ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let G(n)\mathbb{G}^{(n)} be the finite spatial inhomogeneous random graph, let CCG(n)(k)\mathrm{CC}_{\mathbb{G}^{(n)}}(k) denote its clustering function, and let γ(k)\gamma(k) denote the clustering function of the corresponding infinite SIRG. For a sequence kn→∞k_n\to\infty, define

ϕ(kn):={knlog⁡knif β=a+1, a∈(1,∞),knotherwise,\phi(k_n):=\begin{cases} \frac{k_n}{\log k_n} &\text{if }\beta=a+1,\ a\in(1,\infty),\\ k_n &\text{otherwise,} \end{cases}

and

ψ(n):={n1β−1if β≥(a+1)∨2, a∈[0,∞),n2+a−ββ−1otherwise.\psi(n):=\begin{cases} n^{\frac{1}{\beta-1}} &\text{if }\beta\geq (a+1)\vee 2,\ a\in[0,\infty),\\ n^{\frac{2+a-\beta}{\beta-1}} &\text{otherwise.} \end{cases}

Scaling conjecture for finite graphs. Consider the finite SIRG G(n)\mathbb{G}^{(n)}. If ϕ(kn)=o(ψ(n))\phi(k_n)=o(\psi(n)), then

CCG(n)(kn)γ(kn)→P1.\frac{\mathrm{CC}_{\mathbb{G}^{(n)}}(k_n)}{\gamma(k_n)}\xrightarrow{\mathbb{P}}1.

If ϕ(kn)=(1+o(1))ψ(n)\phi(k_n)=(1+o(1))\psi(n), then, for some L0,λ>0L_0,\lambda>0,

CCG(n)(kn)γ(kn)→dL0Poi⁡(λ).\frac{\mathrm{CC}_{\mathbb{G}^{(n)}}(k_n)}{\gamma(k_n)}\xrightarrow{d}L_0\operatorname{Poi}(\lambda).

If ϕ(kn)=ω(ψ(n))\phi(k_n)=\omega(\psi(n)), then

CCG(n)(kn)γ(kn)→P0.\frac{\mathrm{CC}_{\mathbb{G}^{(n)}}(k_n)}{\gamma(k_n)}\xrightarrow{\mathbb{P}}0.

These conjectured limit laws describe the regimes in which the finite model follows the infinite-model clustering asymptotics, exhibits a critical Poisson-type fluctuation, or has negligible clustering relative to γ(kn)\gamma(k_n). The conjecture is motivated by local convergence of the finite model to the infinite SIRG and the known fixed-kk convergence, but the simultaneous growth of knk_n with nn remains unresolved.

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.