
,satlsfythe|lnearsystem
x = -/, x
, .
x= /
-
, x
-
, - / x
,/= z.3, .; .

aaa




X l X2 X3
FIGUR 5.5. 1. Three
spring-coupled masses.
5. 5 Second-Order Systems and Mechani cal Appl i cati ons 381
where / = rJ. App|y the elgenva|ue method to so|ve thls system wlth lnltla|
condltlons
Graphtheso|utlonfunctlonsandestlmategraphlca||ythemaxlmumamountofsa|t
thateachtankevercontalns.
Forana|tematlvelnvestlgatlon, supposethatthe system oftanks lsc|osed
as ln Flg. 5. 4. 5, so that tank I recelves as lnHow the outHow fromtank(rather
thanfreshwater) . Thenthenrstequatlonshou|dberep|acedwlthx -/
-
x
-
- /, x,
Now show that, ln thl s ./osea system, as t -- +> the sa|t orlglna||y lntank I
dlstrlbutes ltse|fwlth constant denslty throughout the varlous tanks. A p|ot |lke
Flg. 5. 4. 6shou|dmake thls falr|yobvlous.
Inthl ssectlonweapp|ythematrlx methods ofSectlons 5. 3 and5. 4tolnvestlgate
theoscl||atlonsoftyplca| mass-and-sprlngsystemshavlngtwoormoredegreesof
freedom. Ourexamp|esarechosentol||ustratephenomenathataregenera||ychar-
acterlstlcofcomp|exmechanlca|systems.
Flgure5. 5 . I showsthreemassesconnectedto eachotherandtotwowa||sby
the four lndlcated sprlngs. We assume that the masses s|lde wlthoutfrlctlon and
thateachsprlngobeysHooke` s|awltsextenslonorcompresslonxandforce iof
reactlon are re|ated by the formu|a i = -/x If the rlghtwarddlsp|acementsx, .
x
:
.andx

ofthethreemasses (from thelrrespectlveequl|lbrlumposltlons)area||


posltlve,then

Thenrstsprlngls stretchedthedlstancex, ,

Thesecondsprlngls stretchedthedlstancex
:
- x, ,

The thlrdsprlngls stretchedthedlstancex

- x
:
,

Thefourthsprlnglscompressedthedlstancex

Therefore,app|lcatlonofNewton` s|aw i= -.tothethreemasses(aslnExamp|e


| ofSectlon5 . I ) yle|dsthelrequatlonsofmotlon.
-
:
x
- -/
:
,x
:
- x, +/

,x

- x
:
.
-

x= -/

,x

- x
:
- /
-
x

( I )
A|thoughweassumedl nwrltlngtheseequatlonsthatthedlsp|acementsofthemasses
are a|| posltlve, they actua||y fo||ow slml|ar|y from Hooke' s and Newton` s |aws,
whatevertheslgnsofthesedlsp|acements.
Intermsofthe....= x, x
:
x

the....
-,
.= -
:

(2)
t-...,...-...-....-.,--....-+..-...........-..-..., |.,....+-....,.-...--..-
.-.--.....,,.......-...-.,--....-...,-,.....-+--,.---..-,,..-.-.
382 Chapter 5 Li near Systems of Di fferenti al Equati ons

.

' '

. .
FIGURE 5.5.2. A system of
spring-coupled masses.
andthe..
(3)
thesystemln, i takesthematrlxform
.

=. (4)
ThenotatlonlnEqs. , | , through(4)genera|lzeslnanatura|waytothesysten
ofsprlng-coup|edmassesshownlnFlg. 5. 5. 2. Weneedon|ywrlte
and
.
-,/,+/
:
/
:
/
:
-,/
:
+/

+
+
+
+
+
+
+
forthemassandstlnessmatrlceslnEq. (4) .
+
+
+
+
-,/
-
-

+/
-
/
-
/
-
,/
-
+/
-

,;
(6)
Thedlagona| matrlx .l sobvlously nonslngular, togetlts lnverse.

we
needon|yrep|aceeachdlagona|elementwlth lts reclprocal. Hencemu|tlp|lcatlon
ofeach slde lnEq. (4)by.

yle|dsthehomogeneous.

(7)
where.

. Therelsawldevarletyof)/./o/essmechanlcal systemsfor
whlch a dlsp|acement or posltlon vector a nonslngular mass matrlx , and a
stlffnessmatrlx.satlsfylngEq. (4) canbedenned.
Solution of Second-Order Systems
To seekaso|utlonof Eq.(7) , wesubstltute(aslnSectlon5. 4foranrst-ordersystem)
atrla|solutlonoftheform
(S)
where vlsaconstantvector. Then

a
:
ve . so substltutlon ofEq. (S)ln (7)
glves
. Second-Order Systems and |ecOani cal Appl i cati ons 383
whlchlmp|lesthat
Av = a
:
v ,9)
Thereforex(t) = year lsaso|utlonotx Axlfaadoalylta
2
= A, aaelgeavalce
ofthematrlxA,aadvlsaa assoclatedelgeavector
Ifx = Axmode|samechanlca|system, thea |tlstyp|calthattheelgeavalces
ofAarenegative realacmbers lf
a
2
= A = u *
thena = Joi Iathlscasetheso|ctlonglvenbyLq. (S)l s
x(t ) = ve

= v(cosot+ i sl aot )
Therea| andlmaglnary parts
x, (t)= v cos ot and x
2
,t : = v sla ot ( l 0)
ofx(t) arethenllnearlylndependeatreal- valued so|ut| oasofthesystem Thlsaaal-
yslsleadstothefol l owlngtheorem
THEOREM 1 Second- Order Homogeneous Linear Systems
IfthematrxAhasdlstlnctnegatlveelgenva|ues-o, -o, . . . ,-owlth
assoclated rea|] elgenvectors Vj , v , . . . ,v, ,thenagenera|so|utlonof
lsglvenby
x = Ax
,
x(t) = )ai coso, t+bi sln o t ) v
i =|
, | |
wlth a and b arbltrary constants In the specla| case of a nonrepeated zero
elgenva|ue

wlthassoclatedelgenvectorv,,
x,(t)= (a,+b,t ) v, , i z
ls thecorrespondlngpartofthegenera| so|utlon.
Remark: Thenonzerovectorv,lsaaelgeavectorcorrespoad|agto

