Series Solution
Series Solution
CHAPTER
Bessel's Functions and Legendre's
(8 Polynomial, Solution in Series
For Linear Equations
SECTION-I
A.INTRODUCTION
In the early study of differential equations, we have seen that linear differential equations have
solutions in terms of elementary functions. e.g., algebraic, trigonometric, inverse trigonometriC
exponential, logarithmic functions and all other functions which can be obtained by adding, subtracting
multiplying and dividing. But there are many second order differential equations which have no solution
-+P2 y = 0 ..(1)
where Po. Pi. P2 are polynomial functions of x, in terms of infinite convergent series.
dy+p+Qy=
dx
0 ..2)
dx
where and Q
P2
P Po
=
Po
X0 is called an ordinary point of equation (1) 0 itPo (0))*
When x = 0 is an ordinary point of (1), its solution can be expressed as a series in the form
y = G + a,X + a z t o « o =
k0
) X= 0 is called singular point of (1) if P, (0) = 0. Further if xP and -Q possess derivatives
rders in the neighbourhood of x = 0, then x = 0 is called a regular singular point of equation (1)
Wisex= 0 is an irregular singular point of ().
when x= 0 is a regular singular point of (), then at least one of its solutions can be expressed as
y = a ) + a, + a + 4a oo)
149
150 ODERN'S INTEGRAL CALCULUS AND THEORY OF EQUATIONS (B.A./B.Sc. 1, Sem. 11, Kashmir Unis
where k is any rational number i.e. k is any positive or negative
integer or
a tracuon.
When x 0 is an irregular singular point of cquation (1) then equation (1) has no series solution,
solution.
ethod for solution in series when x = 0 is an ordinary point of the linear difterential equation
on
Po +P+P2 y = 0 )
dx
Step (). Let y = ao + aX + a r + agr+ .- (2)
be solution of (1)
y = +
a ax +
ar +a +
4,"+ * =
2= 0aX
Then a + 2ar + 3a,x+ .
dx + na"-l+. kay
k =1
Coefficients of x" : (n + 2) (n + 1) a, + 2+ n a, t+ an =0
+1
n+2 ( : n + 1 z 0]
n+2) (n +1) " +2n3)
1
as 5-3 4 4-2 6-4
1 1
a4 -3314 48642
Substituting the values of a;'s in (2), we get
1 x
+ 2.4-6:8
a+2y = 0 ...(1)
Then
dy
ka, and d=
dr2
k = (0
ddx k = 0
dy dy in (), we get
Substituting the value of y.and
dx dx
a2+. + n{n -
1) a ** (7* )na, + -
(1 ) [2.1 a, 3.24
+ + 4.3
*'+.
+n
*7* 2) (n + 1) a,n +2r" . . . 2r [a, + 2a,x + 3ax+
+ 2 [4 + ax
+
az * . *. 0
of x, we get
Equating to zero the coefficients of various powers
Coefficient of Constant term: 2a+ 240 0 a2 =-ao
Coeficient ofx 643 -2a, + 2a,= 0 ag0
0
Coetficient of "; (n + 2) (n + 1) a,2 -n (n - 1)4,- 2na, + 24,=
n(n-1)+20n-1),
(n+2)(n-1).
an+2 n(n-1)+2nan
(+2) (n +1) (n+2)(n
(n+2)(n +1) == n
of ..3)
On+2 n+1
( : n+2 0)
Putting n = 2, 3, 4, 5, 6 . . in (3), we get
2 3 1 4
y =
a a - a+0x3-
*
+0-$-ox
y = 1-- solution
.+axis the required
of equation (1) containing arbitrary constants a and a.
EXERCISE 8.1
Solvethe following equations in series
d2
+xy = 0
d'y-y
d2
=0
+xy =0
4. 2-2y
dx
0
5. d-y+y =
d2 dx
dx dx
aESSEL'S FUNCTIONS AND
LEGENDRE'S POLYNOMIAL, SOLUTION IN SERIES FOR. 153
153
ANSWERS
9!
2. y= ao
4!
3. y= 4 1
3-4.7.8
4. y 1
5. y
a
6. y =
a 8
128
Method for solution in seríes of a linear differential
ofof the equation.
equation when x = 0 is a regular singulari Yon
dx .(1)
where Po, P, P2 are polynomial functions of x.
his equation in k is càlled the Indicial Equation. in terms of a by equating to zero the c o
ep 4. Determine the values of a, az, 43, . .
of y indeterminate.
of the second order linear differential
Note. General solution of the equation. The indicial equation,
to these two
equation, is a quadratic and hence it gives two values of k in general. Corresponding
Values of k, we shall have two solutions (from 2). If y = U and y = V are the two solutions of (1), then
y A U + BV, where A, B are constants, is called a general solution of (1).
