Prove Differentiability of g(x) at x1: mx1+b=x0

  • Level: Graduate 
  • Thread starter Thread starter John O' Meara
  • Start date Start date
  • Tags Tags
    Proof
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
5 replies · 2K views
John O' Meara
Messages
325
Reaction score
0
Suppose that a function f is differentiable at x0 and define g(x)=f(mx+b), where m and b are constants. Prove that if x1 is a point at which mx1+b=x0, then g(x) is differentiable at x1 and g'(x1)=mf'(x0).
Definition: A function f is said to be "differentiable at x0" if the limit
[tex]f'(x_{0})= lim_{h \rightarrow 0} \frac{f(x_{0}+h)-f(x_{0})}{h}[/tex]
exists. If f is differentiable at each point of the open interval(a,b), then we say that f is "differentiable on (a,b)", and similarily for open intervals of the form [tex](a,+\infty),(-\infty,b)and (-\infty,+\infty)[/tex]. In the last case we say f is "differentiable everywhere". I think this is a simple proof, but since I am doing this no my own with a book that doesn't do too many example proofs, I am lost. g(x1) = f(mx1+b) = f(x0), which I think proves the limit of g(x1) exists but g'(x1)=? Thanks for the help.
 
Physics news on Phys.org
I think they want you to do it from the definition. I.e. show that the limit of
[tex] \frac{g(x_{1}+h)-g(x_{1})}{h}[/tex]
for h to zero exists by substituting the definition of g and using that the appropriate limit for f exists.
 
Sorry, if I gave the wrong impression, but the definition I included myself thinking it relevant. The definition was not part of the original question.
 
If you don't want to use the definition, you can directly prove if from the chain rule, I suppose.

I checked the proof from the definition, it works out nicely. Just remember to bring the limit you get back in the form
[tex]\lim_{\delta \to 0} \frac{f(x_0 + \delta) - f(x_0)}{\delta}[/tex]
where [itex]\delta[/itex] is some new quantity depending on h.

(I cannot really give you more hints without doing the calculation for you, so I suggest you give this a try and post what you get).
 
[tex]\lim_{h \rightarrow 0} \frac{ g(x_{1} +h) - g(x_{1})}{h} = \lim_{h \rightarrow 0} \frac{f(m(x_{1}+h)+b) - f(mx_{1} + b)}{h}[/tex]
[tex]= \lim_{h \rightarrow 0} \frac{ f(mx_{1} + mh + b) - f(mx_{1} + b)}{h}[/tex]
[tex]= \lim_{mh \rightarrow 0} m\frac{ f(mx_{1}+mh+b)-f(mx_{1}+b)}{mh}[/tex]
[tex]= \lim_{mh \rightarrow 0} m\frac{ f(mx_{1}+b+mh)-f(mx_{1}+b)}{mh}[/tex]
[tex]= m \lim_{\delta \rightarrow 0} \frac{ f(x_{0}+ \delta)-f(x_{0})}{\delta}[/tex]
[tex]= mf'(x_{0})[/tex]

If this is ok, now I got to use this result to prove that [tex]f'(x)=nm(mx+b)^{n-1}[/tex] if [tex]f(x) = (mx+b)^{n}[/tex], where m and b are constants and n is any integer. I know that [tex]\frac{dx^n}{dx}=nx^{n-1}[/tex] where n is any integer or do I use the definition given above in #1? Thanks again.
 
Last edited:
Very well.

The quickest way indeed seems to be, to use that
[tex]\frac{d}{dx}f(x) = n x^{n-1}[/tex]
where f(x) = xn is differentiable anywhere, and use what you have just proven.