= 0
provldedthatAv, = 0. ltx(t) (ao + /,t) v, , thea
x = 0 v = ,a,+ b,t ) 0 = (ao + b,t )

(Av,) = Ax,
thcsverlfylagtheformlaLq. , i z:
384 Chapter 5 Li near Systems of Di fferenti al Equati ons
cXump| e 1


La| || e;|ames|t|eas
FIGUR 5.5.3. The mass-and
spring system of Example
.. . .. .. .
Ceas|ae::aemassaaas:|a,sys:emw|:a= zsaewa|at|, se.aase:ae:e
|sae:a|:as:|a,.eaae.:ea:ea:|,a:aaaawaii,wese:/

=O. iim = z.m2 = i .


/, = i cc.aaa/

= c. :aea:aeeaa:|eaMx" = Kx is
, i
wa|.a:eaa.es:ex" = Ax w|:a
1ae.aa:a.:e:|s:|.eaa:|eaeiA |s
,-:- ,-c- - c

z=
2
+i z+zcc
= ,+z ,+i cc=c.
seA aas:aeae,a:|vee|,eavaiaes, = -z aaa2 = -i cc sy1aee:em i .
:aesys:em| a, i :ae:eie:eaasseia:|easw|:a .|:.aia:[i:eaea.|esuj = aaa
u2 = i c
CASE 1 : , = -z 1aee|,eave.:e:eaa:|ea(A - U)v =0 |s

-c z

u c
c -z / c

seaae|,eave.:e:asse.|a:eaw|:a, = -z|sV, = i z
CASE 2: 2 = -i cc 1aee|,eave.:e:eaa:|ea(A - U)v =0 |s
seaae|,eave.:e:asse.|a:eaw|:a2 = -i cc|sV2 = i -i ]

sy, i i |:ieiiews:aa:a,eae:aiseia:|eaei:aesys:emia, i |s,|vea|y


x(t) = ,u,.es+/,s|a ) v
. +,a

.esl Ot +/

s|al Ot)v2 . , i :
As |a:aea|s.ass|eaeisamie eise.:|ea z. :ae:we :e:ms ea:ae:i,a:ia
, i ::e:esea:.ei:aemassaaas:|a,sys:em 1aeyaes.:|oe
:aeays|.aisys:em s:we.....a:|:s:we,.i:.aia:[...
..u, = aaau2 = i c1aeaa:a:aimeae
,w|:a c, = ,u+ / , .es a, =uj }cj , aaas|a a, =/, }c, )aas:aes.aia:.emeaea:
eaa:|eas
XI (t) = cj .es ,t - a, ) ,
X2 (t) = zc, .es ,t - a,
, i
cXump| CZ
FIGURE 5.5.6. The three
railway cars of Example
5. 5 Second-Order Systems and Mechani cal Appl i cati ons 385
aaa:ae:eie:e aes.:||es ai:ee es.|iia:|ea|awa|.a:ae:wemassesmeve |asya
.a:eay|a:aesamea|:e.:|eaaaaw|:a:aesamei:eaea.y., = 5, |a:w|:a:ae
ami|:aaeeime:|eaei

:w|.e:aa:ei, ,seet|,5. 5. 4). 1aeaa:a:aimeae


aas:aes.aia:.emeaea:eaa:|eas
x,,= .

.es , l c- a

) ,
x

,)= -.

.es , l c- a

) ,
, l :)
aaa:ae:eie:eaes.:||es ai:eees.|iia:|ea|awa|.a:ae:wemassesmeve|asya
.a:eay|aees|:e a|:e.:|eas w|:a :aesamei:eaea.yo

= lcaaaw|:aeaai
ami|:aaeseies.|iia:|ea,seet|, 5. 5. 5).
0
x X2(t)
a
/
FIGURE 5.5.4. Oscillations in
the same direction with frequency
WI = the amplitude of motion
of mass i s twice that of mass
_____ _.___
|
. \ . 1 . V . t t t t t . 1

X2(t)
0 a/l R 1a/l
FIGURE 5.5.5. Oscillations in
opposite directions with frequency
l = l O; the amplitudes of motion
of the two masses are the same.
t|,a:e5. 5. 6saews:a:ee:a|iway.a:s.eaae.:ea|y|ane:s:|a,s:aa::ea.:waea
.em:essea,|a:a|sea,a,e|as:eaaeis::e:.a|a, w|:a = 3, /

= /

= /. aaa
/, =/
-
=c|as ,z:a:ea,a(4), we,e::aesys:em

wa|.a|sea|vaiea::e
w|:a
c

c
+ -/
+ = /

c
., -cj
= c -z.

/
-z/
/
c

-.

/
,/ = i . z. ., =

,

-/
, l )
, l s)
, l )
iiweassameia::ae::aa:, =

,se:aa:cj =.

,:aeaa|:|ei.ema:a:|ea,ives
,z
386 Chapter 5 Li near Systems of Di fferenti al Equati ons
ie::ae.aa:a.:e:|s:|.eaa:|eaei:ae.eeu.|ea:ma::|s|a , l s) uea.e:ae
ma::|saase|,eavaiaes
,z| a)
.e::eseaa|a,:e:aeaa:a:aii:eaea.|es
,zi o)
ei:aeays|.aisys:em
te:aaame:|.aiesamie,saese:aa::aea:s:aaa:a|:a:a|iway.a:swe|,alz
:easea.a,:aa::aem|aaie.a:we|,ass:eas, aaa:aa::aes:|a,.eas:aa:|s/i
:eas}i:, | e. , / 3000i|}i: 1aea,as|a,saa|:sw|:amassmeasa:ea|asia,s,a
we|,a:eizeaaasaasamasseil sia,),weaave
aaa
3000
.,
750

uea.e:ae.eeia.|ea:ma::|s|s

3000
= :
500

- l z

,zz)
aaa:aee|,eavaiaei:eaea.ya|:s ,|vea|y,zl a)aaa,zl |)a:e

0, o, * 0,

o,= z,aaa-l :,
CASE 1 : 0, o, c. 1aee|,eave.:e:eaa:|ea 0 |s

-l z

se |: |s .iea::aa: v, l l l|s aa e|,eave.:e:asse.|a:ea w|:a

O.
A..e:a|a,:e1aee:eml , :ae.e::eseaa|a,a::eia,eae:aiseia:|eaei

is
CASE 2:

