奇异非局部输运方程向热流的定量扩散极限

My Thoughts on the Nonlocal Continuity Equation

Okay, so I'm diving into this nonlocal continuity equation on a closed connected Riemannian manifold. The equation itself, [ \partial_tμ_b =\operatorname{div}!\left( μ_b\nabla\log\bigl((I-b^2Δ)^{-1}μ_b\bigr) \right) ], is what I'm dealing with. The really interesting bit is that for smooth, strictly positive initial data, the global solution, as b tends to zero, converges to heat flow at a sharp, time-uniform rate. I'm seeing that the key here is a uniform dissipation of a b-weighted higher-order resolvent energy, which yields exponential relaxation. The thing is, this is happening despite the absence of a Wasserstein gradient-flow structure, which is pretty neat. I'm also extending the analysis to a corresponding deterministic N-particle dynamics on the circle. A weak-strong modulated energy argument is giving me [ \mathbb E!\left[ \sup_{t\ge0}W_1(μ_b^N(t),μ_b(t)) \right] \le C(Nb)^{-1/2} ] for iid initialization, which is cool. So, the upshot? Choosing b proportional to N to the negative one-fifth, I can approximate the heat flow uniformly in time at a rate of N to the negative two-fifths. That's a good result.

我们研究闭连通黎曼流形上的非局部连续性方程 [ \partial_tμ_b =\operatorname{div}!\left( μ_b\nabla\log\bigl((I-b^2Δ)^{-1}μ_b\bigr) \right) ] 对于光滑且严格正的初值，我们证明当 $b \to 0$ 时，其全局解以最优的时间一致速率收敛至热流 $μ(t)$： [ \sup_{t\ge0}|μ_b(t)-μ(t)|{L^1}\le Cb^2. ] 关键估计在于一个 $b$ 加权的高阶预解式能量的一致耗散，尽管缺乏 Wasserstein 梯度流结构，该估计仍给出了指数松弛。在圆周上，我们还分析了相应的确定性 $N$ 粒子动力学。通过弱-强调制能量论证，对于独立同分布初值可得 [ \mathbb E!\left[ \sup{t\ge0}W_1(μ_b^N(t),μ_b(t)) \right] \le C(Nb)^{-1/2} ] 由此，选取 $b\asymp N^{-1/5}$ 能够以 $N^{-2/5}$ 的速率在时间上一致逼近热流。