Self-similar solutions with fat tails for Smoluchowski's coagulation equation with singular kernels (Q327753)

Smoluchowski's coagulation equation \[ \partial_t f(t,x) = \frac{1}{2} \int_0^x K(x-y,y) f(t,x-y) f(t,y) dy - \int_0^\infty K(x,y) f(t,x) f(t,y) dy \] describes the dynamics of the particle size distribution function \(f(t,x)\) of particles of size \(x\in (0,\infty)\) at time \(t>0\). When the coagulation kernel \(K\) is homogeneous of degree \(\lambda<1\), it is expected to have two kinds of self-similar solutions: mass-conserving self-similar solutions of the form \[ (t,x)\mapsto t^{-2/(1-\lambda)} g\big(xt^{-1/(1-\lambda)}\big), \] where the profile \(g\) has a finite first moment and is expected to decay exponentially fast as \(x\to\infty\), and self-similar solutions with an infinite first moment of the form \[ (t,x)\mapsto t^{-1-(1+\lambda)\beta} g\big(xt^{-\beta}\big), \] the profile \(g\) decaying slowly as \(x\to\infty\) (fat tail profile). Due to this slow decay, the existence of self-similar solutions belonging to this second class is harder to establish but has been proved for coagulation kernels with no singularity as \((x,y)\to (0,0)\) in [the first and third author, Commun. Math. Phys. 318, No. 2, 505--532 (2013; Zbl 1267.82086); erratum ibid. 318, No. 2, 533--534 (2013)]. The purpose of the paper under review is to extend this existence result to coagulation kernels behaving as \((x^{-a} y^b + x^b y^{-a})\) for some \(a>0\) and \(b\in (-\infty,1)\) and thus having homogeneity \(\lambda = b-a\) and featuring a singularity as \(x\to 0\) or \(y\to 0\). It includes in particular Smoluchowski's coagulation kernel \((x^{1/3}+y^{1/3})(x^{-1/3}+y^{-1/3})\) which corresponds to \(a=b=1/3\). More precisely, it is shown that, for this class of coagulation kernels, there is a self-similar solution of the second type provided \(\rho:=b-a+1/\beta \in \left( \max\{b,0\},1\right)\). In that case, the profile \(g\) is continuous on \((0,\infty)\) and satisfies \(g(x) \sim C x^{-1-\rho}\) as \(x\to\infty\) and vanishes fast as \(x\to 0\). Several steps are needed to construct such a solution: first, a dynamical approach is used to obtain a solution of a suitably regularized problem which satisfies the expected upper bounds. The next step is the derivation of appropriate lower bounds and is the toughest part of the proof. A compactness argument completes the existence proof.