ORDINARY DIFFERENTIAL EQUATION

CHAPTER 2

FUNDAMENTAL SET OF SOLUTION

We shall consider a homogeneous linear equation of order m of the form

dm w dm−1 w dm−2 w
= p 1 (z) + p 2 (z) + ..... + pm (z)w (1)
dz m dz m−1 dz m−2

Let z0 be an ordinary point of (1) where w(z), w ′(z), ....., w m−1 (z) take arbitrary
prescribed values at z0 .

Now corresponding to different sets of prescribed initial values, we get different

holomorphic solutions of equation (1) in domain of an ordinary point and as such
an infinite number of different solutions are possible. But all these solutions are
not essentially different and we shall prove that m solutions can be found s.t. any
solution can be written as a linear combination of these m solutions . For this we
start with idea of linear dependence a set holomorphic functions.

Definition 1.2 :
Let f1 (z), f2 (z), ....., fn (z) be a set of n functions holomorphic in an open region R.
The set of functions f1 (z), f2 (z)....., fn (z) is said to be linearly dependent if constants
c1 , c2 ...., cn not all zero can be found s.t. c1 f1 (z) + c2 f2 (z) + ..... + cn fn (z) = 0 in R
and the functions f1 (z), f2 (z)....., fn (z) are said to be linear independent if they are
not linearly dependent.

Definition 1.3 :
The wronskian of the function f1 (z), f2 (z)...., fn (z) denoted by ∆(f1 , f2 , ..., fn ) or

1
∆(z) is defined by

f1 f2 ... fn
f1′ f2′ ... fn′
∆(f1 , f2 , ...., fn ) =
... ... ... ...
(n−1) (n−1) (n−1)
f1 f2 . . . fn

Theorem 1.4 :
A necessary and sufficient condition for linear dependence of a set of functions
f1 , f2 , ......, fn holomorphic in an open region R is that ∆(f1 , f2 , ...., fn ) = 0 in R.
Proof :
i) Condition necessary :
If f1 , f2 , ....., fn are linearly dependeny then we can find constants c1 , c2 , .., cn not all
zero s.t. in R
c1 f1 + c2 f2 + ........... + cn fn = 0

Differentiating successively

c1 f1′ + c2 f2′ + ........... + cn fn′ = 0

c1 f1′′ + c2 f2′′ + ........... + cn fn′′ = 0

........................................

c1 f1n−1 + c2 f2n−1 + ........... + cn fnn−1 = 0

Eliminating c1 , c2 , ...., cn we get

∆(f1 , f2 , ....., fn ) = 0

ii) Condition Sufficient

We shall prove sufficiency of condition by induction. For n = 2, the theorem can be

2
easily seen to be true.
Let
∆(f1 , f2 ) = 0 =⇒ f1 f2′ − f1′ f2 = 0 in R

If f1 vanishes identically in R i.e., f1 ≡ 0 in R , then c1 f1 + 0.f2 ≡ 0 in R for any

c1 6= 0 which shows that f1 , f2 are linearly dependent.
If f1 6= 0 in R and E denotes the set of zeros of f1 in R then E has no limit point
in R − E . Then in R − E, f1 6= 0 and
 
d f2 f1 f2′ − f1′ f2
= = 0
dz f1 f12

f2
=⇒ = constant = A (say)
f1
=⇒ f2 = A f1 in R − E

Since f1 , f2 are holomorphic in R, we have f2 = Af1 everywhere in R i.e. f1 , f2

are linearly dependent in R.
Now assume that the theorem holds for a particular value of n. Let

∆(f1 , f2 , ....., fn , fn+1 ) ≡ 0 in R

If ∆(f1 , f2 , ....., fn ) ≡ 0 in R, then by hypothesis f1 , f2 , ....., fn are linearly de-

pendent and so f1 , f2 , ....., fn , fn+1 are also linearly dependent.
If ∆(f1 , f2 , ....., fn ) 6= 0 in R identically, then let G be the set of zeros of holomor-
phic function ∆(f1 .....fn ) in R. Then we can find functions u1 (z), u2 (z), ....., un+1 (z)
in R − G
u1 f1 + u2 f2 + ................ + un fn + un+1fn+1 = 0

u1 f1′ + u2 f2′ + ................ + un fn′ + un+1fn+1

′
= 0

...............................................................

.................................................................

u1 f1n + u2 f2n + ................ + un fnn + un+1 fn+1

n
= 0.

3

What can be the functions u1 (z), u2 (z), ....., un+1 (z) so that above relations hold?
The answer may be the following
u1 (z), u2 (z), ....., un+1 (z) may be the cofactors of last row of

f1 f2 ... fn fn+1
f1′ f2′ ... fn′ ′
fn+1
∆(f1 , f2 , ...., fn , fn+1 ) = ... ... ... ...
(n−1) (n−1) (n−1) (n−1)
f1 f2 . . . fn fn+1
(n) (n) (n) (n)
f1 f2 ... fn fn+1

so that
u1 f1 + u2 f2 + ................ + un fn + un+1fn+1 = 0

u1 f1′ + u2 f2′ + ................ + un fn′ + un+1fn+1

′
= 0

...............................................................

