Chapter 2
Chapter 2
Chapter
Aorm
he
ependen
2. Qualitative Properties
quation of
Solutions and
though Oscillation Theory
endent
2.1. INTRODUGTION
In this chapter we
shall study the basic
0scillationof solutions of properties of self-adjoint
linear differential and the
entia
2 (B)
:
shall discuss some
basic
case sturm separation
Sturm.
equations
theorems, sturm comparison
of second order.
Also, we
theorem and their special
theorem which provides is after
Jacques Charles Francois
ndent
20001
nsbe 2.2. THE ADJOINT EQUATION
Definition. Consider the
this nth-order linear differential
operator
L, =a0(x)
d dn-1
+ aj(r) t...+ an
dx + an (x)
-1() ..(1)
val. dan dx
1
000)
and correspon ding nth-order differential equation
Ln y = 0
d"y
Or ao(1)
dx"
+ aj(*) d"-ly +...+ an -1 dx + an (x)=0
daxn-1
called the
adjoint equation tothe equation L, ()= 0.
ao()
d"y
+a(o)
y +...+ an - 1(a)+ a,(x)y =0 .(2)
dan dn-1 dx
isthe
equation
28 Qualitative Properties
-1 d"-1 [aj(*) y]+...
+(-)2
(-)7
dn ()
1o dyn
yl
d [an - 1(*) y]+ an (x) y =0
We have
dx and do(x)
on an interval asxsb and thas Putting these y
continuous (n - th
ay are all
a; have continuous derivative ao(x)
where 0 1 **,an -l, the coefficients We have
ao(0) *
on a ssband
O
(k=0, ...,n -
1) on a<xsb.
(2c+
1,
Case equation
The Second-Order order linear differential
the second ag(*) y+ y=0 which is the requi
Consider
ao(*)y'
a(*) +
function
of r and a(o) i
(x) are continuous (ii) Compar
a (x)and
a
in which ao(x) ÷ 0,
differentiable
on a sx b.
az(x)y =0
'
to (4) is with
The adjointequation
+ +
(-1)' [a(*)
yl We have
(-1) [ag(r)yl'
+ y]+ ag(*)y =0
+ do(*)y -la(x) y i(*)
y and
Or [ao(*) y d'o()y+o (x)
+ do(x)y + - (*)y -di (*) y+ ag(x) ysi6 Putting the
Or ao(r) y' a
ag(x)]y =0
+ (*)-a;(*)+
Or ao() y'+ [2a' o(x)
-a(1)] [a'o We have
w.r.t. 'x.
denote differentiation
where the primes (5)i:
of the adjoint equation
be noted that the adjointequation
Note :It should Or
equation (4) itself.
always the original
to each of the following
equations: which is the re
|
and
y=0 We have Oscillation
Theory
values in the
adjointequation
hderiNaive
We have
ao(*)y' +
[2a'o (x) -
aj(x)] y +[a"o(x) - di
(2x +1)y'+(4-x°)y+ + (x)
ag(c)] y =0
(0– 3 +1)y =0
Or (2r +1) y'+ (4-
and x')y+(1- 3r)y
a()i which is the required adjoint =0
equation.
(iii)
Comparing the given equation
xy+7xy + 8y =0
with ao(*) y"' +aj(*) +ag()y =
y ..(1)
We have
and do (x)=2c, a''o
ao(x) =x2 ,
aj(x)
0.
ag(*) y=
(x) 2, d'y (o)=7. =
Putting these values in
=0 ..5) the adjointequation
ao(*)y' [2a' o(*) a()]
+ y [a'o - + 0)- di(x) +ag()] y =0.
We have
ation (5)i xy'+ [4-7]y + [2-7+8] =0 y
or
xy'-3xy + 3y =0
uations ;
which is the required adjoint equation.
ar 2002, 96
...2)
0. P(u, v) = E
k=1 j=1
The notation u(R- ) denotes the (k- Dth derivatives of u
derivative of Vdn -
w.r.t. 'x, and in
have cor
Further,
for any
t
differentiating (3) w.r.t.
'*,we have
|Uk-)y-uk- 2)y' +.
