Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The question I have in mind is (see here, page 60, the solution is at page 297):

Assume $f_{n}$ is a sequence of functions from a metric space $X$ to $Y$. Suppose $f_{n}\rightarrow f$ uniformly and has inverse $g_{n}$. Now assume $f$'s inverse $g$ is uniformly continuous on $Y$. Prove that $g_{n}\rightarrow g$ uniformly.

I could not prove it using standard techniques as I do not know how to bound $|g_{n}(y)-g(y)|$ when $n$ becomes very large. The authors argue that the convergence of $g_{n}(y)\rightarrow g(y)$ is similar to $f(g_{n}(y))\rightarrow f(g(y))$ because the mapping by a uniformly convergent function series keeps uniform convergence. Thus they give the following argument that $$d(f(g(y)),f(g_{n}(y)))=d(y,f(g_{n}(y)))\le d(y,f_{n}(g_{n}(y)))+d(f_{n}(g_{n}(y)),f(g_{n}(y)))=d(f_{n}(g_{n}(y)),f(g_{n}(y)))$$

So since $f_{n}\rightarrow f$ uniformly by hypothesis the statement is proved. My question is: Is the step of substituting $|g_{n}(y)-g(y)|$ by $|f(g(y))-f(g_{n}(y))|$ really justified? I could not get the "keep uniform convergence" thing the author is talking about. But I also could not come up with a better proof.

share|cite|improve this question
up vote 3 down vote accepted

If $h_n\colon S\to S'$ converges uniformly to $h$ on $S$ and $\varphi\colon S'\to S''$ is uniformly continuous on $S$, fix $\varepsilon>0$. There is a $\delta>0$ such that if $d_{S'}(x,y)\leq \delta$ then $d_{S''}(\varphi(x),\varphi(y))\leq\varepsilon)$. Now, we use the fact that there is an integer $n_0$ such that for all $n\geq n_0$, we have $$\sup_{x\in S}d_S(h_n(x),h(x))\leq \delta.$$ Then $$\sup_{x\in S}d_S(\varphi(h_n(x)),\varphi(h(x)))\leq\varepsilon.$$ Now, we apply this result with $\varphi=g$ and $h_n:=f\circ g_n$.

share|cite|improve this answer
In translation we have $|f(g_{n})-f(g)|\le \delta$ when $n\ge N$, and $|g(x)-g(y)|\le \epsilon$ when $|x-y|\le \delta$. Thus we have $|g(y)-g_{n}(y)|=|g(f(g(y))-g(f(g_{n}(y))|\le \delta$ once $n\ge N$. But this feels wierd - why would $f(g_{n})$ converges uniformly to $g$? – Bombyx mori Jul 26 '12 at 9:10
Sorry I was confused. I should have written $f(g_{n})$ to $f(g)=I$. – Bombyx mori Jul 26 '12 at 9:31
Davide, where are you..? – Bombyx mori Jul 26 '12 at 9:39
It has been shown that $f\circ g_n$ converges uniformly to $f\circ g$. By what I showed, $g\circ f\circ g_n$ converges uniformly to $g\circ f\circ g$, that is $g_n$ converges uniformly to $g$. – Davide Giraudo Jul 26 '12 at 9:45
Davide, I do not think it is given $f\circ g_{n}$ converges to $I$ uniformly. We only know $f_{n}\rightarrow f$ uniformly and $g_{n}\rightarrow g$, the later's uniformity is skeptical unless we proved it. – Bombyx mori Jul 26 '12 at 9:47

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.