Gradient Flow of Wasserstein-Quantization Supports

Definitions and mathematical statement

Let $d\geq 1$ be an integer and let $K\subset{\mathbb{R}}^d$ be compact. The source concerns a density supported in $K$. In the usual probability-measure formulation, this would mean a measurable function $\rho:{\mathbb{R}}^d\to[0,\infty)$ with $\rho=0$ almost everywhere outside $K$ and $\int_{{\mathbb{R}}^d}\rho(x)\,{\,\mathrm{d}} x=1$, defining $\mu=\rho\,{\,\mathrm{d}} x$. Probability normalization is needed for this formulation but is not explicitly stated in the source; no additional regularity or positivity assumption on $\rho$ is supplied.

For an order $1\leq p<\infty$, the $p$-Wasserstein distance between probability measures $\alpha,\beta$ on ${\mathbb{R}}^d$ with finite $p$-th moments is \[ W_p(\alpha,\beta)=\left(\inf_{\pi\in\Pi(\alpha,\beta)}\int_{{\mathbb{R}}^d\times{\mathbb{R}}^d}|x-y|^p\,{\,\mathrm{d}}\pi(x,y)\right)^{1/p}, \] where $\Pi(\alpha,\beta)$ denotes their couplings and $|\cdot|$ is the Euclidean norm. The source does not specify the Wasserstein order; this family of definitions does not select one.

For a positive integer $N$, represent candidate sites by $X_N=(x_1,\ldots,x_N)$, with associated set $S_N=\{x_1,\ldots,x_N\}$. One conventional quantization energy, allowing the atomic weights to be optimized, is \[ Q_{N,p}(X_N)=\inf_{\substack{a_i\geq0\ (1\leq i\leq N)\\\sum_{i=1}^N a_i=1}}W_p\!\left(\mu,\sum_{i=1}^N a_i\delta_{x_i}\right)^p, \] where $\delta_x$ is the unit point mass at $x$. A different convention prescribes equal weights and uses \[ \nu_N=\frac1N\sum_{i=1}^N\delta_{x_i},\qquad W_p(\mu,\nu_N)^p. \] The source does not distinguish these conventions, specify whether to use $W_p$ or its $p$-th power, or say whether sites must remain distinct. These choices must therefore remain unresolved. The admissible sites might lie in $K$ or in a larger ambient domain; the source supplies no constraint or boundary rule.

Statement of the supplied problem. For the intended $N$-point Wasserstein quantization model of $\mu$, characterize the evolution of the support set $S_N(t)$ under steepest descent, and determine how this discrete evolution compares, as $N\to\infty$, with the Wasserstein gradient flow of the limiting energy.

Unresolved formulation. Steepest descent requires a specified energy and a metric on the evolving configurations or atomic measures. For example, a differentiable configuration energy $E_N$ with the Euclidean product metric would give the conditional definition $\dot x_i(t)=-\nabla_{x_i}E_N(X_N(t))$; this is only an illustration of a possible convention, not an evolution law asserted by the source. A precise limiting comparison also requires the definition and normalization of the limiting energy, its measure domain, its Wasserstein metric, the representation of configurations as measures, initial-data assumptions, any time rescaling, and the topology and time interval of comparison. None is specified. Consequently, the supplied record determines a research question but does not determine a unique evolution equation or a fully quantified convergence assertion.

Short English statement

For a density supported on a compact subset of ${\mathbb{R}}^d$, describe how the sites of an $N$-point Wasserstein quantization move under steepest descent and compare their evolution, as $N\to\infty$, with the Wasserstein gradient flow of a limiting energy. The source leaves the quantization convention, descent metric, limiting energy, scaling, initial data, and boundary conditions unspecified, so a unique precise convergence statement cannot yet be extracted.

Sources

• Local registry: data/unsolvedmath/problems.json, with catalog/problems.json as the supplementary registry specified by the project; supplied permanent UnsolvedMath ID $30003665$, exact source problem code catalogue entry, title Gradient Flow of Wasserstein-Quantization Supports. The supplied record belongs to the Oberwolfach Reports Open Problems collection. No individual proposer is identified.
• Supplied report citation: Variational Methods for Evolution (2018), Oberwolfach Reports, pp. 3185--3261, DOI 10.4171/owr/2017/54, https://doi.org/10.4171/owr/2017/54. The record gives proposed year 2017; the citation's parenthetical year is retained as supplied.
• Supplied literature source: A gradient flow perspective on quantization, https://arxiv.org/abs/1711.01652.
• Supplied literature source: Quantization in space dimension two, https://doi.org/10.1016/j.anihpc.2017.12.003.
• Supplied literature source: OWR Report 54/2017, https://ems.press/content/serial-article-files/46718.
• Literature assessment dated 2026-08-21: the supplied curation labels the problem partially solved and reports that optimal-measure asymptotics and a gradient-flow viewpoint have been developed in one dimension and selected two-dimensional settings, while the general evolution and Wasserstein-flow comparison remain open. This is supplied dated curation, not an independently verified assessment. The background explicitly requests independent verification of time-sensitive status and cited claims.
• The supplied statement status is corrected_verified; its accompanying qualification says that mathematical content was preserved while notation, typography, and sentence structure were repaired for publication. This is a registry annotation, not independent mathematical verification of this draft or of the cited literature.