Moment computation conjecture for the optimal estimator

Let Pt\mathcal{P}_{\bm{t}} and Vs\bm{V}_s be as in Proposition. For a coordinate kk, define the moment

μk:=PtxkdϱP=vol(Pt)x^k.\mu_k:=\int_{\mathcal{P}_{\bm{t}}} x_k\,d\varrho_{\mathcal{P}}=\operatorname{vol}(\mathcal{P}_{\bm{t}})\hat{x}_k.

Let Δ\Delta, λ\boldsymbol{\lambda}, and the inverse Laplace operators Lλm1\mathcal{L}_{\lambda_m}^{-1} be as in the preceding construction. Moment computation conjecture. Provided that the integrals on the right-hand side exist,

μk=(mLλm1)(Δxkexp(Vsλ,x)dx).\mu_k=\left(\mathop{\circ}_{m}\mathcal{L}_{\lambda_m}^{-1}\right)\left(\int_{\Delta}x_k\exp\bigl(-\langle\bm{V}_s\boldsymbol{\lambda},\bm{x}\rangle\bigr)\,d\bm{x}\right).

This conjecture supplies the final moment-computation step for the optimal estimator. The source reports that it was numerically verified, and the parser status indicates that its numerical correctness was checked extensively against integration by simplicial decomposition, although a rigorous proof is not supplied here.

Sources & referencesView supporting material

Primary source

Steffen Limmer and Slawomir Stanczak, “Optimal deep neural networks for sparse recovery via Laplace techniques”, arXiv:1709.01112 (2017).

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.