ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let ZnZ_n be the population measure at generation nn, let g∈Cc+(Rd)g\in C_c^+\left(\mathbb R^d\right), and write ⟨Zn,g⟩\langle Z_n,g\rangle for its integral against gg. Let γ\gamma be the interaction parameter, and let Nd(m,Σ)\mathcal N^d(m,\Sigma) denote the dd-dimensional normal distribution with mean mm and covariance matrix Σ\Sigma. Local-mass asymptotics conjecture. If γ>0\gamma>0 (attraction), then there exists a random variable N∼Nd(0,2)N\sim\mathcal N^d(0,2) such that, conditional on N=x0N=x_0,

lim⁡n→∞2−n⟨Zn,g⟩=⟨(γπ)d/2exp⁡(−γ∣⋅−x0∣2),g⟩a.s.\lim_{n\to\infty}2^{-n}\langle Z_n,g\rangle=\left\langle\left(\frac{\gamma}{\pi}\right)^{d/2}\exp\left(-\gamma|\cdot-x_0|^2\right),g\right\rangle\quad\mathrm{a.s.}

Also, there exists a random variable M∼Nd(0,(2+14γ2)Id)M\sim\mathcal N^d\left(0,\left(2+\frac{1}{4\gamma^2}\right)\mathbf I_d\right), with corresponding law P\mathbb P, such that

lim⁡n→∞2−nE⟨Zn,g⟩=E⟨M,g⟩a.s.\lim_{n\to\infty}2^{-n}E\langle Z_n,g\rangle=\mathbb E\langle M,g\rangle\quad\mathrm{a.s.}

Here Id\mathbf I_d is the dd-dimensional unit matrix. If γ<0\gamma<0 (repulsion), then

lim⁡n→∞2−n⟨Zn,g⟩=⟨1,g⟩a.s.\lim_{n\to\infty}2^{-n}\langle Z_n,g\rangle=\langle 1,g\rangle\quad\mathrm{a.s.}

The conjecture combines the asymptotic decomposition into a limiting center of mass and a branching Ornstein–Uhlenbeck process with the strong law for local mass. The stated monotonicity of 2+1/(4γ2)2+1/(4\gamma^2) suggests that stronger attraction gives a smaller variance for the limiting distribution; the claims themselves are presented as conjectural.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.