.................................................................

u1 f1n + u2 f2n + ................ + un fnn + n
un+1fn+1 = ∆(f1 , f2 , ...., fn , fn+1 ) = 0.

u1 , u2......un+1 do not all vanish simultaneously in R − G. If at a point un+1 = 0,

then u1 , u2 , .., un do not all vanish simultaneously and so

∆(f1 , f2 , ...., fn ) = 0

which is not true. Hence un+1 6= 0 in R − G.

ui
Setting vi = un+1
i = 1, 2, ....., n , we have

v1 f1 + v2 f2 + ................ + vn fn + fn+1 = 0

v1 f1′ + v2 f2′ + ................ + vn fn′ + fn+1

′
= 0

...............................................................

................................................................. (2)

v1 f1n−1 + v2 f2n−1 + ................ + vn fnn−1 + fn+1

n−1
= 0

v1 f1n + v2 f2n + ................ + vn fnn + fn+1

n
= 0

4
 Since ∆(f1 , f2 , ...., fn ) 6= 0 in R − G , it follows from first n equation of (2) that
v1 , v2 , ...., vn are holomorphic function in R − G.
Differentiate first equation of (2) and using the second

v1′ f1 + v2′ f2 + .... + vn′ fn = 0

Similarly, differentiate second equation and using the third

v1′ f1′ + v2′ f2′ + ................ + vn′ fn′ = 0

...............................................

v1′ f1n−1 + v2′ f2n−1 + .......... + vn′ fnn−1 = 0

Since ∆(f1 , f2 , ...., fn ) 6= 0 in R − G, so

v1′ = v2′ = ....... = vn′ = 0

in R − G and hence v1 , v2 , ....vn are constants in R − G .

Since v1 f1 + ...... + vn fn + fn+1 = 0 in R − G and f1 , f2 , ...., fn+1 are holomorphic
in R, so
v1 f1 + v2 f2 + ....... + vn fn + fn+1 = 0 in R

so that f1 , ..., fn+1 are linearly dependent.

Thus the theorem holds for n + 1. Hence by induction the theorem holds for n ≥ 2.

Theorem 1.5 :
If w1 , w2 , ...., wm are m analytic solutions of differential equation (1) in domain D
of an ordinary point z0 and ∆(z) = ∆(w1 , w2 , ...., wm ), then
Rz
p1 (z)dz
(i) ∆(z) = ∆(z0 ) e z0

where the path of integration joining z0 to z lies in D.

5
 Proof :

w1 w2 w3 ... ... wm
w1′ w2′ w3′ ... ... ′
wm
∆(z) =
... ... ... ......
(m−1) (m−1) (m−1) (m−1)
w1 w2 w3 ... ... wm

w1 w2 w3 ... ... wm
w1′ w2′ w3′ ... ... ′
wm
∆′ (z) = ... ... ... ......
(m−2) (m−2) (m−2) (m−2)
w1 w2 w3 ... ... wm
(m) (m) (m) (m)
w1 w2 w3 ... ... wm
(m) (m−1) (m−2)
Since wi = p1 w i + p2 w i + ...... + pm wi i = 1, 2, ...., m, so
Rz
p1 (z)dz
∆′ (z) = p1 (z)∆(z) =⇒ ∆′ (z) − p1 (z)∆(z) = 0 =⇒ ∆(z) = ∆(z0 ) e z0

Theorem 1.6 :
Let w1 , w2 , ...., wm be m solutions of differential equations in the domain D of any
ordinary point z0 . If ∆(w1 , w2 , ...., wm ) = 0 at a point in D then ∆(w1 , w2 , ...., wm )
vanishes identically in D .

Proof :
Given that w1 , w2 , ...., wm be m solutions of differential equations in the domain D
of any ordinary point z0 . By Theorem 1.5, we have
Rz
p1 (z)dz
∆(z) = ∆(z0 ) e z0

If ∆(z0 ) = 0 then ∆(z) ≡ 0 ∀z ∈ D. Since z0 is any ordinary point so the proof

follows.

Theorem 1.7:
If w1 , w2 , ...., wm are m linearly independent solutions of (1) in the domain D of an
ordinary point then ∆(w1 , ...., wm ) 6= 0 in D.

6
 Definition 1.4 :
The linearly independent set of m solutions of (1) is called Fundamental Set or sys-
tem of solution.

Theorem 1.8:
A necessary and sufficient condition for a set of m solutions w1 , w2 , ...., wm of (1) in
domain D of ordinary point z0 to a fundamental set is that ∆(w1 , w2 , ...., wm ) 6= 0
at z0 .