= z 1aee|,eave.:e:eaa:|ea 0 | s

0
+ :
~s
+
se|:|s.iea::aa:v

l 0 - i ]

|saae|,eave.:e:asse.|a:eaw|:a

A..e:a|a,:e1aee:eml , :ae.e::eseaa|a,a::eia,eae:aiseia:|eaei

is
5. 5 Second-Order Systems and Mechani cal Appl i cati ons 387
CASE J: x

= -l :,o

=: 1aee|,eave.:e:eaa:|ea- xi)v 0|s


l z
l :i)v
se|:|s.iea::aa:v

= l - l |saae|,eave.:e:asse.|a:eaw|:ax

* l :
A..e:a|a,: e1aee:eml , :ae.e::eseaa|a,a::eia,eae:aiseia:|eaei

=is
1ae,eae:aiseia:|ea= x,x

ei

=|s:ae:eie:e,iveaoy
1eae:e:m|aeaa::|.aia:seia:|ea,ie:assaese:aa::aeiei:mes:.a:|smevia,:a
:ae:|,a:w|:aveie.|:y t aaaa::|me = cs::||es:aee:ae::we.a:s,wai.aa:e
:e,e:ae:|a:a::es: 1ae.e::eseaa|a,|a|:|ai.eaa|:|easa:e
x, ,c)=x

,c)=x

,c)= c.
x,c) = t, x ,c)=x ,c)=c
1aeasa|s:|:a:|eaei,z+a)|a,z),|ves:aes.aia:eaa:|eas
a, a

=c,
a, a

= c,
a, - a

c,
,z:a)
,z:o)
wa|.a:eaa|iyy|eiaa, = a

= a

= c uea.e:aees|:|eaiaa.:|easei:ae:a:ee
.a:sa:e
x, ,)= /, /

s|a z/

s|a :,
x

,)=/, - /

s|a :,
x

,)= /, - /

s|az/

s|a + ,
aaa:ae|:veie.|:yiaa.:|easa:e
x ,)= /, z/

.es z+/

.es + .
x ,)=/, - l z/

.es : ,
x ,)=/, - z/

.esz:/

.es:
sa|s:|:a:|eaei,z+|)|a,z:),|ves:aeeaa:|eas
/, z/

:/

= t,
/, l z/

= c,
/, - z/

:/

= c
,z)
,z:)
388 Chapter 5 Li near Systems of Di fferenti al Equati ons
:aa::eaa|iyy|eia/, ,,/

= ,,aaa/

_, t|aaiiy,:aees|:ieaiaa.:iaas
|a,z)a:e
x, ,)= _c, , l zs sia zs|a : ) ,
- 3sia + ) ,
x

,) _c, , l z - s s|a z+ s|a : )


,z
sa::aeseeaa:|easaeiaeaiyseiea,as:ae:we|ane:s:|a,s:emaia.em
:essea,:aa:|s, wa|ie|e:a
x

x, * c aaa x

- x

* +
1ea|s.eve:waa::a|s|mi|esa|ea: , we.ema:e
aaa,s|m|ia:iy,
x

,)- x,,) _c, ,-ss|az- +sia +)


-_c, ,ssiazss|az.esz)
= - c, ,s|az) , l .esz)
x

,)- x

,)= - c, ,s|a z) , l .es z )


i:ieiiews:aa:x

- x, * caaax

- x

* caa:|i r } z l ,se.eaas),a:
wai.a:|me:aeeaa:|eas|a,z:)aaa,z),ive:aevaiaes
we.ea.iaae:aa::ae:a:ee:a|iway.a:s:ema|aea,a,eaaaameia,:e:ae:i,a:
aa:|i aisea,a,emea:e..a:s a::|me r }z 1ae:eaue:, .a:s l aaaz:ema|aa:
:es:, , ,wa|ie.a:3 .ea:|aaes:e :ae:i,a:w|:aseea, ii,ie:ias:aa.e,, :s
iee:e:se.eaa,a|ea:33 m|iese:aea:), :aea:ae:a:ee.a:s::aveiaais:aa.eai
9r zs z,i:)aa:|a,:ae|:l se.eaaseiea,a,emea:,aaa
x,,)=x

,)= 9r, x

,)= :s- l - ,z)


ie: >r }zt|,a:e |iias::a:es:ae|eie:eaaaai:e:si:aa:|eas,aaati, s
saews:ae,:aasei:aeiaa.:|easx,, ) , x

,) ,aaax

,)|as,z)aaa,z )
Forced Oscillations and Resonance
saeseaew:aa::ae/ :amassei:aemassaaas:ia,sys:emiati, zissao]e.:
:eaa es:e:aaiie:.e F, ,/ i . z, , ; |aaaa|:|ea:e:aeie:.esese::ea|y:ae
s:|a,sa::a.aea:e|: 1aea:aeaeme,eaeeaseaa:|ea.

