Create and explore imaginative 3D worlds with our new product Dreamfield (opens in a new tab)

Optimal Wp–SWp comparisons for p>1

Establish sharp quantitative inequalities controlling the p‑Wasserstein distance Wp(μ,ν) by the sliced distance SWp(μ,ν) for p>1 on ℝ^d, determining the optimal exponent and its dependence on dimension under natural support or moment assumptions.

Background

Beyond the case p=1 studied in this work, the authors highlight that the regime p>1 is largely unexplored. While one can derive bounds by combining known p=1 estimates with elementary arguments (e.g., yielding an exponent 1/(p(d+1)) for compactly supported measures), such exponents are believed to be far from optimal.

Resolving this would advance the theoretical justification for using sliced distances as proxies for higher‑order Wasserstein metrics in applications, by pinpointing the true quantitative relationship for p>1.

References

Let us finally mention the (widely open, in our opinion) case p>1.

— Sharp comparisons between sliced and standard $1$-Wasserstein distances  (2510.16465 - Carlier et al., 18 Oct 2025) in Introduction (final paragraph of contributions)

Our proof techniques do not handle the case of non-integer $p\geq 1$. The corresponding result for non-integer $p\geq 1$ remains open.

— Embedding large subsets of Wasserstein spaces into Banach spaces  (2610.12029 - Letrouit et al., 8 Oct 2026) in Section 1, subsection “Embedding via sliced Wasserstein distances,” immediately before Proposition \ref{th:nonequiv-sw}