What's wrong with this proof of the chain rule

  • Thread starter quasar987
  • Start date
  • #1
quasar987
Science Advisor
Homework Helper
Gold Member
4,773
8

Main Question or Discussion Point

Can anyone see where the flaw is in the developement below, where I prove that (g o f)'(x)=g'(f(x)) instead of g'(f(x))f'(x), as it should be.

Consider the usual hypothese under which the chain rule for real-valued function applies.

Consider [itex] \epsilon>0[/itex]. Since g is differentiable at f(x_0), there exists [itex] \delta_1(\epsilon)>0[/itex] such that [itex]|y-f(x_0)|<\delta_1(\epsilon) [/itex] implies [itex]|g(y)-g(f(x_0))-g'(f(x_0))(y-f(x_0))|\leq\epsilon|y-f(x_0)|[/itex]

On the other hand, f being differentiable at x_0, enjoys the Lipschitz property there. That is to say, there exists positive constants c and M such that [itex]|x-x_0|<c [/itex] implies [itex] |f(x)-f(x_0)|\leq M|x-x_0|[/itex].

So for [itex]\delta=\min(c,\delta_1(\epsilon/M)/M)[/itex], we have that [itex]|x-x_0|<\delta [/itex] implies [itex]|f(x)-f(x_0)|\leq M|x-x_0|<M\delta\leq M\delta_1(\epsilon/M)/M= \delta_1(\epsilon/M)[/itex]

which in turns implies that

[tex]|g(f(x))-g(f(x_0))-g'(f(x_0))(f(x)-f(x_0))|\leq(\epsilon/M) |f(x)-f(x_0)|\leq (\epsilon/M)M|x-x_0| = \epsilon |x-x_0| [/tex]

And this is equivalent to saying that (g o f)'(x) = g'(f(x))...
 
Last edited:

Answers and Replies

  • #2
quasar987
Science Advisor
Homework Helper
Gold Member
4,773
8
Ok, I see it.
 

Related Threads for: What's wrong with this proof of the chain rule

Replies
5
Views
2K
  • Last Post
Replies
4
Views
1K
  • Last Post
Replies
2
Views
2K
  • Last Post
Replies
15
Views
20K
  • Last Post
Replies
2
Views
1K
  • Last Post
Replies
5
Views
1K
  • Last Post
Replies
4
Views
5K
  • Last Post
Replies
8
Views
3K
Top