V=(-1)* v()U + -D1. ...(4)
dx in
+(-1y- 2yyk
- +
va,-
,successively
2) (-
1
Pro
Apply the formula (4)
:
with U =u and V= vdo: Oa,...,
du
vL, [u]=vao d"u dn-lu +... + Uan - 1 dx On
dx"
+vay
We obtain dn - 1
+...
n- )
Y
vL, [u]=(-1)"(vao))u+ [u
-
(van) u (n
- 2) (vao
dx (n-Du
-'(va,)
+ (-1)"-1 u(vag) -1D]+(-1)" 2.5.
d -u7 - 3)(va)+...
[un- 2)(va) Le
dx
]+...+ (-1) (va,-1)!
-
+(-1)"- u(va,)n d
[u(van -1)] + (va,))
dx where
n
d (vag)-)
- k) +
E (-1)"-k(va,y(n
=u k= dx j=1 and ad
0
d n-1
T
dx j=1
1
d
+...+
dx j=1
Therefore,
d
vLn lu] -uL, lo]= E dxj=1
k
(-1)'-l ,(k-) (van--h)U-D
where
k=1
Hence,
n
(-1)/-1 (va,-)U - )
dx k =1 j=1
d
P(u, v)l.
dx
2,
for any two having nth
points x and 0f a sx
x derivatives on a asxsb.
b,
sxsb. Then,
auccessively [(vLn lu] -un dx= P(u, v)lz= x
lv]}
- P(u,v)l,=
Proof. We have by Lagrange's [Meerut2006;Gwalior
identity 2006]
d
vLn u] -uL, v] =- [P(u,v)].
dx ...(1)
On integrating (1)w.r.t. 'xbetween
thelimits x to x9, Wehave
vLn [u] -uin [vl} dx= (P(u, v)l,=x -[P(u,
v)lx=x
2.5.
LAGRANGE'SIDENTITY:THE SECOND-ORDER CASE
1
[Kolkata 2001]
(van-1) Let us consider the Lagrange's identity in the second-order
[P(u,v)]
)]+(va,) dx ...(1)
where operator d
(vag)U-D
L =ao(r) + a,(*) dx
+ag(x)
dy2
d
and adjoint operator L,=ao(x) + [2d'o(x) -a,()] dx
da2
+ [a'o (x) - d (*)+ ag(*)]
Therefore, the left hand side of Lagrange's identity (1) reduces as
vL, lu] -uLg [v]= v [agu''+ ajw'+ agu]
-u [aou' +{2do - a} +{a'o -i +ag} u]
=agvu'' + ajvu' + agvu-agu' 2d'o u -
..2)
'x.
where the primes denote differentiationsw.r.t.
d
is - P(u,v)],where
Now, the right hand side of Lagrange's identity (1) dx
2 k
P(u, v) =
k=1|j=1
1
=
j=1 2
(van)(U-)
+j=1
E (-1)'-'y2-D
+(-1)' u(va,)
-(-1) u(va) +(-1)° u (va,)
ao ()y''
+ a(x) y +a(*) y=0 Fr
is called self-adjoint is identical adjointequationwith its ...(2)
if it Jy=0.
(*) - (*) + az(*)] He
ao(1) y'+ [2a6(x) - a (*)] y' +a''o
II
Here, we observe that if
(*) =a'o (). write t
(x) do(x)then = ' we are in a
aj
identically (1). Hence,
As a result of this equation (2)
becomes below.
eguation in an other way as given
position to define
the self-adioint form if
eguation (1) is said to be self-adjoint
Definition. The differential
a (x) a'o(x) = form
equat
in a particular
(1) may be written
and in that case equation ...(3)
ag(x) y =0
ao(*) y'+ aj(x) y t
ther
derivative
where () ao(*) has a continuous second
a continuous first derivative
(ii) a (r)has
and ao(*) 0on a <x<b.
and (ii) ao(x) iscontinuous Or
that equation (1) be self-adjoin
and sufficient condition
Then, a necessary
equation isthat
d'o (x) = ()ona Sx<b.
a, th
Hence, equation
(1)is self-adjoint.
