In Velenik's Statistical Mechanics of Lattice Systems, Exercise 6.22 claims that if $\pi =\{ \pi_\Lambda:\Lambda \Subset\mathbb{Z}^d\}$ is a translationally invariant specification with nonempty Gibbs state $\mathscr{G}(\pi)$ (probability measures compatible with $\pi_\Lambda$), then the set of translationally invariant Gibbs state is nonempty.

Velenik provides a hint in which we take $\mu\in \mathscr{G}(\pi)$ and use $$ \mu_n = \frac{1}{|B(n)|} \sum_{j\in B(n)} \theta_j \mu $$

I would assume that we are to take the limit (passing under subsequence by Banach-Alaoglu) and obtain a vague convergence probability measure, but it seems that this proof requires that $\mathscr{G}(\pi)$ be closed under the vague topology, which may not necessarily be true. Is there another way to prove the claim? Or is the statement missing a requirement?


1 Answer 1


You are completely correct. We should have added the assumption that the specification $\pi$ is quasilocal. (For instance, using the setting described right before the exercise; I guess that's what we had in mind when stating the exercise, but I don't really remember. Of course, in that case, one can remove the assumption that $\mathcal{G}(\pi)\neq\emptyset$ by Theorem 6.26.)

With the quasilocality assumption, $\mathcal{G}(\pi)$ is closed by Lemma 6.27 and the exercise is proved as described in the book (solutions are given in Appendix C).

I have updated the errata and corrected the "preprint" version of the book. (Thanks for pointing this out!)

(By the way, if you are interested in the corresponding result in a much more general framework than we cover in our book, you should have a look at Corollary 5.16 in Georgii's book.)

  • $\begingroup$ Thank you for the recommendation! I have seen Georgii's book before, but it is a little too general for me at the moment. $\endgroup$ Sep 30, 2020 at 21:38
  • 2
    $\begingroup$ @AndrewYuan The fact that Georgii's book is too abstract to learn the theory from it is one of the reasons we decided to write our book: there was a need for an introductory book on this subject. ;) $\endgroup$ Oct 1, 2020 at 6:03

Your Answer

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

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