0% found this document useful (0 votes)
46 views2 pages

Implicit Function Theorem

Download as pdf or txt
Download as pdf or txt
Download as pdf or txt
You are on page 1/ 2

Implicit Function Theorem

This document contains a proof of the implicit function theorem.


Theorem 1. Suppose F (x, y) is continuously differentiable in a neighborhood of a point (a, b) ∈ Rn × R
and F (a, b) = 0. Suppose that Fy (a, b) 6= 0. Then there is δ > 0 and  > 0 and a box B = {(x, y) :
kx − ak < δ, |y − b| < } so that
1. For each x such that kx − ak < δ there is a unique y such that |y − b| <  for which F (x, y) = 0.
This correspondence defines a function f (x) on {kx − ak < δ} such that

F (x, y) = 0 ⇔ y = f (x) for (x, y) ∈ B.

2. f is continuous.

3. f is continuously differentiable and


Dx F (x, f (x)
Df (x) = − ,
Fy (x, f (x))

where Df = [fx1 , . . . , fxn ] and Dy F = [Fx1 , . . . , Fxn ].


Proof. 1. Choose δ1 > 0 and 1 > 0 so that Fy (x, y) > 0 for kx − ak < δ1 , |y − b| < 1 . Since F (a, b) = 0
and F (a, y) is strictly increasing in y, F (a, b + 1 /2) > 0 and F (a, b − 1 /2) < 0. Let  = 1 /2 and
choose δ < δ1 so that F (x, b + ) > 0 and F (x, b − ) < 0 if kx − ak < δ. These dimensions define B.
For fixed x with kx − ak < δ, since F (x, b − ) < 0, F (x, b + ) > 0, and F (x, y) is strictly increasing
in y, the intermediate value theorem implies that there is a unique y with |y − b| <  such that
F (x, y) = 0. The uniquely determined y defines a functionf (x). This proves the first statement.

2. We prove that f is continuous at a. Let e > 0 be given. Assume that e <  Then by the proof of
the first statement, there is a d > 0 (we may choose d < δ) so that the uniquely defined f (x) in
{kx − ak < d} satisfies |f (x) − b| < d. This proves continuity at a. We can repeat this argument at
any point (a1 , f (a1 )) ∈ B, proving that f is continuous on {kx − ak < }.

3. By differentiability
X
0 = F (x, f (x)) = F (a, b) + Pj (x, f (x))(x − aj ) + Q(x, f (x))(f (x) − f (a))
j
X
= Pj (x, f (x))(x − aj ) + Q(x, f (x))(f (x) − f (a)),
j

where Pj (x, f (x)), Q(x, f (x)) are continuous at a. Rewrite this as


X
Q(x, f (x))(f (x) − f (a)) = − Pj (x, f (x))(x − aj ).
j

1
Since Q(x, f (x)) is continuous at a and Q(a, (f (a)) = fy (a, b) > 0, Q(x, f (x)) > 0 for x near a and
we can divide by it to get
X Pj (x, f (x))
f (x) = f (a) + − (x − aj ).
Q(x, f (x))
j

Pj (x, f (x))
Each term is continuous at a so f is differentiable at a. Moreover
Q(x, f (x))

Fj (a, b)
fxj (a, b) = − .
Fy (a, b)

You might like this bad notation:


∂y ∂Fxj
=− .
∂xj ∂Fy

You might also like