Professional Documents
Culture Documents
Equation (5.2) is not, in general, exact because M y = p(t) and N t = 0 ; exactness would
µ (t) = e !t
p(s)ds
(5.5)
Multiplying equation (5.2) by the integrating factor gives M (t, y)dt + N(t, y)dy = 0 where
!" !"
= M (t, y) = µ (t)p(t)y # µ (t)q(t),!! = N(t, y) = µ (t) (5.7)
!t !y
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.1
From the second of equations (5.7),
where h may depend on t but not on y. Differentiating with respect to t and setting the
!" dh
= µ #(t)y + = µ (t)p(t)y $ µ (t)q(t) (5.9)
!t dt
d "t p(s)ds
= p(t)e "t
p(s)ds
µ !(t) = e = p(t)µ (t) (5.10)
dt
dh
= ! µ (t)q(t) (5.11)
dt
THEOREM. The one-parameter family of solutions to the general first order linear
1 "
y= C + ! µ (s)q(s)ds $ (5.13)
µ (t) # t %
where µ (t) = e !t
p(s)ds
.
The following constructive algorithm gives an alternative proof of (5.13) and is generally
more useful for solving the general linear equation than memorization of the formula.
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.2
ALGORITHM 1. To solve y! + p(t)y = q(t)
(2) Multiply the ODE through by µ (t) , giving µ (t) y! + µ !(t)y = µ (t)q(t) .
d
(3) Apply the product rule and observe from this that [ µ (t)y ] = µ (t)q(t) .
dt
This has the form y! + p(t)y = q(t) , where p(t) = 2t and q(t) = t . An integrating factor
µ (t) = e !t = e !t
p(s)ds 2sds 2
is = et . Multiplying equation (5.14) through by the integrating
factor
d ! t2 # t2 2
ye = e ( y% + 2ty) = tet (5.15)
dt " $
Hence
'1 2 * 1
y = e!t # " ses ds % = e!t ) et + C , = + Ce!t . ■
2 2 2 2
(5.16)
$ t & (2 + 2
Putting this into standard form gives y! " y / t = te"t . Hence µ = e "
! (1/t )dt
= e! ln t = 1 / t .
Multiplying throught,
d
[(1 / t)y ] = (1 / t)( y! " y / t) = te"t (5.17)
dt
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.3
Therefore
d t
! e y # = et ( y% + y) = et cost (5.20)
dt " $
#1 &
y = e!t " es cos s!ds = e!t % et (sin t + cost) + C ( (5.21)
t $2 '
The initial condition gives C = 1 / 2 and hence the solution of the initial value problem is
1
y= " cost + sin t + e!t $ . (5.22)
2# %
We write the general linear first order initial value problem in terms of the linear
differential operator D = d / dt as
The operator D has several useful properties, among them the following.
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.4
D[eat y] = eat (D + a)y (5.25)
Property (5.24) is the fundamental theorem of calculus (let f (t) = F !(t) and the result
follows), and property (5.25) follows immediately from (5.26) with f (t) = at. Equation
(5.26) follows from the product rule for derivatives, and equation (5.27) is a special case
D "$ e !t y %' = e !t
p(s)ds p(s)ds
q(t) (5.28)
# &
= D "1D #% e !t
p(s)ds &
ye !t y ( = D "1 #% e !t q(t) &( = !t e !
p(s)ds p(s)ds p(r)dr
s
q(s)ds + C (5.29)
$ ' $ '
Then by (5.26),
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.5
D[ µ (t)y] = µ (t)[D + p(t)]y (5.32)
Substituting (5.23),
t
µ (t)y = C + ! µ (s)q(s)ds (5.34)
t0
theorem.
# to &
The proof holds regardless of the choice of initial condition on µ (t) and hence we are
free to set µ (t 0 ) = 1 . This is equivalent to defining µ (t) = exp " ! p(s)ds $ where we
# t %
1 "
µ (s)q(s)ds %'
t
y= $
µ (t) #
y0 + !t 0 &
(5.37)
uniqueness, we observe that the differential equation can be written in the form
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.6
f (t, y1 ) ! f (t, y2 ) = q(t) ! p(t)y1 ! q(t) + p(t)y2
= p(t) y1 ! y2 (5.38)
< M y1 ! y2
t
An integrating factor is µ = exp " (!2)dt = e!2t . From equation (5.36)B
0
Bernoulli Equations
These equations are similar in form to equation (5.1), although they are not linear, and
Were it not for the factor of y n equation (5.40) would be linear, and in fact it is linear if
z = y1! n (5.41)
Differentiating,
1 n
y! = y z! (5.43)
1"n
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.7
Substituting (5.43) into (5.40) gives
1 n
y z" + p(t)y = y n q(t) (5.44)
1! n
1
z" + p(t)z = q(t) (5.45)
1! n
z! = 4y3 y! (5.48)
z! t
3
+ ty = (5.49)
4y y3
Multiplying through by 4y 3 ,
z! + 4tz = 4t (5.50)
An integrating factor is µ = e !
