MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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

Let's consider an interval $I\subseteq\mathbb R$, and let $\mathcal F(I)$ be the set of bijective functions $f:I\to I$ so that the graph of $f$ is a analytic curve in $I\times I$.

The set $\mathcal F(I)$, together with the function composition $\circ$, is a group. Let $\mathcal H(I)$ be its subgroup consisting in all analytic functions from $\mathcal F(I)$ having the inverse an analytic function (here and everywhere in this question by "inverse" and $f^{-1}$ we mean inverse w.r.t. function composition).

For example, if $I=\mathbb R$, the function $f(x)=x^3$ is invertible, and its graph is an analytic curve in $I\times I$, but $x^3\notin \mathcal H(I)$.

Q1. What do we know about the cosets of $\mathcal H(I)$ in $\mathcal F(I)$? Do you know some references?

Let $Q\in I$ and let $\mathcal H_Q(I)\subset\mathcal F(I)$ so that $f(Q)=Q$ and both $f$ and $f^{-1}$ are analytic on $I-Q$. For example, the function $x^3\in\mathcal H_0(\mathbb R)$, but $x^3\notin\mathcal H(\mathbb R)$. Another example is $x^2$, where $I=[0,+\infty)$ and $Q=0$.

The set $\mathcal H_Q(I)$ is a subgroup of $\mathcal F(I)$, and $\mathcal H(I)\cap \mathcal H_Q(I)$ is a subgroup of $\mathcal H_Q(I)$.

Q2. What do we know about the cosets of $\mathcal H(I)\cap \mathcal H_Q(I)$ in $\mathcal H_Q(I)$? Do you know some references?

What about the case when the graph is not necessarily analytic, but only smooth?

share|cite|improve this question
Is "$f$ is analytic" the same as "the graph of $f$ is an analytic curve"? If so, then $H=F$, since a group contains the inverse of each of its elements. If no, then what is the difference? – Gerry Myerson Feb 18 '12 at 8:14
@Gerry Myerson: For example, the function $f(x)=x^{1/3}$ is not analytic, although its graph is an analytic curve in $\mathbb R\times \mathbb R$. – Cristi Stoica Feb 18 '12 at 9:47
$\mathcal{H}(I)$ is not necessarily a subgroup of $\mathcal{H}_Q(I)$, since for example, $x+1\in\mathcal{H}(\mathbb{R})$ but not in $\mathcal{H}_0(\mathbb{R})$. – Steve D Feb 18 '12 at 15:28
@Steve D: Thanks for pointing this out, I updated the question accordingly. – Cristi Stoica Feb 18 '12 at 19:01
This is hardly worth mentioning, but perhaps in the definition of $\mathcal{H}(I)$ the member functions themselves (in addition to their inverses) should be required to be analytic. – Tiankai Mar 4 '12 at 5:35
up vote 1 down vote accepted

Let $f$ be a function in this group. $f$ fails to be analytic at the points where its "derivative" is infinite, and this is a finite set. Similarly $f^{-1}$ is not analytic where its derivative is $0$. So count the number of points where $f$ or $f^{-1}$ is analytic, and list them in order. This should be finite. Furthermore, list at each point the order of vanishing of the zero of the derivative of $f$, or the order of vanishing of the zero of the derivative of $f^{-1}$.

This seems to be sufficient information to determine the coset. Suppose we have two functions with the same list. We can multiply one on the right to move the non-analytic points to the same $x$ coordinate as the non-analytic points of the other. We then look at $fg^{-1}$. This is clearly analytic away from those points, and it is also analytic at those points - indeed, we can calculate its derivative as the nonzero finite ration of the $n$th derivatives of $f$ and $g$.

Similarly, two functions in the same coset must have the same list, up to reversal of the order.

This gives a complete description of the cosets.

Smoothness seems like it might be harder.

share|cite|improve this answer
Yes, that's it! Thank you very much for your time and for considering my question. – Cristi Stoica Mar 15 '12 at 4:20

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.