8
$\begingroup$

Glaeser's Theorem says that a $C^\infty$ function $F$ on $\mathbb R^n$ which is invariant under permutation of the variables is a smooth function of the symmetric polynomials of $(x_1, \dots, x_n)$.

Question 1: What remains (if anything) of this statement if $F$ is $C^k$ ?

Question 2: In the statement above, is it clear that you can replace $\mathbb R^n$ by a symmetric open subset of $\mathbb R^n$?

Improve this question
$\endgroup$
6
  • 1
    $\begingroup$ Can you provide a reference for Gleaser's theorem? $\endgroup$
    – Piotr Hajlasz
    Apr 24, 2020 at 12:46
  • 2
    $\begingroup$ For 1), take $F(x_1,\ldots ,x_n)=\sum |x_i|^k$ with $k$ odd. $\endgroup$
    – abx
    Apr 24, 2020 at 13:20
  • 1
    $\begingroup$ jstor.org/stable/1970204?origin=crossref&seq=1 $\endgroup$
    – Robert Israel
    Apr 24, 2020 at 15:26
  • $\begingroup$ @abx Your example is $C^{k-1}$ and invariant under permutation. Is it not a $C^{k-1}$ function of the symmetric polynomials? In the case $n=2$, $k=1$, $|x_1| + |x_2| = \sqrt{(x_1+x_2)^2 - 2 x_1 x_2 + 2 |x_1 x_2|}$. Or are you just saying it's not a smooth function? $\endgroup$
    – Robert Israel
    Apr 24, 2020 at 15:38
  • $\begingroup$ @Robert Israel: maybe I was too hasty. Do you think you could do that for any $k$? $\endgroup$
    – abx
    Apr 24, 2020 at 16:06

2 Answers 2

Reset to default
9
$\begingroup$

A more direct reference (quoted in Rumberger's paper) is: G. Barbançon, Le théorème de Newton pour les fonctions de classe $C^r$. Ann. Sci. École Norm. Sup. 5 (1972), 435–458. He proves that a symmetric function of class $C^{nr}$ on $\Bbb{R}^n$ is a function of class $C^{r}$ of the symmetric polynomials.

Improve this answer
$\endgroup$
8
$\begingroup$

My collegue Armin Rainer pointed me to the following paper, which has some answers, also in its references. There is a loss of differentiability involved.

  • Matthias Rumberger: Finitely differentiable invariants. Math. Z. 229, 675–694 (1998)
Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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