Aksum University

College of Natural & Computational Sciences

Department of Mathematics
Ordinary Differential Equations assignment-1

Name: Sahle Weldemichael Tesfagiorgis

ID No: AKU1300003

2013 Geez Calendar.

 ∂f
1. Let D ⊆ Rn be open set. If f : D → R is a function and exists and continuous on D , then
∂x
prove that f is locally Lipschitz on D.
Proof:
Let x0 ∈ D. Since D is open , there exists δ > 0 such that
Bδ (x0 ) = {x ∈ D : |x − x0 | < δ} ⊆ D.
Choose  > 0 small such that the compact set
B (x0 ) = {x ∈ D : |x − x0 | ≤ } ⊆ D.
∂f ∂f
Since is continuous on D , it is continuous on B (x0 ). In addition, since Bδ (x0 ) is compact ,
∂x ∂x
∂f
is uniformly continuous on Bδ (x0 ). So, is bounded on B (x0 ). It follows that, ∃M ∈ (0, ∞) such
∂x
that  
 ∂f 
 (x) ≤ M for all x ∈ B (x0 ).
 ∂x 
Claim :

(a) B (x0 ) is convex set (b) f is Lipschitz on B (x0 ).

(a) Let x, y ∈ B (x0 ) and z = x + t(y − x) , 0 ≤ t ≤ 1. Then
|z − x0 | = ||x + t(y − x) − x0 | = |x + t(y − x) − x0 + tx0 − tx0 |
= |(1 − t)(x − x0 ) + t(y − x0 )|
≤ (1 − t)|x − x0 | + t|y − x0 | ≤ (1 − t) + t =  ⇒ z ∈ B (x0 ).
Therefore, B (x0 ) is convex set.
(b) Define a function g : [0, 1] → R by
g(t) = f (a + t(b − a)) , a, b ∈ B (x0 )
∂f (z)
Let z = a + t(b − a) , 0 ≤ t ≤ 1. Then Since z is differentiable on [0, 1] and exists at
∂x
z ∈ B (x0 ) , g is differentiable on [0, 1] by chain rule and
∂ dz ∂
g 0 (t) = f (z) = f (z) (b − a).
∂(x) dt ∂(x)
The existence of g 0 on [0, 1] implies that g is continuous on[0, 1]. Thus,by mean value Theorem ,
∃s ∈ (0, 1) such that
∂f
g(1) − g(0) = g 0 (s) ⇒ f (b) − f (a) = (a + s(b − a)) (b − a)
∂x 
 ∂f 
⇒ |f (b) − f (a)| =  (a + s(b − a)) (b − a)
∂x
 
 ∂f 
≤ 
 (a + s(b − a)) |b − a| ≤ M |b − a| for all a, b ∈ B (x0 )
∂x
which implies f is Lipschitz on B (x0 ).


Now, by choosing so small δ , say δ = , we get f is Lipschitz on Bδ (x0 ). Since x0 ∈ D was
2
arbitrary, f is locally Lipschitz on D.

1
2. Let G ⊆ Rm be compact set. Suppose fk : [a, b] → G be continuous on [a, b] for k = 1, 2, 3, · · · and
fk → f uniformly on [a, b]. If g ∈ C ([a, b] × G, Rm ) , then prove that {g (t, fk (t))}∞
k=0 converges
uniformly to g (t, f (t)) on [a, b].

Proof:

By the compactness of G on Rm and the compactness of [a, b] on R, we see that the set
D = G × [a, b]
is compact on Rm+1 . The compactness of D and continuity of g on D implies that g is uniformly
continuous on D. So, for any
 > 0 , ∃δ = δ() > 0
such that
(t1 , x1 ), (t2 , x2 ) ∈ D & |(t2 , x2 ) − (t1 , x1 )| < δ ⇒ |g(t2 , x2 ) − g(t1 , x1 )| < .
In particular, for
t ∈ [a, b], x1 , x2 ∈ G ,
we get
|x2 − x1 | ≤ |(t, x2 ) − (t, x1 )| < δ ⇒ |g(t, x2 ) − g(t, x1 )| < .
Since fk → f uniformly on [a, b] , ∃N ∈ N such that
k ≥ N & t ∈ [a, b] ⇒ |fk (t) − f (t)| < δ.
Hence,
k ≥ N & t ∈ [a, b] ⇒ |fk (t) − f (t)| < δ
⇒ |g (t, fk (t)) − g (t, f (t)) | < .
Hence, {g (t, fk (t))}∞
k=0 converges uniformly to g (t, f (t)) on [a, b].