TYPE
When the Roots of the Indicial Equation are different and they do not differ by an integer
t h e indicial equation has wo roots k =a , Bwhere I a -Bl is a non integer, then we get two
Mo solutions then the general solution of the equation is given by y = AU + BV, where A, B are
arbitrar constants.
Example 3. Solve completely in series the equation.
2x(1- dx +1-)+3y
Solution. The given equation is:
2x(1-x)+1-)+3y= 0
dr
y = la +a,x + a + s , o] .(2)
Let
where ao# 0, a az .... are all constants, be a trial solution of (1).
and =
ak(k -1)*i+a (k +
1) kr* -i+a(k + 1) (k+ 2) x* +
dx
Substituting these values in (1), we get
(:40)
ak (2k ) =
0 k =
0 or
roots of the indicial equation are different and do not differ by an integer.
Clearly the
coeff. of x; 2a, (k + 1)k- 240 k (k 1) + a (k
-
+ 1) -
ag k+ 34 =0
-k 3) = 0
a (k + 1) (2k+ ) - % [2k -
a k+ ) (2k +
1)-ag (2k-3) (k + ) =0
(2k-3) .3) k+ 1 #0
2k+1
0
2) (k 1)- 2a, (k+ 1) k+ a (k + 2) a, (k + 1) + 3a1 =
of x+l; 2a, (k + +
-
coef.
BESS FUNCTIONS AND LEGENDRE'S POLYNOMIAL, SOLUTION IN 155
a k +2) (2k +3)
SERIES FORR . *******
-a [2k + 2k + k+ 1 -3] =0
a2 (k + 2) (2k + 3) -a, (2k +3k 3]= 0
2) 0
a, (k+ 2) (2k + 3) a (2k - 1) (k + 2) 0
-
=
=
2k-1
a2 a1 2k+3 k+2 0]
This equality can also be obtained
from (3) by changing k to k +1 and a to a2 and 4o to
a
Similarly we can
prove ag =
a2
2k+1
and so on.
When k =0, then
a =
4o
- 300
5)--)G)-
and
-2
When k =
.then a o-4o
2 =(0) (a,) = 0 and ag = 2 = ) =0 and so on
3
ie.
( - * ) y =o .(1)
Let y =
*lao+a+a+ +s. ] ..(2)
where * O, aq, az, a.4.. are constants, be a trial solution of (1).
a = dkrk- 1+a, (k + 1) **+a, (k + 2) **+ s . o
dr
EQUATIONS (B.A./B.SC. I1,
Sem.
Kashmir Uni
I, Kashmi.
THEORY OF
156 CALCULUS AND
NS INTEGRAL
+ 2) (k
+ 1)x* +
o. 0
and
agk (k - 1) sk-2+a, (k
+ 1) kr* -I+ a, (k
2
Substituting these values in (1), we get * o] +x lakrk-1,
2) (k + )x a.4.
I +
a, (k
+
lak k 1) - 2 + a, (k + 1) kr - +1.
a2*t . . o]=0
-
identically
of x, we get
Equating to zero the co-efficients of various power Indicial equation)
coct. of r* : ak (k - 1) + a k - n'a = 0
differ by an
integer. =0
coeff. of rk*l;a, (k +1) k +a, (k + 1) -a,n 3)
0
a [k +1) - n']
=
+
a =
and so on
a,ICk+r-a+a,-2 = 0 .6)
[Note it carefulyl
From (6) -2
(k+r)-n
Putting r = 2, 4, 6, s . , we get
do
(k+2)-2
(&+4- &+4-(&+2-
When k = n,
4
(n+2- 2 (n+1)
n+4)-|(n+2- 2'21(n+2(«t)
Similarly,
203:(n+3) (n+2)(n+1)and so on.
Substituting the values of a's in (2), one solution of (1) is
y= 4g
2 22 (n+1) 22!(n+2)(n+1) 2°3! (n+3) (n +2) (n+1)
ta
cuS FUNCTIONS AND LEGENDRE'S POLYNOMIAL, SOLUTION IN SERIES FOR.
B E S S E L ' S
157
**s***
Now when k= - n, the other solution of (1) is obtained by changing to -n in the first solution,
and we get
x
t a,V (say)
y do 22 (1-n) 2'21(2-n)(1-n) 23!(3-n)(2-n)(1-n)
Hence the general solution of (1) is
U and V
y and
dy we get two linearly independent solutions of the equation. If
these solutions are
ak
where A, B are arbitrary constants.
then y = AU + BV is the general solution of the given equation,
in series the equation
Example 5. Integrate a
= 0
0 .(1)
2)
Let y
=
l a + at+ax+ o ]
all constants, be a trial
solution of (1).
where a # 0, d1, a2 . . .
are
a2 (k + 2) (k+ 1) x* + . . 0 0
+ a, (k + 1) k*
+
and aqk (k- 1) - 2
dx
Substituting these values in (1), we get
1) * 2) (k+ o]+ (1-5x) lakk-14
- ) [ak (k -1) *-2 + a, (k + 1) kr*+a2 (k+
+...