= .is:eia.eawi:a
:aeaeaaeme,eaeeaseaa:|ea
. . ,zs)
. SCCCOOOC SylCOOOO|CCOOO| CO' /' | COl| CO 389

_
:
50
25
:
CarJ `
. v
conunt|cs
Cars l and 2
' _

0. 5 | . 0 | . 5 2. 0 2. 5
'
FIGURE 5.5.7. (a) Before; (b) after. FIGURE 5.5.8. Position
functi ons of the three rai lway cars
of Example
cXump| e
FIGURE 5.5.9. The forced
mass-and-spri ng system of
Example
a-:- F, F2 FIl . :a- external force vector |e: :a- ,:-m Ma|
:.| . .+:.ea |, ' ,.-|!
x" Ax + f (29)
a-:- f . :a- -s:-:a+| |e:.- -.:e: per unit mass. w- +:- --..+||, . a:-:-:-! .a
:a- .+- e| + periodic exteral force
f, .e . ,1I
,a-:- . + .ea:+a: -.:e: v- :a-a +a:...+:- + -:.e!.. +::..a| +: e|a:.ea
x_ , ) e .e . ,1 1
.:a :a- |aea -s:-:a+| |:-a-a., . +a! . :a + .e-u..-a: -.:e: c ,-: :e |- !-:-:
m.a-! s-.+a- x = .
2
c .e . . a|:.:a:.ea e| ,1I +a! ,1 | .a (29), |e||e-!
|, .+a.-||+:.ea e| :a- .emmea |+.:e: .e . . ,.- :a- | .a-+: ,:-m
(12
:e |- e|-! |e: c.
O|-:- :a+: :a- m+::. s A + .
2
. aea.a,a|+:-.a a..a .+- ,1: .+a
|- e| -! |e: e-aa|- .` A, +a -.,-a+| a- e| A. 1aa + -:. e!.. +::..a|+:
e|a:.ea e| :a- |e:m .a (3 1 ) -s.: :e.!-! :a+: :a- -s:-:a+| |e:..a, |:-a-a.,
!e- not -a+| ea- e| :a- a+:a:+| |:-a-a..- ., W2 , . . . , Wn e| :a- ,:-m 1a-
.+- .a a..a . is + a+:a:+| |:-a-a., .e::-ea! :e :a- a-aem-aea e| resonance
!..a-! .a s-.:.ea 2. 6.
sae- :a+: : a- -.ea! m+ .a s+m|- ; . a|]-.:-! :e :a- -s:-:a+| -:.e!..
|e:.- I .e . 1a-a . :a m, 2, m2 ; . k, ; Ic. k2 c. +a! I .a
t., ^. (29) :+|- :a- |e:m

: 2

I
x
c c
x +
c
.e . .
+a! :a- a|:.:a:.ea x c .e . |-+! :e :a- -a+:.ea
.` : 2 c
c .` I
c
c
, 11
,1:
390 Chapter Li near Systems of Di fferenti al Equati ons

`
orccdIrcgucncy
FIGURE 5. 5. 10. Frequency
amplitude plot for Example
ie::ae.eeia.|ea:ve.:e:c = ., .

1a|y:em|:e+a|iyeiveaie:
l zc
.,
- --

,o

- z) ,o

- l cc)
.
c,.

- )
.
:
= -
,o

z) ,.

- l cc)
,)
te:|as:aa.e,|i:aees:e:aaia+:eai:eaea.y|.

= c,:aea,)y|eia
., -l , .
:
= -l 1ae:eai:|a,ie:.eae:|ea|.e.|ii+:|ea|saes.:|oeaoy
x, ,)= - .e o . x

,)= - .e o
1aas:ae:wemasseses.|iia:e|asya.a:eayw|:aea+i+mi|:aae+aa|a:ae+me
a|:e.:|ea
ii :aees:e:a+isa+:eai:eaea.y|.

= l z, :aea, ) y|e|a., . .

=
-l 1ae:esai:|a,ie:.eae:|ea|.e.|ii+:|ea|ae.:|oeaoy
x, ,)= .es o . x

, ) = - .e . .
aaaaew:ae:wem+ssees.|ii+:e|aya.a:eay|aee|:ea|:e.:|eas.oa:w|:a:ae
ami|:aaeeime:|eaei

:w|.e:a+:ei,
i:|s ev|aea:i:em:aeaeaem|a+:e:|a,):a+:., +aa.
:
+:e+.a +
o a:ea.aese|:ae:ei:ae:wea+:a:+|i:eaea.|es., = +aa., = l c,ieaaa
|asamiel ) t|,a:e l csaews+ie:ei:aeam||:aae, ccei:aeie:.ea
e:|ea|.se|a:|eax,)= c .eoas+iaa.:|eaei:aeie:.eai:eaea.y. 1aee+|s
+:o

= aaao

= l cesa|||:v|a+|iy:aeaeaemeaeaei:eea+a.e
Periodic and Transient Solutions
i:ieiiewsi:em1aee:em:eise.:|ea :aa:+a::|.aia:eia:|eaei:aeie:.ea
ys:em
x= Axg .es. ,:)
w||i|eei:aeie:m
,)
wae:ex
,
, ) |saa::|.a|a:seia:|eaei:aeaeaaeme,eaeeasy:em+aax ,)|+
seia:|eaei:ae.e::eeaa|a,aeme,eaeeays:em i:|s:y|.+iie::aeeiie.:ei
i:|.:|eaai:es|s:aa.e|ame.aaa|.aiy:em:ea+mea::ae.emiemea:+:yiaa.:|ea
seia:|eax , ) , e:aa:
x ,)-0 as - ,s)
Hea.ex ,)|+transient solution :a+:aeeaaea| yea:ae|a|:|+|.eaa|:|ea, |:
a|es ea:w|:a :|me, ie+v|a,:aesteady periodic solution x
,
,):esai:|a,i:em:ae
es:e:a+ia:|v|a,ie:.e
x,)-x
,
, ) + -

,)
Asa:a.:|.aima::e:,eve:yays|.+isys:em|a.iaaei:|.:|eaa|:es|:+a.e,aeweve:
smaii):aa:aamsea:::aa|ea:aia:|aa|a:a|sm+aae:
5. 5 Second-Order Systems and Mechani cal Appl i cati ons 391
Problems
.1 ....7 ............,.....
......5.5. 1 1 ...,....
,

..

......................,...
............,...........
...........
FIGURE 5. 5. 11. The mass-and-spring
system for Problems through
1.

. .

(no walls)
2.

. .

. .

. .


4.

. .

5.

..

. .

.
7.

. . .

.
..8 ....10 ............,.....
.........,.
.,....

.......
....

............

.,
..................
.....,,........., . ,...
8. The mass-and-spring system of Problem with
cos

=
9. The mass-and-spring system of Problem with

..
10. The mass-and-spring system of Problem with
..

