Characterizations of the Ricci flow

Robert Haslhofer*, Aaron Naber

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

18 Scopus citations

Abstract

This is the first of a series of papers, where we introduce a new class of estimates for the Ricci flow, and use them both to characterize solutions of the Ricci flow and to provide a notion of weak solutions to the Ricci flow in the nonsmooth setting. In this first paper, we prove various new estimates for the Ricci flow, and show that they in fact characterize solutions of the Ricci flow. Namely, given a family (M, gt)t?I of Riemannian manifolds, we consider the path space PM of its space-time M = M × I. Our first characterization says that (M, gt)t?I evolves by Ricci flow if and only if an infinite-dimensional gradient estimate holds for all functions on PM. We prove additional characterizations in terms of the C1/2-regularity of martingales on path space, as well as characterizations in terms of log-Sobolev and spectral gap inequalities for a family of Ornstein–Uhlenbeck type operators. Our estimates are infinite-dimensional generalizations of much more elementary estimates for the linear heat equation on (M, gt)t?I, which themselves generalize the Bakry–Émery–Ledoux estimates for spaces with lower Ricci curvature bounds. Thanks to our characterizations we can define a notion of weak solutions for the Ricci flow. We will develop the structure theory of these weak solutions in subsequent papers.

Original languageEnglish (US)
Pages (from-to)1269-1302
Number of pages34
JournalJournal of the European Mathematical Society
Volume20
Issue number5
DOIs
StatePublished - 2018

Funding

Acknowledgments. R.H. has been supported by NSF grant DMS-1406394, and A.N. has been supported by NSF grant DMS-1406259.

Keywords

  • Path space
  • Ricci flow
  • Wiener measure

ASJC Scopus subject areas

  • General Mathematics
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'Characterizations of the Ricci flow'. Together they form a unique fingerprint.

Cite this