Title: Uniform error estimate of an asymptotic preserving scheme for the Lévy-Fokker-Planck equation
We establish a uniform-in-scaling error estimate for the asymptotic preserving (AP) scheme proposed by Xu and Wang [Commun. Math. Sci. 21 (2023), pp. 1–23] for the Lévy-Fokker-Planck (LFP) equation. The main difficulties stem not only from the interplay between the scaling and numerical parameters but also the slow decay of the tail of the equilibrium state. We tackle these problems by separating the parameter domain according to the relative size of the scaling parameter ε<#comment/>\varepsilon : in the regime where ε<#comment/>\varepsilon is large, we design a weighted norm to mitigate the issue caused by the fat tail, while in the regime where ε<#comment/>\varepsilon is small, we prove a strong convergence of LFP towards its fractional diffusion limit with an explicit convergence rate. This method extends the traditional AP estimates to cases where uniform bounds are unavailable. Our result applies to any dimension and to the whole span of the fractional power.  more » « less
Award ID(s):
1846854
PAR ID:
10515946
Author(s) / Creator(s):
;
Publisher / Repository:
AMS
Date Published:
Journal Name:
Mathematics of Computation
ISSN:
0025-5718
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Let (R,m) (R,\mathfrak {m}) be a Noetherian local ring of dimension d≥<#comment/>2d\geq 2 . We prove that if e( R^<#comment/> red )>1 e(\widehat {R}_{red})>1 , then the classical Lech’s inequality can be improved uniformly for all m\mathfrak {m} -primary ideals, that is, there exists ε<#comment/>>0\varepsilon >0 such that e(I)≤<#comment/>d!(e(R)−<#comment/>ε<#comment/>)ℓ<#comment/>(R/I)e(I)\leq d!(e(R)-\varepsilon )\ell (R/I) for all m\mathfrak {m} -primary ideals I⊆<#comment/>RI\subseteq R . This answers a question raised by Huneke, Ma, Quy, and Smirnov [Adv. Math. 372 (2020), pp. 107296, 33]. We also obtain partial results towards improvements of Lech’s inequality when we fix the number of generators of II . 
    more » « less
  2. In this paper we consider which families of finite simple groups GG have the property that for each ϵ<#comment/>>0\epsilon > 0 there exists N>0N > 0 such that, if |G|≥<#comment/>N |G| \ge N and S,TS, T are normal subsets of GG with at least ϵ<#comment/>|G| \epsilon |G| elements each, then every non-trivial element of GG is the product of an element of SS and an element of TT . We show that this holds in a strong and effective sense for finite simple groups of Lie type of bounded rank, while it does not hold for alternating groups or groups of the form PSLn (q) \mathrm {PSL}_n(q) where qq is fixed and n→<#comment/>∞<#comment/>n\to \infty . However, in the case S=TS=T and GG alternating this holds with an explicit bound on NN in terms of ϵ<#comment/>\epsilon . Related problems and applications are also discussed. In particular we show that, if w1,w2 w_1, w_2 are non-trivial words, GG is a finite simple group of Lie type of bounded rank, and for g∈<#comment/>Gg \in G , P w1(G),w2(G) (g) P_{w_1(G),w_2(G)}(g) denotes the probability that g1g2=g g_1g_2 = g where gi∈<#comment/>wi(G) g_i \in w_i(G) are chosen uniformly and independently, then, as |G|→<#comment/>∞<#comment/> |G| \to \infty , the distribution P w1(G),w2(G) P_{w_1(G),w_2(G)} tends to the uniform distribution on GG with respect to the L∞<#comment/>L^{\infty } norm. 
