Optimal transportation, dissipative PDE's and functional inequalities (Q4428796)

The author studies three fundamental problems in the theory of mass transportation: Problem A (Rate of convergence as \(t\to \infty\)): Consider the diffusive, nonlinear partial differential equation where the unknown \((f_t)_{t\geq 0}\) is a time-dependent probability density on \(\mathbb{R}^d\), \[ \frac{\partial f}{\partial t}= \sigma\Delta f+\nabla \cdot (f\nabla V)+\nabla \cdot (f\nabla (f\partial W)), \qquad t\geq 0,\;x\in \mathbb{R}^d \tag \(*\) \] (here \(\nabla \) stands for the gradient operator in \(\mathbb{R}^d\), while \(\nabla \cdot\) is its adjoint, the divergence operator; moreover, \(V\) and \(W\) are smooth potentials). Does this equation admit a stationary state? If the answer is yes, do solutions of \((*)\) converge towards this stationary state? Problem B (Rate of convergence as \(N\to\infty\)): Consider a bunch of \(N\) particles in \(\mathbb{R}^d\), with respective positions \(X_t^i\) \((1\leq i\leq N)\) at time \(t\geq 0\), solutions of the stochastic differential equation \[ dX_t^i= dB_t^i- \nabla V(X_t^i)\,dt- \frac1N \sum_{j=1}^N \nabla W(X_t^i- X_t^j)\,dt, \qquad 1\leq i\leq N, \] starting from some chaotic initial configuration \(X_0= (X_0^1,\dots, X_0^N)\) with law \({\mathcal L}(X_0)= f_0^{\otimes N}\,dx\) on \((\mathbb{R}^d)^N\). Under some assumptions on \(V\) and \(W\) it is known that for each time \(t>0\) the density of particles in \(\mathbb{R}^d\) associated with this system converges, as \(N\to \infty\), towards the unique solution of \((*)\). This means for instance, that for any bounded continuous function \[ \frac1N \sum_{i=1}^N \varphi(X_t^i) @>> N\to\infty> \int_{\mathbb{R}^d} f_t \varphi\, dx, \] in the sense of convergence in law for random variables (note that the left-hand side is random, while the right-hand side is not). Can one estimate its speed of convergence as \(N\to \infty\)? Problem C (Optimal constants): What is the optimal constant, and how do minimizers look like, in the Gagliardo-Nirenberg interpolation inequality \[ \| w\|_{L^p (\mathbb{R}^n)}\leq C\|\nabla w\|_{L^2 (\mathbb{R}^n)}^\theta \| x\|_{L^q (\mathbb{R}^n)}^{1-\theta} \] (with some compatibility conditions on \(p,q\))? What about the Young inequality \[ \| f*g \|_{L^r (\mathbb{R}^n)}\leq C\| f\|_{L^p(\mathbb{R}^n)} \| g\|_{L^q (\mathbb{R}^n)}? \] For the entire collection see [Zbl 1013.00028].