+ d'o v u] Sufficient Condition
...(4)
Let us suppose that equation (1)is
self-
f-adjoint.
identical Therefore,
(3)are equations (1)
and
Equating
the coefficients of y and y in (1) and (3), we have
2a o(x) -a(x) = a (x)
and
d''o (x)-(x) +ag(*) = a (*),
..(4)
tion
a''o (x) = i (*)
...(1)
which, on integrating, gives
do (x) =a()+ G, where cis an arbitrary constant. ...(6)
0. ...(2) From equations (4) and (6), constant cvanishes.
)
self-adjoint
Solution. (i) We know that the second order homogeneous linear differential
equation
...(3)
ao ()y" + a(*) + az()y =0 ...1)
cdjoint
0r
(i)Comparing
lao(x) y
[sin x y '
Y+ag(x) y =0
+ 2y =0.
given equation
with (1), we have
=x, a ()=3,ag()
=1
ao(r)
then
d'o (r) =32 =a(x). be written
as
So (i)can
equation.
:.
Given equation (ü)is self-adjoint
...(2)
°yl+y = 0.
Differential
Equ
(x)in the
34 (x) and a differential
ao(*), aj =0 equ Oualitative Properti
coefficients
'
aj(*)y' +ag (x) y
ao(x)y +
3. Ifthe
Theoremn
# 0 on
asx<h Then, equation (1) () y' - 2acy
ao(x)
on a sxSb and equation (ii) f(o)
y'+
sxsb
self-adjoint
are continuous
transformed
into the
equivalent
r(x) y+ p) y=0, a (ii) -tan
*y'
y
r(x) =e
a (*)
ao() )p)=
dr ag(*)r(*)
(iv)
Solution. (i
where aj (*) dx
1 ao (*) Comparing
Proof. In order
factor H(%)
by a suitable
throughout the equation
determined by
function H(r)
[ao(x) H()}
aj(x) H(*) =
ap(x) H (x) 'o + ()H() = aj(x)H(x).
L.e.,
Dividing (4) by ao
()H(*), we get
H (*) do(1) aj(*)
dx.
log H(r) =-log ao(x)+
ao(*) Or
aj (x)
dx This
ao(*)
So that H%)=
ao(*)
dx
W
r(*) =e (eol)
and pc) = a(3)
wegetequivalent
self-adjoint
a() M
equation
Ir(x) yI +t px)y
Illustration3. =0.
self-adjoint Transform each of
equation: the
following equations
into an equiva
Owalitative
Properties of Solutionsand
uation Oscillation Theory
(1) (i) xy'- 2xy'+ 2y =0 |35
(ii) f(x) y"+ gx) y' =0
(iii) y- tan y' + y=0x
xy'-2acy' 2y=0. +
Comparing with ...(1)
elog x2 1
x2
1
X=0, we get
1
2 2
Or y+ y= 0.
|y =0.
We get a, =
(*) 0.
=f), a (x)=gx) and
ao(x)
Multiplication factor
1 a () dx
exp
ao(1)
lao()
gr) dx
f(x) f()
Equatio,
Differential
a s*Sb,
36
interval Multiplyin
on any
Multiplying equation g)
f(*)
get
form
f(x) in the or
be written
is self-adjoint equation and
This equ
may
This equation
d
dx
is
(iii)The given equation
y''- tan x y + y=0. T
2.7.
Compare with y'+ aj(*) y t a(c) y = 0.
an(x) BasicRes
tan x and a (*)
=1. Let u:
do(x)=1, aj(x) =-
We get
factor
Multiplication 1 aj(*) dx
where r(
ao(*)
ao(*) Bolzano
1
1 exp|
(- tan )dx Sup
ex
there
x]
= exp [logcos
=COS X. Th
on any interval a xsb,we get
equation (1) by
cos x
Multiplying at least
y'- cos x tan x y'+ cos x y=0
cos x
1
exp
2 dx
1
exp
1,
dx th
Properties of
,sawation, Qualitative Solutions and
Oscillation
1
Theory
stsb,we exp (log
x)=- 1
atervala x2
Multiplying g equation (1) by on any
interval,
excluding the point
(x2y +xy + y)=0 x
0,we get
the form
or xy'+ y +
1
y=0.