..
11. Consider a mass-and-spring system containing two
masses and

whose displacement func


tions . and . satisfy the diferential equations
... .
. . .
(a) Describe the two fundamental modes of free oscilla
tion of the system. (b) Assume that the two masses start
in motion with the initial conditions
and
. .
. .
and are acted on by the same force,

cos Describe the resulting motion as a superpo


sition of oscillations at three diferent frequencies.
..12 ...13, ......,....
........5. 5. 1, ..............
.,.............,.......
. .

.,.....,........
..cos ..

cos ..

cos .
12.
13.
14.
15.
16.

. .

. .


.One eigenvalue is .
In the system of Fig. assume that , .
.

and in mks units, and that .Then


fnd

so that in the resulting steady periodic oscillations,


the mass ml will remain at rest( ! ) . Thus the effect of the
second mass-and-spring pair will be to neutralize the ef
fect of the force on the frst mass. This is an example of
a ......,It has an electrical analogy that some
cable companies use to prevent your reception of certain
cable channels.
FIGURE 5.5.12. The mechanical
system of Problem 14.
Suppose that

. .

, and . (all in mks units) in the forced


mass-and-spring system of Fig. Find the solution of
the system .
,
.F that satisfes the initial condi
tions .. .
Figure shows two railway cars with a bufer spring.
We want to investigate the transfer of momentum that oc
curs after car with initial velocity 0impacts car at rest.
The analog of Eq. in the text is
.

X
with . for Show that the eigenvalues of
the coefcient matrix A are and

with associated eigenvectors v

and .

|
X2' (O) =0
FIGURE 5.5.13. The two railway
cars of Problems 1 6 through
17. If the two cars of Problem 1 6 both weigh 16 tons ( so that

(slugs and .ton/ft (that is,


lb/f), show that the cars separate afer J/seconds, and
that .and . 0 thereafer. Thus the original
momentum of car is completely transferred to car
392 Chapter 5 Li near Systems of Di fferenti al Equati ons
18. If cars and 2 weigh and tons, respectively, and
.= lb/f, show that the two cars separate afer /
seconds, and that
.

| c, and . = + c,
thereafer. Thus the two cars rebound in opposite direc
tions.
19. If cars and 2 weigh .and tons, respectively, and
.= lb/f, show that the cars separate afer J/2 sec
onds, and that
.

= +| c, and . = + c,
thereafter. Thus both cars continue in the original direc
tion of motion, but with diferent velocities.
.20 ....23 ...........
.................,.,.....
.,...............5. 5. 6 ...........
.,2. ......... = ..XI =
X = X_ = .............
c,= .......................
= /2 ......,....,
....................
........ . .... ....
/2. ...,.............,
2) ........................
..,,,....
20. .= c,..= .= c,
21. .= c,.. = . = c,
22. .= c,..= c,..= c,
23. .= c,.. = c,.. = c,
24. In the three-railway-car system of Fig. suppose that
cars and 3 each weigh 32 tons, that car 2 weighs tons,
and that each spring constant is .tons/ft. If . = c,
and .= .= show that the two springs are
compressed until = /2 and that
. =

c, and . = . = + c,
thereafter. Thus car rebounds, but cars 2 and 3 continue
with the same velocity.
The To-Axle Automobile
...,4 .2. 6 .........
...................
....................
.,..........,.........5. 5. 14 ,
.......,.................
........................
L = LI +L
2
. ..........
............L I ......
.............,....,...... ..
.........

.,........
.......,.......
..,.. ..........,.
................ ....
....................
.,....
25.
26.
La|||ot|am
es|t|ea
FIGUR 5.5.14. Model of the
two-axle automobile.
.
Suppose that = slugs (the car weighs .lb), LI =
f, L
2
= ft (it' s a rear-engine car), .= .

=
lb/f, and = f l b s
2

Then the equations in .
take the form
.+
,
ooox =
.+ =
(a) Find the two natural frequencies WI and W
2
of the car.
(b) Now suppose that the car is driven at a speed of cfeet
per second along a washboard surface shaped like a sine
curve with a wavelength of .f. The result is a periodic
force on the car with frequency W = -c}
,
o = -c}o
Resonance occurs when with W = WI or W = W
2
. Find the
corresponding two critical speeds of the car (in feet per
second and in miles per hour) .
Suppose that . = .

= .and L
I = L
2
= | L in
Fig. .(the symmetric situation). Then show that ev
ery free oscillation is a combination of a vertical oscilla
tion with frequency
and an angular oscillation with frequency
..27 ....29, .....5.5. 14 ....
..........,.......,....
,.... (u) ........,......
.. (b) ...................
........................
...,..
27. = = LI = L
2
= .= .

=
28. = = LI = L
2
= . .= .

=
29. = = L
I
= L
2
= .= .

=
5. 6 Mul ti pl e Ei genval ue Sol uti ons 393
luItipl Eialue Solutions
cXump| e 1
iase.:|ea :we+w:aa:|i:ae m+::|sna+a/s/.,:e+|a:.am|es)
e|,eava|aes, ,x

x
-
w|:a:ee.:|ve+e.|a:eae|,eave.:e:v, ,v

v
-

:aeaa,eae:aieia:|eaei:aey:em
|s,|vea|y
as
~ = ns
a
, l )
,z)
w|:a+:||::a:y.ea:aa:., ,.