3. Assume f : [a, b] → R be differentiable with −∞ ≤ a < b ≤ ∞. If f 0 is continuous on (a, b) , show

that f is locally Lipschitz on (a, b).

Proof:

Let x0 ∈ D = (a, b). Since D is open , there exists δ > 0 such that
Bδ (x0 ) = {x ∈ D : |x − x0 | < δ} = (x0 − δ, x0 + δ) ⊆ D.
Choose  > 0 small such that the the convex and compact set
B (x0 ) = {x ∈ D : |x − x0 | ≤ } = [x0 − , x0 + ] ⊆ D.
Since f 0 is continuous on D , it is continuous on B (x0 ). In addition, since Bδ (x0 ) is compact , f 0 is
uniformly continuous on Bδ (x0 ). So, f 0 is has maximum value on B (x0 ). Let
M = max |f 0 (x)| (|f 0 (x)| ≤ M for all x ∈ B (x0 ))
x∈B (x0 )

Define a function g : [0, 1] → R by

g(t) = f (a + t(b − a)) , a, b ∈ B (x0 )

2
 Let y = a + t(b − a) , 0 ≤ t ≤ 1. Then Since y is differentiable on [0, 1] and f is differentiable on
B (x0 ) , g is differentiable on [0, 1] by chain rule and
df (y) dy df (y)
g 0 (t) = = (b − a).
dx dt dx
The continuity off 0 on B (x0 ) and y on [0, 1] implies g 0 is continuous on[0, 1] by composition . Thus,
g 0 is integrable on [0, 1] and
Z 1 Z 1
0
f (b) − f (a) = g(1) − g(0) = g (s) ds = f 0 (x)(b − a) ds , x = a + s(b − a) for 0 ≤ s ≤ 1
0 0

(Note: We can apply the MVT simply. )

1 Z 1
Z 
0
|f 0 (x)| |(b − a)| ds
 
⇒ |f (b) − f (a)| =  f (x)(b − a) ds ≤
0 0
Z 1
≤ M |(b − a)| ds = M |b − a| for all a, b ∈ B (x0 )
0

which implies f is Lipschitz on B (x0 ).


Now, by choosing so small δ , say δ = , we get f is Lipschitz on Bδ (x0 ). Since x0 ∈ D was arbitrary,
2
f is locally Lipschitz on D.
4. Discuss the existence and uniqueness of the solutions of the IVP:
(
3
x0 = sin(tx) + (1 − x2 )
x(0) = 2

Solution:

Let
3
f (t, x) = sin(tx) + 1 − x2 .
Then
f (t, x)
and
∂f
2
(t, x) = t cos(tx) − 6x 1 − x2
∂x
are continuous on
D = (−∞, ∞) × R.
Thus, f is locally Lipschitz in x on D.

So, ∃α > 0 such that the given IVP has unique solution which exists on

I = [−α, α].
But, nothing to say about maximal interval of existence as f is not bounded in x− space Rn .

3
5. Suppose f ∈ (R × Rn , Rn ) and is locally Lipschitz in x on D = R × Rn . Assume f (t, 0) = 0. If φ(t) is
solution of x0 = f (t, x) such that φ(0) 6= 0 , then show that φ(t) 6= 0 for all t.

Proof:

Since f is continuous on
D = R × Rn
and locally Lipschitz in x on D , the IVP :
(
x0 = f (t, x)
x(t0 ) = x0 , x0 ∈ Rn

has unique solution which exists on the maximal interval I0 of existence of the solution , t0 ∈ I. Since
φ(t) is solution of
x0 = f (t, x),
we get
φ(t) = f (t, φ(t)) .
Let I1 be the maximal interval of existence of the solution φ(t). Assume that ∃t1 ∈ I1 , t1 6= 0 such
that
φ(t1 ) = 0.
Then φ(t) is the unique solution of the IVP:
(
x0 = f (t, x)
x(t1 ) = 0.

Since f (t, 0) = 0 , x = 0 is also solution of the IVP:

(
x0 = f (t, x)
x(t1 ) = 0.

By the uniqueness of the solution, we have

x = φ(t) = 0 for all t ∈ I1 .
This implies that φ(0) = 0 which contradicts to the assumption φ(0) 6= 0.

Hence , φ(t) 6= 0 for all t ∈ I1 .