Dhis equation is
self-adjoint equation and may be written in the form
[xy Y+ y=0.
Basic Result
Let us consider the equation
Suppose E isa set of points on the x-axis. A pointx0 is called alimit point of E if
there existsa sequence of distinctpoints x1, X, X3,...of E such that
lim xp =X0
The Bolzano-Weierstrass theorem statesthat every bounded infinite set E has
Integratingby part
Theorem
4. If
ao(*)y'
+ aj(*)
# 0on a<x<b
ao(x)
r()yl +
are continuous self-adjoint
px) w' (*) on a The right handn
derivative
The equivalent first r(*) [u
(2) having
of equation on a<xsb. is the require
be a solution
ofzeros which
number
Let u(x)
an infinite zeros on (a,bl, by Note :The exp
u(r) has
of
all x
onasx<b. number xo.v
where xo Wof tw
Suppose
u(x) =
0for
has an infinite has limitpoint a la, Wronskian
Then u(r) zeros to xo
Proof.
Since theset of
which converges (wh
theorem } of zeros
Weierstrass fx, i.e.s
Bolzano- a sequence
exists
Thus,there Thusthe con
Theorem 6.
is continuous
Since, u(x)
lim u(x) =u(r0),
of zeros n. such that
u(x) a
the sequence depend
xo through linearly
where
lim u(x) =0= w(x0 ).
Proof. W
Then
r(;
exists
Now,since u'(x0) u(r) - u(x0)
Let xoel
w'(x0)= k=
lim
we obtain
Thus
n.
where x o through the
sequence
u()- u(xo) =0and thus u'(xo)=0. Since, v
For such points
Thus.
2.9. ABEL'S FORMULA solutions
is
Theorem 5. A self-adjoint linear differentialequation of second order Thee
r(x) yI + p) y =0.
Let u(x) and vz) be any two solution of (1), we shall show that and supr
The
r(*) [u(c) v (e) – w (x)v(x)] =k
where k is a constant.
Pr
[Gwalior21 exist co
Proof. Since u(*) and v(*) are solutions of (1),
we have
r(o) u' () + px) u(*) =0 .for all
r(x) (o)} + px)u() =0 N
Multiplying (3) and (4)by -u(r),
u(x)respectively and then adding,we e.
u(x) [r(x)v (o)l -uo) r(x) (}=0. u
Properties of
Qualitative Solutions and
-uation Oscillation
Theory
Integrating by parts 39
under limits a to x, we have
(a u(a) r(*) - v() r(x) (*)Y
(x)Y dx
w
dx=0
[u(x) (o)-w ()
r(x)
u*)]=r(a) u(a) (a) -
...(2 The right hand member (5) of
w (a) (a)l.
constant. is ..5)
r(x) [u() (o) -w ()v)]
=k
is the required Abel's formula.
1.
by tha
which
Note :
The expression inside the
W of two solutions
bracket of left
hand side is nothing but the
Wronskian u(*) and v(x) of
equation (1).
o (wher W u(x) U(x)
w (*) (0)
Thus the constant kis zero if and only if
u(o)and vlo) are linearly dependent.
Theorem 6. Let u(x) and v() be two solutionof
r(*) yI+ *)y =0, .(1)
u(x) and Uc) have a common on asx<b.
such that zero Then, u(¢) and ulo) are
Let xo € [a,b]be common zero of u(x) and v().Letting x = xgin the formula (2),
we obtain 0.k=
Thus
r(x) [u(x) (x) – u'()ux)]=0 for all x e (a, bl. .3)
Since, we have assumed throughout that r(x) > 0 ona <x<b, the quantityin
bracket of (3)must be zero for all x onasxsb.
But this quantity is W(u, v) (x)
r(x) W(u, v) (x) =0
since r(x) >
.
forall x e (a, b].
Or
W(u, v) =0, 0
Therefore the
be zero for all x on a <xsb.
