The paper studies a type of partial differential equation called a nonlinear Fokker-Planck equation, which describes how a probability distribution evolves over time when particles are spreading out (diffusion) and being pushed by a force (drift). What makes this version especially challenging is that the diffusion part is "non-diagonal," meaning the spreading in different spatial directions can be coupled together in a complicated way. The central mathematical question is whether there is only one solution to this equation given a starting distribution, a property called uniqueness. The authors prove that a natural candidate solution, built using a well-established mathematical tool called a semigroup, is in fact the only solution within a very broad class of possible solutions called distributional solutions.
A key technical step is proving uniqueness for a related, simpler "linearized" version of the equation. This linearized result then acts as a lever to establish uniqueness for the full nonlinear problem. The authors also derive a new bound showing that if you start with an initial condition that is both integrable and bounded, the solution remains bounded for all later times, with a controlled estimate on how large it can get. These results extend earlier work that only handled the simpler "isotropic" or diagonal case, where diffusion acts the same way in every direction, to the more general and physically realistic anisotropic setting.
On the probabilistic side, the paper connects these analytical results to a class of stochastic differential equations called McKean-Vlasov equations, where the random motion of a particle depends not just on its current position but on the overall distribution of all particles simultaneously. The authors prove that these equations have a unique probability law, meaning the randomness is well-defined and not ambiguous. They further show that the collection of probability laws associated with these solutions forms what is called a nonlinear Markov process in McKean's sense, a rigorous framework for describing interacting particle systems where the future depends only on the present state of the whole distribution.