Singular-limit conjecture for adaptive weight dynamics

Let the assumptions of Theorem~ hold. For ϵ>0\epsilon>0, let (X(ϵ)(t),w(ϵ)(t))(\bm{X}^{(\epsilon)}(t),\bm{w}^{(\epsilon)}(t)) solve, and let (X(0)(t))(\bm{X}^{(0)}(t)) solve the limiting system with the same initial data. Define wij(0)(t):=β(Xi(0)(t),Xj(0)(t))\bm{w}^{(0)}_{ij}(t)\vcentcolon=\beta(\bm{X}^{(0)}_i(t),\bm{X}^{(0)}_j(t)). Singular-limit conjecture for adaptive weight dynamics. There exists C(t)>0C(t)>0, independent of ϵ\epsilon and dependent on tt and the bounds in Assumptions~ and, such that

1N2w(ϵ)(t)w(0)(t)1([N]×[N])+1NX(ϵ)(t)X(0)(t)2([N])ϵC(t).\frac{1}{N^2}\|\bm{w}^{(\epsilon)}(t)-\bm{w}^{(0)}(t)\|_{\ell^1([N]\times[N])}+\frac{1}{N}\|\bm{X}^{(\epsilon)}(t)-\bm{X}^{(0)}(t)\|_{\ell^2([N])}\leq\epsilon C(t).

This formalizes the expectation that rapidly relaxing weights become instantaneously determined by the states, reducing the coupled system to the limiting state equation.

Sources & referencesView supporting material

Primary source

Datong Zhou, “Non-exchangeable mean-field theory for adaptive weights: propagation of dissociatedness and graphon sampling lemma”, arXiv:2506.13587 (2025).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.