4t 2
= e2t . Multiplying equation (5.50) through by µ and
rearranging,
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.8
d ! 2t 2 #
= e2t ( z% + 4tz ) = 4te2t
2 2
ze (5.51)
dt " $
Integrating,
!t
2 2 2
ze2t = 4se2s ds = e2t + C (5.52)
2
z = 1 + 15e2!2t (5.53)
and hence
( )
2 1/4
y = z1/4 = 1 + 15e2!2t (5.54)
Exercises
sin t C ! cost
(1) t 2 y! + 3ty = , t <0 ans: y =
t t3
et (t ! 1) + C
(2) ty! + 2y = et , t > 0 ans: y =
t2
2 cost sin t
(3) y! + y = 2 , y(" ) = 0, t > 0 ans: y =
t t t2
(4) y! + 4t 3 y = t 3 ,!!y(0) = 1
dy 1
(6) y(1) = ! / 2 + y = et y(a) = ya
dt t
n
(7) y! + y = At " n ,
t
5Ct
(8) t 2 y! + 2ty " y 3 = 0 ans: y = ±
2C 2 ! t 5
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.9
3 139t 18 ! 3t
(9) y! " y = 4y "5 ,!!y(1) = 2 ans: y = 2 6
t 17
dy 1
(11) = y ans: y = ln !t ± t 2 + C #
dt e ! t " $
ecos y
(12) y! = cos y
te sin y + y
(13) Ricatti's equation is y! = p(t) + q(t)y + r(t)y 2 . Suppose that we already know one
particular solution u(t).
(a) Show that a more general solution is given by
1
v(t)= u(t) +
w(t)
for some function w(t).
(b) Show that w!(t) = "[q(t) + 2r(t)u(t)]w(t) " r(t)
Historical Notes
Jacob (Jacques) Bernoulli (1654-1705) was the first member of the great Bernoulli family to study mathematics
(including his nephew Nicolaus (1662-1716), his brother Johann (1667-1748), Johann’s sons Nicolaus (1695-1726),
Daniel (1700-1782) and Johann II (1710-1790) and Johann II’s sons Johann III (1744-1807) and Jacob II (1759-1789)
was born into an influential Swiss family; his father ran a lucrative spice business, was a magistrate and member of the
Basel town council, and his mother came from a family of bankers. He attended the University of Basel, receiving a
master’s degree in philosophy (1671) and a licentiate in theology (1676). He traveled as a tutor, first in Geneva and
later in France, studying with the leading mathematicians of Europe including Boyle and Hooke, finally returning to
teach mechanics at the University of Basel in 1683. He married Judith Stupanus the following year and eventually had
two children, neither of whom studied mathematics. Bernoulli was appointed professor of mathematics in 1687. His
younger brother Johann also began to study mathematics at about this time. Johann was jealous of his brother and
become one of his greatest critics, eventually being dismissed from the University and leading to a total breakdown in
sibling relations by 1697. Jacob Bernoulli’s contributions to mathematics include a study of logic and algebra (1685);
probability (1685); geometry (1687); several works on infinite series (1682-1704); and the law of large numbers
(1689). He showed that the isochrone (a curve of constant rate of descent in a gravitational field) can be solved with a
first-order nonlinear differential equation (1689); developed the technique of separation of variables and used the word
“integral” in calculus for the first time (1690); a studied the solution of the Bernoulli Differential Equation (1696);
properties of curves including parabolas, epicycloids, evolutes, and spirals and lemniscates (1692-1694); found a
method to keep a drawbridge balanced (1695); and probability and games of chance (published posthumously, 1713).
© 2005 BE Shapiro [Math 351 Spring 2005 revised 3/25/05] Page 5.10