.
-
ia:a|se.:|eawea|s.as:aes|:a+:|aaaea
:ae.aa:a.:e:|:|.eaa:|ea
n- =c
aeesoaavea|s:|a.::ee:s, aaa:aasa+s+:iea:eae:ee+:ea:ee:
Aae|,eav+iae|seimUltiplicity / |i|:|s+/ieia:ee:ei ,: te:e+.a
e|,eavaiae,:aee|,eave.:e:eaa:|ea
,n- v=0 ,:)
aasa:ieas:eaeaeaze:eeia:|eav,se:ae:e|+:ie+s:eaee|,eave.:a:+a.|+:ea
w|:a sa:aae|,eavaiaeeimai:|i|.|:y/ > l may aave)ee:aaa/||ae+:i,
|aaeeaaea:asse.|a:eae|,eave.:e:s.ia:a|.+ewea:eaaa|ie:eaaaa.am|e:e
se: ei i|aea:iy |aaeeaaea:e|,eave.:e:s ain, a aeeaea:eie:m:ae,eae:+|
seia:|ea|a,z)
Le:a.aii+ae|,eav+iaeeimai:|i|.|:y/ complete |i|:a+/ i|ae+:i, |a
aeeaaea:ase.|+:ea e|,eave.:e: iieve:ye|,eavaiaeei:aem+::|sn|.am
ie:e,:aea-|e.aasee|,eave.:e:sae.|a:eaw|:aa|ne:ea:e|,eav+iaesa:ei|ae+:|,
|aaeeaaea:-|:ieiiews:aa:naeesaavea.emie:ee:eii|aea:iy|aaeeaaea:
e|,eave.:e:sv, , v

v
-
asse.|a:ea w|:a :aee|,eav+iae , ,


-
,e+.a
:ee+:eaw|:a |: mai:|i|.|:y) ia:a| .+e +,eae:ai seia:|eaeis = ns|:||i
,|vea|y:aeaaai.em||aa:|ea|a,z)

t|aaa,eae:aieia:|eaei:aesy:em
:
-l
:
,
Sol ution 1ae.aa:a.:e:|s:|.eaa:|eaei:ae.eeia.|ea:ma::|sn|a,)|
n- =
-
-:
:
:
-i -
:
+
+
-
= ,- ,- ) , -l - z:[
= ,- ) , l - s

)
=,- ) ,- )

= +
1aasnaa:aea|s:|a.:e|,eavaiae
,
= aaa:ae:eea:eae|,eavaiae

= ei
mai:|i|.|:y/=z
394 Chapter 5 Li near Systems of Di fferenti al Equati ons
CASE 1:
, = 1aee|,eave.:e:ea+:|ea,n-: =0, ae:ev . / c

,
|s
(A - iv

|
+.aei:aea:s::waea+:|eas,:.+:/ = c+aa-:.- :/ = c,y|eia/ = -.
1aea:ae:a|:aea+:|ea:eaa.e:ez.- z.=c.e:a+:c . 1ae.ae|.e. =
:aeay|eias:aee|,eave.:e:
v, = i -l i
+se.|+:eaw|:a:aee|,eav+iae, =5.
CASE 2:
:
= New:aee|,eave.:e:ea+:|ea|
: :
,n- : v= : -:
: :
e:aeaeaze:eve.:e:v= . / . |+ae|,eave.:e:|i+aaeai,|i
:.+ :/=c, ,::
:a+:|, /= -. 1aei+.::a+:,::aeeae:|avaivec me+as:a+:.|+:o|::+:y,
ao]e.::e:ae.eaa|:|eav=0. ii.= l , :aeawem+y.aeee.=/=c,:a|,|ve
:aee|,eave.:e:
v
:
= c c l
+sse.|+:ea w|:a
:
= ii. = c, :aea wema:.aeee. :e oeaeaze:e te:
|as:+a.e,|i.=z,:e+ve|ai:+.:|eas),:aea/=-. se
v

= z - c
|s+e.eaai|ae+:iy|aaeeaaea:e|,eave.:e:+se.|+:eaw|:a:aemai:|||.|:yze|,ea
v+iae
:
=
1aawea+veieaaa+.emie:ee:v
,
.
:
.v

ei:a:eee|,eave.:e:+e.|+:ea
w|:a:aee|,eav+iaes, , 1ae.e::eeaa|a,,eae:+iseia:|eaei(5)|
w|:as.+i+:.emeaea:iaa.:|aa,|veaoy
x,,:= ., e

+ z.

.
x
:
,=-.,
e

- .

.
x

,:= ., e

+.
:
e

,)

cXump| eZ
. |O' l| ' C L| _COvO' OCSC' Ol| CO 395
Remark: Oa: .ae..- .a s+m|- ei :a- :e -.,-a-.:e:
V
2

+a! V
3
: I [
+e..+:-! .:a :a- :--+:-! -.,-a+|a- A
2
|-+: .emm-a: 1a- i+.: :a+:
- ie: +ay -.,-a-.:e: +e.i+:-! .:a A
2
m-+a :a+: +a, a.a -.,-a-.:e:
.+a |- :.::-a +
+a! :aa . + |.a-+: .em|.a+:.ea e| V
2
+a! V
3
. 1a-:-ie:-. ,.-a +a! c ae: |e:a
z-:e. - .ea|! .aee- v :+:a-: :a+a V
3
+ ea: :a.:! -.,-a-.:e:. +a! :a- a- ,-a-:+|
e|a:.ea
CI VI e5t + c
2
v
2
e
3
t + C
3
ve
3
t
ea|! |- -a.+|-a: :e :a- ea- .a (7) . 1aa - a--! ae: e::, +|ea: m+|.a, :a-
:.,a: .ae..- ei ia!--a!-a: -.,-a-.:e: +e..+:-! .:a + ma|:.|- -.,-a+| a-
Aay .ae..- .|| !e, - ,-a-:+||, m+|- :a- . m|-: ea- - .+a
Defective Eigenvalues
1a- ie||e.a, -s+m|- ae :a+:-aaie::aa+:-|,-ae: +|| ma|:.|- -.,-a+|a- +:-
.em|-:-
1a- m+::. s
a+ .a+:+.:-:. :i. -a+:.ea
n

A
7 - A
I A) (7 - A)
A
2
- 8A + l (A - 4)
2
O.
( 8)
1aa n a+ :a- .a,|- -.,-a+|a- A j 4 e| ma|:.|...:, 2 1a- -.,-a-.:e: -a+:.ea