+ + 1) -
1) a (k
-
5agk 440
-
a (k + 1) (k+ 1) =
an (k* + 4k + 4)
k+2]
+
coef. of * * ' ; a , (k + 2) (k + 1) - a , (k + 1) k + a , (k + 2) - 5a, (k + 1) - 4a, = 0
a (k + 2) (k + 2) (k+ 6k + 9)
(k+412
Similarly k+10, and so on.
Substituting these values in (2), we have
y ag . (3)
ak log 1 * +
+40 xtk+1 (k+1
+1 (k+12° k+l (k+1
.. (4)
Putting k = 0 in (3) and (4), we get
y =
ag [ 1+ 2x + 32 + 43 + ..
0o] =
an U (Say)
and
ok
=
ag log x [l + 22x +3x + 4+ . .
o] -
240
1.2 x + 2.32+ 3.41 .
koy logx(4, U) - 24, [1.2x+2.312 + 3.4+. o =bo V(Say)
..
0o)
are arbitrary constants.
Example 6. Bessel's Equation of zero order.
Solve
++y
dx
= 0.
ecCSELS FUNCTIONS AND LEGENDRE'S POLYNOMIAL, SOLUTION IN SERIES FOR 159
.
.
Solution. The given equation is
d*y +xy = 0
dr dx .(1)
Let y =
[aq t ax + a,x +.. co] .(2)
where ag *0, a1, a2. are constants be a trial solution of (1).
d
= apk -+a (k + 1) x* + a, (k +2) *l+...
az (k + 2) + a = 0
a2 (k+2)
Similarly we can prove a= 0 and a4 t44?a+n2 and so on.
from (2), we have
1
(k+2)2 (k+4 a+2 | (3)
Differentiating (3) w.r.t. k, or both sides, we get
x
ok
o log.* +.. o
(k+2) (k +4) (k +2)
+ 22
(k+2 (k+4) (k+2) \k +4 k+2|
.(4)
Putting k = 0 in (3) and f4), we get
2 = ay U(Say)
=ok Ulog r+
=
a [U log x +
V]
EQUATIONS (B.A./B.Sc. ,
Sem. li, Kashmiri
THEORY OF
160 MODERN'S INTEGRAL CALCULUS AND
where V l
Hence the general solution of (1) is
+ C240 (U log
x + V)
y =
C (a6U)
= AU + B (U log x+ V)
BV where
=
(A + B log x) U +
.. 00 andd
U=1,*2.22 642*
*******
TYPE II.
and one root of it makes some co-
When the root of indicial equation differ by an integer
efficients of y as infinite.
integer. Let k a makes
is =
... (1)
x-) -3x-y 0
d (2)
y = * (4 + a,x + a +... 00)
Let
where ag # 0, a1, ag.are constants be a trial solution of (2)
and
dy = ak (k-1)k-2 +a,k + 1) krk=+ a, (k + 2) (k + 1)+
d
Substituting these values in (1), we get
(x-)lak (k -1) ** +ak + 1) kr*- + az (k+ 2) (k + 1) *+
-3r l4k x-+ a,(k + 1) *+a, (k +2) * ' t .
ldentically
[4 +a,*+ a, +2
-
o] 0 =
a (k+ 1) -a k +2) = 0 k + 2 * 0]
a2 k+2)2a
k+1 . (4) [Using
and so on.
Putting these values in (2), we get
y =
k
y = Ap [k +(k + 1)x+ (k + 2)+ (k + 3) r + . )
. (5)
Differentiating (5) w.r.t., k we get
TYPE IV
When the roots of indicial equation differ by an integer and one root of it makes a co-efficient
y indeterminate.
Let a, B, (a, B) be the mwo roots ofthe indicial equation where ß - a is an integer. Let k = a makes
+n (n+1)y = 0.
Sobve: (-
OF EQUATIONS (B.A.B.Sc. l, Sem.Il, Kashmir tUni.
AND THEORY
162 INTEGRAL CALCULUS
MODERN'S
Solution. The given equation is
d- -2n(n+1y= 0
dr dr
+ a,r+.. ) 2)
Let y =
* (a +a,r
are constants, be a trial solution of (1).
where a *0, a1. az
..
-2x la,- + a
r*+2+ o] = 0 ldenticaly
+n (n + 1)la% *+ a,+l+a,
...
But when k= 1, a
= 0
and of them viz. k = 0, makes a
Thus the roots of indicial equation differ by an integer one
indeterminate.
(k-1)-240k + n (n + 1) ag 0 =
Similarly
and a, (k +4) (k+ 3) a,{k +2) (k+3) n (n + 1)}
= -
and so on.