Thus, the value of Wronskian
are linearly dependent on
a <x<
solutions u(x) and v(x) solutionsof
be nontrivial linearly dependent
is 7.Let u(x) and ux)
Theorem ...1)
=0 on a xsb
r(x) y}+ p)
y
...1)
a sxo <b.
= where xo is such that 04,05;Ujjain 2003]
and suppose u(xo) 0,
[Himachal 2003,
Then U(r0) 0. = on a <xsb. Then,
there
40 Co =). Qualitative
must have
we which
on la, bl, in (2) Properties
Thus, neither C1 nor of So
x C
all
1s not zero for
U(r) #
Since,
0and likewise C2
0.
Thus, c1 #
TakingI=*in Abel
contradiction.
letting x=Xo in (2), we have
Since, u(ro)=0,
C2 U(x0)=0
Thus uro) = 0.
r(x2)
ua)
THEOREM Now
STURM SEPARATION solutionsof )>0,u(ay
heorem U*)must
2.10. linearly independent have
and u*) be real
Theorem 8. Ifu(x) asxsb
r()
y
y+p) =0
zeros of u(x)
there is
on
exactly one zero of uo)
2002; Gwalio
Further, there
Now
interchangi
+here must be at least
car
zerosof
Let x, and *, be two consecutive
Proof.
supposed by the theorem. separate the zeros o
y= v)
Illustratio
Solution.
linearly indepe
obtuse
Now, for
Let
and
Fig. 2.1
'
formula,we have
r(xj) [u(x) v (*)- w (x1) U(x)]=
k, a
k<0, as r(a)> constant
0, u(x)=
r(x2) [u(*2) v (x)>0,ux1)>0
0, u'
(*2)- u (*2) U(x,)] <0
U(xg) <0,as
of Now Ux1) >0,U(*g) < 0 r(*,) > 0,u(*,) =0, (x)<0
-beorem and u(*) is
u(x) must have
at least one zero continuousfunction.
Therefore by Bolzano
in between x1
Further, there and t.
f cannot, be more than
Now interchanging u(x) and one zero.
ulx)
v(o) then
there must be at least one between two consecutive
walior
Note :We zero of u(c). zeros of )
may restate sturm separation
Jijain 2 Zeros of one of two real theorem in the following
form:The
linearly independent
x)wh solutions of
r(x) yl + px) y=0 on a <xsb
separatethe zeros of the other solution.
YA
Fig. 2.2
Illustration 4. Show that the zeros of the functions a sin x + b cos x and
and u) aretwo
Solution.According to the sturm separation theorem, if u(x)
of
linearly independent solutions
r(o) y+ po) y =0 on asxsb,
are distinct and occur alternately.
thenthe zeros of these functions
must be non-zero, i.e.,
independent solution,the Wronskian
Now, for linear
W(u, v) 0.
an
an!
c sin
C COS
-d sin x
x-
a cos –b sin
x)
+X by a sim
Eq
homoge
y*) = L
Fig. 2.3
S
L() (re) z (9)1-y()(ro)z(w)} dx- (z(2)(r(o) y ()1
Or
r(x) [)z (u)-z(*) y (*)]- r*0) [y(*o)z(o)-zl0)y x0)1
+ lg(x) -p)]y(x) z(x) dx =l
erential
Properties of
Qualitative Solutions and
Oscillation
nately Theory
wheneVe r(x) y(x) z()-
z(x) y(x1))+ [ Ig()- *)] y)zl)ds =0
|43
not
(6) is
positive which is absurd. of (6)is
negative
remain positive on the
lvanish
again
Hence, vanish somewhere
z(*)
3e further notice
will
that y(x)
within this
whole
interval
interval xo
xo
<< 1-
0.
12. CONVERSION OF STANDARD FORM TO NORMAL FORM
...(5)
Just to discuss the of solutions of homogeneous equation it is
oscillation
tain
convenient to deal with equation in which the first derivative term is missing. We
now show that any equation of the form
y'+ aj()y +ag() y =0 ..1)
can bewritten as
u''+ g)u =0 ...2)
) dr=0
Onsetting the coefficient
of
=0 Equation (3)
()
...5)
-(a,()}? --d,
2
ql) =ag(a) 4
Eql
Differerntial
of (1); Qualite
ona
44| vanishes, the
abovro
trangformation
closed
given by (4) never solutions.