Theorem 1.9:
Let w1 , w2 , ....., wm be a fundamental set of solution of (1) in domain D of an ordinary
poit z0 . If w be any solution then we can find constants c1 , c2 , ....., cm s.t.

w = c1 w1 + c2 w2 + ........... + cm wm in D

Proof :
Since w1 , w2 , ....., wm form a fundamental set of solution in D, with w1 , w2 , ...., wm
prescribed at z0 and ∆(w1 , w2 , ...., wm ) 6= 0 at z = z0 so the constants c1 , c2 , ..., cm
can be found uniquely s.t.

c1 w1 (z0 ) + c2 w2 (z0 ) + ............... + cm wm (z0 ) = w(z0 )

c1 w1′ (z0 ) + c2 w2′ (z0 ) + ............... + cm wm

′
(z0 ) = w ′ (z0 )

.............................................................
(m−1) (m−1) (m−1)
c1 w 1 (z0 ) + c2 w2 (z0 ) + ............... + cm wm (z0 ) = w (m−1) (z0 )

Hence w and c1 w1 + c2 w2 + ........... + cm wm are both solutions corresponding to

same set of initial value at z0 . Hence by Theorem 1.2

w = c1 w1 + c2 w2 + ........... + cm wm in D

Definition 1.5 :
If w1 , w2 , ....., wm form a fundamental set of solutions of (1) and c1 , c2 , ....., cm are

7
arbitrary constants then w = c1 w1 + c2 w2 + ........... + cm wm will be called general
solution of (1).

Theorem 1.10:
Any set of n (> m) solutions of (1) in domain of ordinary point is linearly dependent.

Special Fundamental Set

1. The set of m analytic solutions w1 , w2, ...., wm of (1) in domain D of ordinary

point z0 subject to following sets of initial values at z0

w1 (z0 ) = 1 ; w1′ (z0 ) = w1′′ (z0 ) = ........ = w1m−1 (z0 ) = 0

w2 (z0 ) = 0 ; w2′ (z0 ) = 1 ; w2′ (z0 ) = w2′′′ (z0 ) = ............. = w2m−1 (z0 ) = 0

.....................................................................
′ m−2 m−1
wm (z0 ) = 0 ; wm (z0 ) = 0 ; .....................wm (z0 ) = 0 ; wm (z0 ) = 1

for which ∆(w1 , w2, ...., wm ) = 1 at z0 forms a fundamental set of solution.

Now if w(z) is any analytic solution then

w(z) = w(z0 )w1 (z) + w ′(z0 )w2 (z) + .............. + w (m−1) (z0 )wm (z)

2. If w1 (z) is a particular non-zero solution of (1) in domain D of an ordinary point

z0 then we can proceed to construct the special fundamental set of solutions in D
as follows. For convenience let us consider the case m = 3. So

d3 w d2 w dw
= p 1 (z) + p 2 (z) + p3 (z)w
dz 3 dz 2 dz
R
Set w = w1 vdz, then
Z Z Z
′ ′ ′′ ′′ ′ ′ ′′′ ′′′
w = w1 vdz+vw1 ; w = w1 vdz+2w1 v+w1 v ; w = w1 vdz+3w1′′ v+3w1′ v ′ +w1 v ′′

8
So the above third order differential equation reduces to a second order differential
equation
d2 v dv
2
= q1 (z) + q2 (z)v
dz dz
3w1′ 3w ′′ 2w ′
where q1 (z) = p1 − w1 , q2 (z) = p2 − w11 + p1 w11 .
R
Let v1 be a particular non-zero solution of this equation and setting v = v1 udz,
then this equation reduces to 1st order equation
du 2v ′
= r(z)u where r = q1 − 1
dz v1
R
The particular non-zero solution of this equation is u1 = e r(z)dz .
R R R
Thus w1 , w1 v1 dz, w1 (v1 u1 dz)dz are the solutions of given third order equa-
tion . They are linearly independent solutions, for, if c1 , c2 , c3 are constants s.t. in
D Z Z Z
c1 w1 + c2 w1 v1 dz + c3 w1 v1 ( u1 dz)dz = 0
R R R
then c1 + c2 v1 dz + c3 v1 ( u1 dz)dz = 0 in D except possibly the zeros of w1
in D. But since L.S. is holomorphic in D the equations hold everywhere in D.
Differentiating we get
Z
c2 v1 + c3 v1 u1 dz = 0 in D

Hence arguing as above

Z
c2 + c3 u1 dz = 0 in D

Again differentiating

c3 u1 = 0 in D

Since u1 is non-zero , so c3 = 0 . Hence c2 = = c1 = 0 in D.

This prove the assertion.
Again let us consider the case m = 2. If w1 is a particular non-zero solution of the
second order differential equation
d2 w dw
2
= p1 (z) + p2 (z)w
dz dz
R
R qdz 2w1′
in the domain of an ordinary point then w1 and w1 e dz where q = p1 − w

form a fundamental set of solution.