When k= 0, we get
n (n + 1)
az (2-1) = d l - n (n + 1)]
2!
(n +2)(n -1).
ag (3-2) = a, [2-n-n]
3!
a (4-3)=az 16 - n*- n] = - az (n +3) (n - 2)
[Using the
as
value
y=)1" 2!
2 (n+3) (n+) n(n-2)4
4!
5!
(n-)(n),5++
is the general solution of (1)
Example 9. Hermite Equation. i t contains two arbitrary constants
Sohe the equation: dy
+2py 0. where p is any constant.
d
Solution. The given equation is
d
d"y-2x+2py =0 dx .. (1)
Let
where
y =
l d +a,x +a^r+... o] (2)
a* 0, a. a.. are constants, be a trial solution
of (1).
dy
a kr*=' +a, (k + 1) *+ a, (k +2).+l..0
and
dy
=ak (k -1) x-2+a (k +1) kr-+ a, (k +2) (k + )+.
dr
Substituting these values in (1), we get
lan k (k- 1) -2 + a, (k + 1) kr*-l + a, (k +2) (k +1) r . + o]
-
2x la +a, (k + 1)*+a, (k + 2) x* ]
+2p lag + a * + a * * .o) = 0
Equating to zero the various power of x, we get
coeff. of r-2; ak (k -
1) = 0 Indicial equation)
k 0,1 (400)
Clearly the roots of indicial equation differ by in integer.
Here k-k l-0=0
0 ak 0
coef. of - ; a, (k+ 1)k = k +l * 0)
When k = 0, a, can have may values and hence indeterminate.
But when k = 1, aj = 0
Thus the roots of indicial equation differ by an integer and one of them viz. k = 0, makes a
indeterminate.
=0
oEl. of ; (k + 2) (k + )a2- 2ak +2p ay
(k+ 2) (k + ) a = 2 k -p)
=0
3) (k + 2) a1- 24,(k )+ 2pa,
+
COelf. of rk+ l; (k +
2(k+1 -pla
k+3) (k + 2) a1
=
When k 0, we get
(2-(1) a = - 2pa%
2!
MODERN'S INTEGRAL CALCULUS AND THEORY
OF EQUATIONS (B.A./B.Sc. I1, Sem.I1, Kashmir Uni
164
-
2 (p-1)a
(3)(2) a = 2(1 -P)a1 a3 3!
2 ( p - - 2 p{p-2)
2 (2 - p)a2 a 4 4!
(4)(3) a = 4:3
22 (p-1Dp=S
Similarly
2p-3)
54
5!
a etc.
5!
y 4!
(: it contains two arbitrary constants]
is a general solution of (1).
EXERCISE 8.2
Ci) 2x3-++1)=0
iv) 2x d2
+(2»2-)+y=0.
dr
(x-)2+(1-x)-y =0 (n 1+1)+2y=0
2. ) dr
(iv) -y0.
cin +*y=0
dr
dx
-(1+3x)-y =0
3. +(a2 -4)y=0o
) 2 d +ax (i) x(1-x)
dx
4. ?y=0 () (1-*)-2+2y
dr
=0
dx
6. () dyy+xy =0
2 t y =0 (ii) +(1+x)+2y =0
d
)
aEsSELS FUNCTIONS AND LEGENDRE'S POLYNOMIAL, SOLUTION IN SERIES FOR . 165
ANSWERS
1. () y AU + BV, where
U-1++ 6:9
+... 0o and
8-11 2 13-16
8 11 14
+t... 0
10-13 10-13 16 T.
y=A-2 B ...0
(ii) y= A
25-4.9 2.5.4.9.6.13
22 2
B 3.4.7.8 3.4-7:8-11-12
(iv) y= A/2
+-B s7
2. (i) y=Al + BV, where
1-2.5
x+... 00 and
U= 1-2r+
V=Ulog x+2|2
AU + BV where
(iii) y =
-x+0o and
=
1-* 3 (21
V= Ulogx*: -Fa*i*Joi*i)'*
(iv) v= (A + B log x) + (3!)
(212
BV, where
3. () y =AU +
and
MODERN'S INTEGRAL CALCULUS AND THEORY Or rQUAIONO (MA/ d0, , ierm, 1, Kashrr Uni)
166
V=U log+
() y= AU + BV, where
U- - nd
V=- U log r + | | -- 5 1 - I 1 t . o
4y 4 34 8.9 3-4:78112
4.58:9 4:5:8.9-12 13
-+0|+aX
5
(ii) y= a
(iv)y= a
L*R165.3-24
8 16
+
t+Br-3/2
(ii) y = -+-
2 212 3 +B 5 57
4
(iv) y= A
2:5245.9 .+B + 2.45-7
6. ()y=A
22.42 2242 6
+B logrta