    more » « less
  3. An nn -vertex graph GG ispancyclicif for every ℓ<#comment/>\ell , 3≤<#comment/>ℓ<#comment/>≤<#comment/>n3\leq \ell \leq n , GG contains a cycle of length ℓ<#comment/>\ell . We consider an analog for hypergraphs. An rr -uniform hypergraph is a set of vertices and a set of edges in which every edge contains rr vertices. A Berge cycle of length ℓ<#comment/>\ell in a hypergraph is an alternating sequence of ℓ<#comment/>\ell distinct vertices and ℓ<#comment/>\ell distinct edges v1,e1,v2,…<#comment/>,vℓ<#comment/>, eℓ<#comment/> ,v1 v_1,e_1,v_2, \ldots , v_\ell , e_{\ell },v_1 such that {vi, vi+1 }⊆<#comment/>ei \{v_i, v_{i+1}\} \subseteq e_i for all ii , with indices taken modulo ℓ<#comment/>\ell . We similarly call a hypergraphBerge pancyclicif it contains Berge cycles of every possible length. For sufficiently large nn we prove a sharp minimum degree condition guaranteeing Berge pancyclicity in rr -uniform hypergraphs where 3≤<#comment/>r≤<#comment/>3 \leq r \leq ⌊<#comment/>(n−<#comment/>1)/2⌋<#comment/>−<#comment/>2\lfloor (n-1)/2\rfloor - 2 . This bound is optimal for all such combinations of nn and rr . 
    more » « less
  4. Let Σ<#comment/>\Sigma be a smooth Riemannian manifold, Γ<#comment/>⊂<#comment/>Σ<#comment/>\Gamma \subset \Sigma a smooth closed oriented submanifold of codimension higher than 22 and TT an integral area-minimizing current in Σ<#comment/>\Sigma which bounds Γ<#comment/>\Gamma . We prove that the set of regular points of TT at the boundary is dense in Γ<#comment/>\Gamma . Prior to our theorem the existence of any regular point was not known, except for some special choice of Σ<#comment/>\Sigma and Γ<#comment/>\Gamma . As a corollary of our theorem we answer to a question in Almgren’sAlmgren’s big regularity paperfrom 2000 showing that, if Γ<#comment/>\Gamma is connected, then TT has at least one point pp of multiplicity 12\frac {1}{2} , namely there is a neighborhood of the point pp where TT is a classical submanifold with boundary Γ<#comment/>\Gamma ; we generalize Almgren’s connectivity theorem showing that the support of TT is always connected if Γ<#comment/>\Gamma is connected; we conclude a structural result on TT when Γ<#comment/>\Gamma consists of more than one connected component, generalizing a previous theorem proved by Hardt and Simon in 1979 when Σ<#comment/>=Rm+1\Sigma = \mathbb R^{m+1} and TT is mm -dimensional. 
    more » « less
  5. Let ff be analytic on [0,1][0,1] with |f(k)(1/2)|⩽<#comment/>Aα<#comment/>kk! |f^{(k)}(1/2)|\leqslant A\alpha ^kk! for some constants AA and α<#comment/>>2\alpha >2 and all k⩾<#comment/>1k\geqslant 1 . We show that the median estimate of μ<#comment/>=∫<#comment/>01f(x)dx \mu =\int _0^1f(x)\,\mathrm {d} x under random linear scrambling with n=2mn=2^m points converges at the rate O(n−<#comment/>clog⁡<#comment/>(n))O(n^{-c\log (n)}) for any c>3log⁡<#comment/>(2)/π<#comment/>2≈<#comment/>0.21c> 3\log (2)/\pi ^2\approx 0.21 . We also get a super-polynomial convergence rate for the sample median of 2k−<#comment/>12k-1 random linearly scrambled estimates, when k/m k/m is bounded away from zero. When ff has a pp ’th derivative that satisfies a λ<#comment/>\lambda -Hölder condition then the median of means has error O(n−<#comment/>(p+λ<#comment/>)+ϵ<#comment/>)O( n^{-(p+\lambda )+\epsilon }) for any ϵ<#comment/>>0\epsilon >0 , if k→<#comment/>∞<#comment/>k\to \infty as m→<#comment/>∞<#comment/>m\to \infty . The proof techniques use methods from analytic combinatorics that have not previously been applied to quasi-Monte Carlo methods, most notably an asymptotic expression from Hardy and Ramanujan on the number of partitions of a natural number. 
    more » « less