, n 4I) v

:a-a +meaa: :e :a- -a.+|-a: .+|+: -a+:.ea


+ O.
H-a.- .i v . :e |- +a -.,-a-.:e: e| n 1a-:-|e:- +ay -.,-a
-.:e: +e..+:-! .:a A I 4 . + aeaz-:e ma|:.|- ei v
T
. 1aa :a-
ma|:.|...:, 2-.,-a+|a- A j 4 a+ ea|, ea- . a!--a!-a: -.,-a-.:e:. +a! a-a.-
i ia.em|-:-
396 Chapter 5 Li near Systems of Di fferenti al Equati ons
Aae|,eavaiaexeimai:|i|.|:y/> i|s.+iieadefective |i| :|ae: .emie:e
iixaaeaiyp < /i|ae+:iy|aaeeaaea:e|,eave.:e:,:aea:aeaamoe:
a = / - p ,)
eim|ss|a,e|,eave.:e:s|.aiiea:aedefect ei:aeaeie.:|vee|,eav+iaeA. 1aa
:aeaeie.:|vee|,eavaiaex, = :|as+mieza+mai:|i|.|:y/ = z+aaaeie.:
a= i .|e.aasewes+w:aa:|:aaseaiyp= i +e.|+:eae|,eave.:e:
ii:aee|,eav+iaeei:ae m+::|sA +:eae:+ii.emie:e,:aea:aee|,ea
vaiaeme:aeaasye:aes.:||eaw|ii:eaa.e)ee:aaa:aeaeeaeai|ae+:iy|aae
eaaea:seia:|easei:aesys:emx
'
=Ax. we:ae:eie:eaeea:ea|.eve:aew:eaaa
:aem|ss|a,seia:|eas.e::eseaa|a,:e+aeie.:|vee|,eav+iaeA eimai:|i|.|:y
/ > i
The Case of Multiplicity h Z
Le:as|e,|aw|:a:ae.ae/ = z, aaasaese:aa:wea+veieaaa,+s|as+mie
z):aa::a-:e|seaiyas|a,iee|,eave.:e:v,ase.|+:eaw|:a:aeaeie.:|vee|,eav+iae
x 1aeaa::a|se|a:wea+veieaaaeaiy:aes|a,ieeia:|ea
, i c
eix: = Ax. syaa+ie,yw|:a:ae.aeei+:ee+:ea.a+:+.:e:|:|.:ee:ie:+|a,ie
i|aea:a|ne:ea:|aiea+:|ea,se.:|eaz ), wem|,a:aee:eaaa+se.eaaeia:|ea
ei:aeie:m
, l l )
waeawesa|s:|:a:ex =

| ax' =Ax, we,e::aeeaa:|ea


sa:|e.aase:ae.eeu.|ea:ei|e:ae

+aae

ma:o+i+a.e,|:ieiiews:a+:

=0,
aaa aea.e:aa:X

, ~ 0. 1a|sme+a :a+:-.ea::+:y :e ea:aee-:ae y:em


x
'
= Ax aeesae:aaveao/././eia:|eaei:aeie:m+samea|a, i i
ia:eaaeis|miy,|v|a,aea:ae|aeaoea|aa, i i . ie:aes:eaa| :si|,a:iy
aaa:eia.ev
:
w|:av, v

1aaweesie:e:aee|o|i|:yei+e.eaaeia:|ea
ei:aeie:m
, i z)
wae:ev,aaav

a:eaeaze:e.eas:aa:ve.:e:waeawea|:|:a:ex =v, e

+v

|ax
'
=Ax, we,e::aeeaa:|ea
, i
weeaa:e.eeu.|ea:seie

aaae

ae:e,+aa:ae:eoyeo:+|a:ae:weea+:|ea
(A Jv, =0 , i ::
aaa
, l )
:aa::aeve.:e:sv,aaav

mas:s+:|siy|ae:ae:ie:, i z:a,|ve+seia:|eaei x=Ax.


5. 6 Mul ti pl e Ei genval ue Sol uti ons 397
Ne:e:aa:, l :)me:eiy.eaa:ms:a+:v, |+ae|,eave.:e:eiA +sa.|+:ea
w|:a:aee|,eavaiae 1aea , l ) +ys:a+::aeve.:e:v

+:|saes:aeea+:|aa
i:ieiiews:aa:,|ae:ae::eseives|mai:+aeeaiy:ae:weea+:|eas|a( 1 4) +aa, l ),
|:sau.es:eaaaaseia:|eav

ei:aes|a,ieea+:|ea(A - l)

0 sa.a:a+::ae
:esai:|a,ve.:e:v, (A - i)v

|o,eoi::anea::a+::a|s|aiw+,as||ie
|i:aeaeie.:|vee|,eavaiae eiA |eimai:|i|.|:yz Ceaseaea:iy,:ae:a.eaa:e
aes.:||ea|a:aeieiiew|a,ai,e:|:amaiwaysa..eeas|aaaa|a,:we|aaeeaaea:
seia:|easasse.|a:eaw|:asa.aaae|,eavaiae
ALGORI THM Defective Multi pl i city Z Eigenval ues
1. t|:s:aaaaaeaze:eseia:|eav

ei:aeeaa:|ea
sa.a:aa:
(A - i) v

=v,
|saeaze:e,aaa:ae:eie:e|saae|,eave.:e:v
,
asse.|a:eaw|:a
Z. 1aeaie:m:ae:we|aaeeaaea:seia:|eas
aaa
eix' =Ax .e::eseaa|a,:e
l
x

|
_ x.
, l :)
, l )
, l s)
, l )
(?0)
Sol ution iasamiezweieaaa:aa::ae.eeu.|ea:ma::|sA |a(?0) a+:aeaeie.:|ve
e|,eavaiae =4 eimai:|i|.|:yzwe:ae:eie:e|e,|a|y.+i.ai+:|a,
uea.e , l :)|s
v

0,
aaa:ae:eie:e|ssa:|saea|y.y.ae|.eeiv

ia:|a.|ie,| :.eaiaa+ea:a+:
(A - :i) v

|saeaze:e,asaes|:ea)ie:eme.ae|.eseiv

:aeaaaa:ia:e:ae:s ii
398 Chapter 5 Li near Systems of Di fferenti al Equati ons
J
0^

- I
-Z
- J
-+
FIGUR 5.6. 1. Direction feld
and solution curves for the linear
system .= x, - x, .
x= x, :x,of Example
we::yV2 = . c
t
weaaa:a+:
-
( A - 4I) v

= Y(
|aeaze:e,+aa:ae:eie:e|+ae|,eave.:e:+sse.|+:ea|:aA = 4. ,i:| - :|me
:aee|,eave.:e:ieaaa|as+mie2. ) 1ae:eie:e:ae:weseia:|eaei(20) ,|vea
oy, l s)+aa . +:e
, ;
-

s, = vi e =

e .
1ae:eai:|a,,eae:+ieia:|ea
s,;= ., s,,;.

