Image: Íñigo de Amescua/ICMAT

A recent study solves a classical problem about the heat equation that has remained open for decades and is widely regarded as one of the central questions in the area. The problem can be described as follows: if we know how temperature is fixed along the boundary of a region whose shape may be irregular and even change over time, can we determine the temperature everywhere inside in a stable and well-controlled way?

For instance, in a piece of metal with a slightly rough surface whose geometry evolves with time, if some parts of its surface are heated and others cooled, heat diffuses into the interior according to the heat equation, but it is far from obvious whether irregularities of the boundary could amplify and produce pathological behaviour of the hat diffusion. Conversely, one may know that the temperature inside behaves in a stable and controlled way and ask whether this forces the piece of metal to have a boundary possessing some intrinsic geometric regularity.

The article published in Inventiones Mathematicae, entitled “Solvability of the L^p Dirichlet Problem for the Heat Equation is Equivalent to Parabolic Uniform Rectifiability in the Case of a Parabolic Lipschitz Graph” and authored by Simon Bortz (University of Alabama), Steve Hofmann (University of Missouri), José María Martell (ICMAT-CSIC) and Kaj Nyström (Uppsala University), studies precisely this question for domains whose boundary can be described as a Lipschitz graph of a function depending on space and time.

The authors show that analytic stability of heat diffusion inside such a domain is not merely influenced by the geometry of the boundary, but is in fact completely determined by it. More specifically, they prove that three conditions that had previously been understood only partially—solvability of the Dirichlet problem in Lebesgue spaces, quantitative control of caloric measure, which describes the behavior of the temperature inside the domain, and a geometric property known as parabolic uniform rectifiability—are actually equivalent.

In concrete terms, if the heat measure behaves comparably to the natural surface measure, then the boundary must possess a precise and quantifiable geometric structure: although it may appear irregular, it is irregular in a controlled and measurable way. Technically, this corresponds to the defining function of the boundary having the fractional time derivative of order 1/2 in a bounded mean oscillation space, a condition that turns out to be exactly the right geometric criterion.

Earlier work had shown that certain geometric assumptions guarantee solvability, but whether those assumptions were also necessary was unknown. By proving the full equivalence, this paper provides a complete characterization of which parabolic graph domains allow the Dirichlet problem for the heat equation to be solved, thereby closing a longstanding gap and completing a program initiated in foundational work of de John L. Lewis and Margaret A. M. Murray.

Conceptually, the proof introduces a strikingly original perspective: it analyzes the level sets of the Green function, which describes pointwise influence for the heat equation, treating them as extensions of the boundary itself and testing refined Littlewood–Paley-type estimates that measure oscillation across scales. These estimates make it possible to connect analytic behaviour, geometric structure, and boundary solvability within a single unified framework.

By establishing this definitive bridge between geometry, analysis, and heat diffusion, the work delivers a sharp structural theorem where previously only partial results were known, and is therefore widely viewed as a genuine breakthrough that sets a new benchmark for research in parabolic partial differential equations, harmonic analysis, and geometric measure theory.

Reference

Bortz, S., Hofmann, S., Martell, J.M. et al. Solvability of the Dirichlet problem for the heat equation is equivalent to parabolic uniform rectifiability in the case of a parabolic Lipschitz graph. Invent. math. 239, 165–217 (2025). https://doi.org/10.1007/s00222-024-01300-1