Graded cluster expansion for lattice systems (Q2575381)

In this paper plus their follow-up paper in [Commun. Math. Phys. 261, No. 2, 323--378 (2006)], the authors develop the multiscale cluster expansion techniques which were invented by Froehlich and Imbrie to study the Griffiths phase, and apply them to study properties of renormalization group maps. In particular they prove ``weak Gibbsianness'' of renormalized measures, also in cases where these are known not to be Gibbs measures proper. A difference with the Froehlich-Imbrie situation is that the first scale already can be large. The authors obtain close to optimal results in that they can handle the whole uniqueness region. Another follow-up paper with applications to the study of the Griffiths phase is also announced. I found the paper technically impressive. The approach they develop can be applied to various questions and promises to be fruitful, even if sometimes somewhat painful. Somewhat related methods in fact have independently been pursued by the reviewer, in collaboration with \textit{K. Netocný} and \textit{H. G. Schaap}, in a paper in [J. Stat. Phys. 118, No. 5--6, 997--1056 (2005; Zbl 1073.82022)], but with the purpose of analyzing random boundary conditions.