Since. v(x) as
the zeros of
no effect
whatsoever on
0 and u(x) is a non-trivial
solution of asr<i
Proof.
(2) has g(x)< It
10. If
Theorem one zero.
has at most So that u(xo)
=0. 0 asrSb.
we follow u'(0)+
then u(x) zero of u(*),
xo be a
Proof. Suppose solutionof given equation,
Since. ulx)is
non-trivial
where
u'(x0)<-0.
.: Either
(x0)>0or
w'
u(x) is positive
over somne interval to that
(x0)>0, so that Such
that w' Since
We suppose
right ofX0
Since,q()<0,
u''(x) =-g(x) u(x) ()is an increox
the slope w' and
This implies that same wav it
over the same interval.
of xo, and in
the
is positive to the right
cannot have a zero The
function,so u(x)
none to the left of xo. no zeros at al contradi
u'(x0)<0, so u(x) has either
holds when
A similar argunent
only one zero. of equation +
u'' g(*) u
If L q(x) dx =o [A
2. SI
x-axis.
many zeros on thepositive
then ulx) has infinitely
of times
most a finite number
Proof. Assume that u(x) vanishes at
for all x>
>1 exists with the property that u(x)
0
3.
0<x<0,sothat a point Xo 1
Ue) u)=|
- q() dx +
Use (1)to conclude that vx) is positive if x is taken large enough. This sb
that from (2), u(x) and u ()
have opposite signs if x is sufficiently large, S0
negative. This contradiction proves the
theorem.
Theorem 12. If yx) be a non-trivial solution of
equation
y' + ql*) y = 0, q()> 0,
of
(1) Qualitative
Properties of
Solutionsand
Oscillation
Theory
interval.
on
a closed a
sxsb. Then |45
y(x)
has at most
asxsb a
finite
number of
Proof. If possible, let y(x) zeros in
o)*0. asxs It followsthatthere
have an infinite
exist in number
6.
of
zeros,say
a
sxsb a
point x0 and a *1, 02, 3,...in
sequence of zeros
erval to # X0, for n =1,2,
n where
such thatx, - *0:
3,...
EXERCISE 2.1.
q(*) u
1.Find the adjointequation of the equation:
sin x y'"+ cos xy'+ 2y =0.
times ;y'+jy=0.
all x2 self-adjoint equation :
3. Transform each of the followingequations into an equivalent 3
(0) (+)y'+ 2* y' + 3y=0 (Ans. [(+ 1) yl'+y=0)
g) y =
[Ans. (y' )+ y=0]
(i) fo)y''+ 0.
y+ [Himachal
2004]
is continuous
zeros.
and g(x)
y =0
<0throughout a
Differential
s xsb. then, no
Equation,
[Himachal 200,
non triv
Cho
if
Show that
9.
of
solution
y''+ g)
Theorem.
y =0.
adjoint operators.
? Find the condition for second
adjoint equation
adjoint operators
and [Kurukshetra 20
13. What are
equation
tobe self-adjoint. [Meerut 2006; Kurukshetra 206
Identity. 3.1
14. State and
prove Lagrange's
it with thehelp of an example.
illustrate
Abel's formula
15. State and prove [Delhi 2001, 03,
very
ofBessel's equation also
16. Find the normal form
dy+ xy'+ ( - p')y =0
solution has infinitely many positive zeros
and use it toshow that every non-trivial 3.
of
Show that every non-trivial solution
17.
k
1
has an infinite number of positive zeros if k > and only a finite number if k< Or
4 4
and
18. Ify() is a non-trivial solution of +
y'' q(r) y=0,show that y(*) has an infinite number lin
k 1 1
positive zerosif q(x) > for some k> and only a finite number if q(x) <
4 43 the
for
19. Every non-trivial solution of y''+ (sin x + 1) y=0 has an infinite number of post
Zeros.
w)
is