In the following paper (extended version here), at the beginning of section 3, the authors give two axioms about $\mathcal{A}$. Axiom 1 is about $\mathcal{A}$ being an algebra. I do not see where this axiom is being used. Any ideas?

Conference version: Bobkov, Sergey; Tetali, Prasad, Modified log-Sobolev inequalities, mixing and hypercontractivity, Proceedings of the thirty-fifth annual ACM symposium on theory of computing (STOC 2003), San Diego, CA, USA,. New York, NY: ACM Press (ISBN 1-58113-674-9). 287-296, electronic only (2003). ZBL1192.60020.

Journal version: Bobkov, Sergey G.; Tetali, Prasad, Modified logarithmic Sobolev inequalities in discrete settings, J. Theor. Probab. 19, No. 2, 289-336 (2006). ZBL1113.60072.

  • 2
    $\begingroup$ @KenW.Smith: p290, Axiom 1. So I haven’t managed to digest the paper yet, but is it important that $f^2\in \mathcal A$? (I notice that $f^2$ appears in various places; and a linear space of functions is closed under squaring if and only if it’s closed under taking products). $\endgroup$ – Anthony Quas Sep 8 '19 at 19:33
  • 1
    $\begingroup$ @KenW.Smith $\mathcal{A}$ appears in the first paragraph of section 3 ("POINCARE AND LOG-SOBOLEV IN ABSTRACT SETTINGS"). You can easily find it by searching for Axiom 1 using ctrl+F and reading the text above it. $\endgroup$ – Ella Sharakanski Sep 9 '19 at 13:55
  • 2
    $\begingroup$ As a general tip, when you mention a paper, I suggest giving a full citation instead of just a link. It helps people find the paper more reliably. (In particular, your link to the extended version could only be used by people whose institutions participate in OpenAthens.) Also, I feel it is a courtesy to the authors to mention their names and the other info about their paper. There is a "cite" button in the question editor that will give you a full citation with DOI and Zentralblatt link. I've added citations to your post - perhaps you could double check that they are correct. $\endgroup$ – Nate Eldredge Sep 11 '19 at 17:24
  • 1
    $\begingroup$ I also don't immediately see this assumption used explicitly. However, I would note that they are using a very simplified definition of "Dirichlet form". In the more usual development, a Dirichlet form $\mathcal{E}$ is defined on a dense subspace $D(\mathcal{E})$ of $L^2(M,\pi)$, which should be thought of as a space of "finite energy" functions having one "derivative" in $L^2$, like the Sobolev space $W^{1,2}_0$. Then it is a theorem that the bounded functions in $D(\mathcal{E})$, that is $\mathcal{A} := D(\mathcal{E}) \cap L^\infty(M)$, do in fact form an algebra. $\endgroup$ – Nate Eldredge Sep 11 '19 at 17:34
  • 1
    $\begingroup$ The standard reference for Dirichlet forms is Fukushima-Oshima-Takeda, Dirichlet Forms and Symmetric Markov Processes. I am pretty sure this theorem can be found there. $\endgroup$ – Nate Eldredge Sep 11 '19 at 18:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.