The paper tackles a long-standing open problem in the mathematics of partial differential equations, specifically a conjecture made by the mathematician Pierre-Louis Lions about a class of equations called viscous Hamilton-Jacobi equations. These equations appear throughout physics, economics, and control theory, describing how quantities evolve when both diffusion and nonlinear transport effects are present. The conjecture is about "maximal regularity," meaning: if the input data (a source term) belongs to a certain function space, does the solution automatically inherit the best possible smoothness one could hope for? The paper focuses on the hardest case of this question, called the endpoint or critical case, where the source term sits at the exact threshold exponent that makes the problem most delicate.
The central difficulty at this critical threshold is a phenomenon called concentration, where the energy or gradient of a solution can pile up at a single point rather than spreading out smoothly, potentially destroying regularity. The authors rule this out under a mild assumption: that the collection of source terms is "equi-integrable," meaning no mass escapes to arbitrarily small regions in a uniform sense. Their main result is that under this condition, the solutions do indeed achieve the optimal regularity one conjectured, confirming Lions' prediction at the endpoint.
The proof strategy combines two sophisticated tools. First, the authors use a two-stage blow-up argument, which is a way of zooming in on potential trouble spots at different scales to analyze what the solution looks like locally. Second, they apply a Liouville-type rigidity theorem, a result that says certain equations defined on all of space can only have trivial or highly constrained solutions, which eliminates the possibility of bad behavior surviving the zoom-in process. Together, these tools allow the authors to convert weak (incomplete) compactness of approximating solutions into genuine strong compactness, closing off the concentration scenario and establishing the desired regularity.