,;
aa.+i+:.emeaea:iaa.:|ea
x,,;= , -.

- .,e
-

.
x

,;= ,.

., ;e
-

w|:a.

= 0 :aeeseia:|eaeaa:|ea :eaa.e :e:aeea+:|eax, , = -., e


-

.
x

,;= ., e
-

,wa|.a+:+me::|ze:aei|aex, = -x

|a:aex, x

i+aeo 1aee|a:
,x, , . x

,:aea:e.eae+iea,:a|i|ae+w+yi:em:aee:|,|a+ - :e:ae
ae::awe:|i., > +aa:e:aesea:ae+s:|i., < O. A|aa|.+:ea|at|, : l , e+.a
eia:|ea.a:vew|:a.

i 0 |s:+a,ea::e:aei|aex, = -x

+::aee:|,|a,:aee|a:
,x, ,; . x

, a:ea.ae :aee:|,|a+ - +aa+:e+.ae +iea,:ae


eia:|ea.a:ve+s- M
Generalized Eigenvectors
1aeve.:e:V2 |a , l :) |s+aes+mieeia,eae:+i|zeae|,eave.:e: iiA |+a
e|,eav+iaeei:aema::|s :aeaarank generalized eigenvector +e.|+:ea|:a
A |+ve.:e:v sa.a:a+:
(A - Alrv =0 oa: - AIr
-
I
v i O. (2 1 )
ii= . :aea:. |miyme+a:aa:v | +ae|,eave.:e:+se.|+:eaw|:aA ,:e.+ii|a
:ae.eavea:|ea:aa::aec:aewe:ei+saa:ema::|s|:ae|aea:|:ym+::|s) 1aa+
:+a|. ,eae:+i|zeae|,eave.:e:|+ae:a|a+:ye|,eae.:e:1aeve.:e:V2 |a, l :)|+
:aa|z,eae:ai|zeae|,eave.:e:,+aaae:+ae:a|a+:ye|,eave.:e:)
1aemai:|i|.|:yzme:aeaaes.:||eae+:i|e:oe|iaewa:eaaa|a,++|:, , .

|
ei,eae:+i|zeae|,eave.:e:,eaeei:+a|. +aaeaeei:+a|2, a.a:a+:(A -AI)V2
V , u|,ae:mai:|i|.|:yme:aeas|aveiveiea,e:.a+|aei,eae:+i|zeae|,eave.
:a:Alength /chain of generalized eigenvectors based on the eigenvector ,|
+e:, , .V2 , . . . , Vk } ei/,eae:+i|zeae|,eave.:e:a.a:a+:
- AI)Vk = Vk-
I
,
(A - AI) Vk -1 = Vk-2,
- AI) V2 = ,
(22)
cXump| e4
5. 6 Mul ti pl e Ei genval ue Sol uti ons 399
se.aase|saae:a|aa:ye|,eave.:e:, = O. 1ae:eie:e,| : ieiiewsi:em
,zz):aa:
,z)
ii

|saiea,:a.aa|aei,eae:ai|zeae|,eave.:e:sasse.|a:eaw|:a
:aemai:|iee|,eavaiaeei:aema::|s:aea|:|seasy:eve:|iy:aa::a:eei|aea:iy
|aaeeaaea:seia:|easei=a:e,|vea|y
,;=

= +

te:|as:aa.e,:aeeaa:|eas|a,zz),|ve
se
=

+
=

++

+
= +

+
1ae:eie:e,|a,z+)aees,|aaeea,aeaaeaseia:|eaei
,z:)
Ceaseaea:iy, |ae:ae::eaaaaieamai:|i|.|:y e|,eavaiae |:sau.es
:e aaaaiea,:a .aa|a

ei,eae:ai|zeae|,eavaiaesasse.|a:eaw|:a
Lee||a,a: ,z), wesee:aa:weaeeaeaiyaaaaseia:|eaei

=0
sa.a:aa::aeve.:e:s

= aaa =

a:e|e:aaeaze:e,ai:aea,a,aswew|iisee, :a|s|sae:aiwaysess||ie)
t|aa:a:eei|aea:iy|aaeeaaea:seia:|easei:aesys:em
,z)
Sol ution 1ae.aa:a.:e:|s:|.eaa:|eaei:ae.eeu.|ea:ma::|s|a,z)|s
z
= - -7
0
=
.
[ -7 - z

, - +,x) ,x) , +
=

= +

= 0,
400 Chapter 5 Li near Systems of Di fferenti al Equati ons
aaa:aasaas:aee|,eavaiae eimai:|i|.|:y 1aee|,eave.:e:eaa:|ea
0 ie:aae|,eave.:e: . |s

1ae:a|:a:ew.

,|ves-., :aea:aea:s::ew.

,|ves
. 1aas, :e w|:a|aa.eas:aa:mai:|ie,:aee|,eavaiae aaseaiy:ae
s|a,ieasse.|a:eae|,eave.:e:. . -. w|:a. =

aaase:aeaeie.:
ei |s
1eaiy:aeme:aeaaes.:||eaae:eie:::|iee|,eavaiaes,wea:s:.ai.aia:e



aaa

1aas.yaeaze:eve.:e:w|ii|ea seia:|eaei:aeeaa:|ea

O.
se,|aa|a,w|:a
ie:|as:aa.e,we.ai.aia:e

Ne:e:aa: |s:ae:ev|easiyieaaae|,eave.:e:w|:a. :a|sa,:eemea: