Existence and uniqueness: Picard’s theorem

First-order equations

Consider the equation y 0 = f (x, y) (not necessarily linear). The equation dictates
a value of y 0 at each point (x, y), so one would expect there to be a unique solution curve
through a given point. We will prove the following theorem:

THEOREM 1. Let I be a real interval (possibly R itself ), and let f be a continuous real
function on I × R, satisfying the “Lipschitz” condition

|f (x, y2 ) − f (x, y1 )| ≤ K|y2 − y1 |

on this set. Let x0 ∈ I. Then there is a unique function y on I such that

y 0 (x) = f [x, y(x)] for x ∈ I

and y(x0 ) = y0 .

Note. By the mean-value theorem, the Lipschitz condition will be satisfied if we

∂
have | ∂y f (x, y)| ≤ K on I × R.

LEMMA 1. It is enough to prove the result for the case x0 = y0 = 0.

Proof. Suppose the result is known for this case and x0 , y0 are given. Let g(x, y) =
f (x + x0 , y + y0 ). By assumption, there is a unique function z such that

z 0 (x) = g[x, z(x)] when x + x0 ∈ I

and z(0) = 0. Let y(x) = z(x − x0 ) + y0 . Then y(x0 ) = y0 and for x ∈ I,

y 0 (x) = z 0 (x − x0 )
=
=
=

The next step is to show that the required properties of y can be equally expressed by
an integral.

LEMMA 2. Let I be an interval containing 0. For a differentiable function on I, the

following statements are equivalent:

1
 (i) y 0 (x) = f [x, y(x)] for x ∈ I and y(0) = 0;
Z x
(ii) y(x) = f [t, y(t)] dt for x ∈ I.
0

Proof. (i) ⇒ (ii).

Let C(I) be the space of all continuous functions on I. We define a mapping A from
C(I) to itself as follows: given y ∈ C(I),
Z x
(Ay)(x) = f [t, y(t)] dt.
0

By Lemma 2, y is the required solution if and only if A(y) = y.

We take

y0 = 0, y1 = A(y0 ), y2 = A(y1 ), . . . , yn = A(yn−1 ) = An (y0 ).

We will show that (yn ) converges (uniformly on I) to a limit function y. One would then
expect that A(y) = limn→∞ A(yn ) = limn→∞ yn+1 = y, as required.

LEMMA 3. Suppose that |u(x) − v(x)| ≤ C for x ∈ I. Then for each n ≥ 1,

(K|x|)n
|(An u)(x) − (An v)(x)| ≤ C for x ∈ I.
n!

Proof. By the Lipschitz condition, |f [t, u(t)] − f [t, v(t)]| ≤ CK for t ∈ I, so

Z x 
 
|(Au)(x) − (Av)(x)| =   [f (t, u(t)) − f (t, v(t))] dt
0
≤ CK|x|,

i.e. the statement holds for n = 1. Now suppose it holds for a certain n. Then, by the
Lipschitz condition,

(K|t|)n
|f [t, (An u)(t)] − f [t, (An v)(t)]| ≤ KC .
n!
So (for x > 0),
x
Z 
 n+1 n+1
  n

n
(A u)(x) − (A v)(x) =  {f [t, (A u)(t)] − f [t, (A v)(t)]} dt

0

2
 (Similarly for x < 0, with |x| instead of x.)

Proof of the Theorem. For the moment, assume that I is a bounded, closed interval.
Then |x| ≤ R (say) for x ∈ I, and continuous functions are bounded on I, so there exists M
such that |f (t, 0)| ≤ M for t ∈ I.

Uniqueness. Suppose that Au = u and Av = v. Then An u = u and An v = v for all

n. Also, there exists C such that |u(x) − v(x)| ≤ C for x ∈ I. By Lemma 3,

(K|x|)n
|u(x) − v(x)| ≤ C for all n ≥ 1.
n!
But for each x, (K|x|)n /n! → 0 as n → ∞. Hence u(x) = v(x).

Existence. Let y0 = 0, y1 = A(y0 ), y2 = A(y1 ), . . . , yn = An (y0 ), . . .. Then

Z x 
 
|y1 (x) − y0 (x)| = |y1 (x)| = 
 f (t, 0) dt ≤ M |x| ≤ C
0

for x ∈ I, where C = M R. By Lemma 3,

|yn+1 (x) − yn (x)| ≤

P∞ n
P∞
for x ∈ I. Now n=1 (KR) /n! is convergent (to eKR ), so by the “M -test”, n=1 (yn+1 −yn )
is uniformly convergent on I (say to y). But

(y1 − y0 ) + (y2 − y1 ) + · · · + (yn − yn−1 ) = yn − y0 = yn ,

so this says that yn → y uniformly on I. In other words, |yn (x) − y(x)| ≤ δn for x ∈ I,
where δn → 0 as n → ∞. By the case n = 1 in Lemma 3,

|yn+1 (x) − (Ay)(x)| = |(Ayn )(x) − Ay(x)| ≤ K|x|δn

for x ∈ I, so
(Ay)(x) = lim yn+1 (x) = y(x)
n→∞

for x ∈ I, as required.

3
 Finally, suppose that I is an open or unbounded interval. Then we can express it as
∪∞
n=1 In , where I1 ⊂ I2 ⊂ I3 ⊂ . . . are bounded, closed intervals with 0 ∈ I1 . By the above,
there is for each n a unique solution y(n) on In with y(n) (0) = 0. By this uniqueness, y(n+1)
agrees with y(n) on In , so it is consistent to define y on I by:

y(x) = y(n) (x) for x ∈ In .

Clearly, y satisfies the equation on I. 

We now give two quite simple examples to show that both parts of the theorem can
fail if the Lipschitz condition is not satisfied.

Example 1. Consider the equation y 0 = −y 2 , with y(1) = 1, on the interval [−1, 1].
Solving by elementary methods, we have
1 0
− y = 1,
y2
1
= x + c.
y
The condition y(1) = 1 selects the solution y = 1/x. This is clearly the unique solution
on (0, 1]. It cannot be extended to a differentiable function at 0, so there is no solution on
[−1, 1]. To see that the Lipschitz condition fails, note that f (x, y) = −y 2 . We have

f (x, y + 1) − f (x, y) = −(y + 1)2 + y 2 = −2y − 1,

which is unbounded on R.

Example 2. Consider y 0 = 3y 2/3 with y(0) = 0, on I = R.

Clearly, 0 is one solution. Elementary methods give

1 0
y = 1,
3y 2/3
y 1/3 = x + c,
y = (x + c)3 .

Hence x3 is another solution with y(0) = 0. There are actually infinitely many solutions: for
each c > 0, a solution is 
(x − c)3 for x ≥ c
yc (x) =
0 for x < c.
(Then yc is differentiable at c, with derivative 0). The Lipschitz condition fails because

y 2/3 − 0 1
= 1/3 → ∞ as y → 0+ .
y−0 y

4
 Picard’s iteration. The proof of Picard’s theorem provides a way of constructing
successive approximations to the solution. With the initial condition y(x0 ) = y0 , this means
we define y0 (x) = y0 and
Z x
yn (x) = (Ayn−1 )(x) = y0 + f [t, yn−1 (t)] dt.
x0

Example. Consider the equation y 0 = y with y(0) = 1. Of course, the solution is ex .

Picard’s iteration gives the following: f (x, y) = y, hence

y0 (x) = 1,
Rx
y1 (x) = 1 + 0
f (t, 1) dt =

Example. The equation y 0 = 2x + y 2 , with y(0) = 0. Start with y0 (x) = 0. Then

Systems of equations and higher order equations

Now consider a system of first-order equations

yj0 (x) = fj [x, y1 (x), . . . , yn (x)] (1 ≤ j ≤ n).

In vector notation, y(x) = [y1 (x), . . . , yn (x)], and the system becomes

y 0 (x) = f [x, y(x)].

By a vector adaptation of the proof, we shall obtain a generalization of Theorem 1 to systems

of equations. As we have seen before, higher order equations can be seen as a special case
of a system.

For y = (y1 , y2 , . . . , yn ) ∈ Rn , write kyk = max |yj |. This is a “norm” on Rn , giving

kx − yk as a notion of generalized “distance” between x and y. Note that kx − yk ≤ M
simply means that |xj − yj | ≤ M for each j.

5
 Remark. Another norm is kyk2 = ( nj=1 yj2 )1/2 . The two norms are “equivalent” in
P
√
the sense that kyk ≤ kyk2 ≤ nkyk for all y. This means that either can be used to
determine convergence. Though k k2 is arguably the “true” measure of distance, it is more
complicated to use in proofs like those that follow. Our kyk is often denoted by yk∞ .

Our theorem is:

THEOREM 2. Suppose that each fj is continuous on I × Rn and

|fj (x, y) − fj (x, z)| ≤ Kky − zk

on I × Rn . Suppose that x0 ∈ I and real numbers α1 , . . . , αn are given. Then there is a

unique vector function y(x) such that y 0 (x) = f [x, y(x)] for x ∈ I and yj (x0 ) = αj for
each j.

As before, it is sufficient to consider the case where x0 = αj = 0 for all j.

Exactly as in Lemma 2, the required conditions are equivalent to

Z x
yj (x) = fj [t, y(t)] dt (j = 1, . . . , n),
0

or in vector notation Z x
y(x) = f [t, y(t)] dt.
0
Given y = (y1 , . . . , yn ), where each yj is in C(I), define Ay to be z, where
Z x
z(x) = f [t, y(t)] dt.
0

We require y such that Ay = y.

LEMMA 30 . Let u, v be vector functions such that ku(x) − v(x)k ≤ C for x ∈ I.

Then for each k ≥ 1,
(K|x|)k
k(Ak u)(x) − (Ak v)(x)k ≤ C for x ∈ I.
k!

Proof. Just like Lemma 3, with the notation adjusted. We describe the induction step.
Assume the statement holds for a certain k. Then, by the Lipschitz condition, for each j:
k
fj t, (Ak u)(t) − fj t, (Ak v)(t)  ≤ KC (K|t|) .
    
k!
Write Ak+1 u = (g1 , . . . , gn ) and Ak+1 v = (h1 , . . . , hn ). This means that
Z x
gj (x) = fj [t, (Ak u)(t)] dt
0

6
and similarly for hj (x) with v instead of u. Exactly as in Lemma 3, we deduce that for
x ∈ I,
(K|x|)k+1
|gj (x) − hj (x)| ≤ C (j = 1, . . . , n),
(k + 1)!
which is equivalent to

(K|x|)k+1
k(Ak+1 u)(x) − (Ak+1 v)(x)k ≤ C .
(k + 1)!

Theorem 2 now follows exactly as in the case n = 1. Let y (0) = 0 and y (k) =
(k) (k)
(y1 , . . . , yn ) = Ak (0). (Here the notation y (k) is being used temporarily for the kth function
(k)
of the sequence, not the kth derivative.) As before, we see that for each j, the functions yj
tend to a limit function yj , uniformly on I, as k → ∞ (in other words, y (k) → y uniformly
on I), and Ay = y.

THEOREM 3. Suppose that f is continuous on I × Rn and

|f (x, y) − f (x, z)| ≤ Kky − zk

on I × Rn . Suppose that x0 ∈ I and real numbers α1 , . . . , αn are given. Then there is a

unique function y on I such that

y (n) (x) = f [x, y(x), y 0 (x), . . . , y (n−1) (x)]

for x ∈ I and y(x0 ) = α0 , y (j) (x0 ) = αj for 1 ≤ j ≤ n − 1.

Proof. The equation is equivalent to the system

yj0 = yj+1 for 0 ≤ j ≤ n − 2,

0
yn−1 = f (x, y0 , y1 , . . . , yn−1 ).

In the notation of Theorem 2 (with j now ranging from 0 to n − 1), we have fn−1 = f , while
for 0 ≤ j ≤ n − 2,
fj (x, y0 , y1 , . . . , yn−1 ) = yj+1 .

We only need to verify that the functions fj satisfy the Lipschitz condition. This is trivial
for 0 ≤ j ≤ n − 2, and ensured by our hypothesis for j = n − 1. 

Finally, we specialize this to linear equations, showing that the Lipschitz condition is
then automatically satisfied.

7
 THEOREM 4. Suppose that the functions p0 , p1 , . . . , pn−1 and g are continuous on an
interval I. Let x0 ∈ I and real numbers α1 , . . . , αn be given. Then there is a unique function
y such that

y (n) (x) = p0 (x)y(x) + p1 (x)y 0 (x) + · · · + pn−1 (x)y (n−1) (x) + g(x)

for x ∈ I and y(x0 ) = α0 , y (j) (x0 ) = αj for 1 ≤ j ≤ n − 1.

Proof. First assume that I is a bounded, closed interval (then extend to open and
unbounded intervals as before). Then each pj is bounded on I: say |pj (x)| ≤ K for x ∈ I.
Apply Theorem 3 with
n−1
X
f (x, y0 , y1 , . . . , yn−1 ) = pj (x)yj + g(x).
j=0

Since it is a sum of products of continuous functions of single variables, f is continuous on

I × Rn . The Lipschitz condition is satisfied, because:
 
Xn−1  n−1
X
|f (x, y) − f (x, z)| =  pj (x)(yj − zj ) ≤ K |yj − zj | ≤ nKky − zk. 
 
 
j=0 j=0