Riesz transforms, Hodge-Dirac operators and functional calculus for multipliers I

Abstract

In this work, we solve the problem explicitly stated at the end of a paper of Junge, Mei and Parcet [JEMS2018, Problem C.5] for a large class of groups including all amenable groups and free groups. More precisely, we prove that the Hodge-Dirac operator of the canonical "hidden" noncommutative geometry associated with a Markov semigroup $(T_t)_{t \geq 0}$ of Fourier multipliers is bisectorial and admits a bounded $\mathrm{H}^\infty$ functional calculus on a bisector which implies a positive answer to the quoted problem. Our result can be seen as a strengthening of the dimension free estimates of Riesz transforms of the above authors and also allows us to provide Hodge decompositions. A part of our proof relies on a new transference argument between multipliers which is of independent interest. Our results are even new for the Poisson semigroup on $\mathbb{T}^n$. We also provide a similar result for Markov semigroups of Schur multipliers and dimension free estimates for noncommutative Riesz transforms associated with these semigroups. Along the way, we also obtain new Khintchine type equivalences for $q$-Gaussians in $\mathrm{L}^p$-spaces associated to crossed products. Our results allow us to introduce new spectral triples (i.e. noncommutative manifolds) and new quantum (locally) compact metric spaces, in connection with the carr\'e du champ, which summarize the underlying geometry of our setting. Finally, our examples lead us to introduce a Banach space variant of the notion of spectral triple suitable for our context.