Sharp macroscopic blow-up behavior for the parabolic-elliptic Keller-Segel system in dimensions $n\ge 3$

TL;DR

This paper characterizes the sharp macroscopic blow-up behavior of solutions to the parabolic-elliptic Keller-Segel system in dimensions 3 to 9, providing precise asymptotics, estimates, and profiles near blow-up time.

Contribution

It establishes the existence of nonflat backward self-similar solutions describing blow-up, improving previous microscopic scale results to a full space-time scale analysis, and simplifies related proofs.

Findings

Existence of nonflat backward self-similar solutions

Sharp two-sided global estimates for solution behavior

Precise final blow-up profile with nonzero limit

Abstract

We study the space-time concentration or blow-up asymptotics of radially decreasing solutions of the parabolic-elliptic Keller-Segel system in the whole space or in a ball. We show that, for any solution in dimensions $3\le n\le 9$ (assuming finite mass in the whole space case), there exists a nonflat backward self-similar solution $U$ such that $$u(x,t)=(1+o(1))U(x,t),\quad\hbox{as $(x,t)\to (0,T)$.}$$ This macroscopic behavior is important from the physical point of view, since it gives a sharp description of the concentration phenomenon in the scale of the original space-time variables~$(x,t)$. It strongly improves on existing results, since such behavior was previously known (\cite{GMS}) to hold only in the microscopic scale $|x|\le O(\sqrt{T-t})$ as $t\to T$ (and in the whole space case only). As a consequence, we obtain the two-sided global estimate $$C_1\le (T-t+|x|^2)u(x,t)\le C_2\quad\hbox{in $B_R\times(T/2,T)$},$$ whose upper part only was known before (\cite{Soup-Win}), as well as the sharp final profile: $$\lim_{x\to 0} |x|^2u(x,T)=L\in(0,\infty).$$ The latter improves, with a different proof, the recent result of \cite{BZ} by excluding the possibility $L=0$. We also give extensions of these results, in higher dimensions, to type~I and to time monotone solutions. Moreover, we extend the known results on type I estimates and on convergence in similarity variables, and significantly simplify their proofs.