Searcharxiv⌕ Search

arXiv · 2611.22492

Equivariant Flow Matching on Homogeneous Spaces

Abstract

Flow matching enables the training of networks that can flow samples from an initial to a target distribution. The network is regressed against simple vector fields that are conditioned on end points in the target distribution. In the Euclidean case, these conditional vector fields induce flows along straight lines. Various generalisations have been proposed to perform flow matching on spaces where straight lines are not well-defined, such as Riemannian manifolds and Lie groups. Riemannian flow matching applies to many domains, but typically requires expensive simulation, while flow matching on Lie groups is simulation-free but only works on Lie groups. In this work, we develop equivariant flow matching on homogeneous spaces, which generalises flow matching on Lie groups, while remaining simulation-free. We also derive all generators connecting pairs of position-orientations, generalising, and perform a similar derivation for event-velocity space, the Lorentzian analogue of position-orientation space, so that we can perform flow matching on these spaces.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Finn M. Sherry, Gijs Bellaard. 2026-11-17. Equivariant Flow Matching on Homogeneous Spaces. https://searcharxiv.com/abs/2611.22492

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Almost minimizing Yang$-$Mills fields: log-epiperimetric inequality, non-concentration, and uniqueness of tangents

We establish a direct log-epiperimetric inequality for Yang$-$Mills fields in arbitrary dimension and we leverage it to prove uniqueness of tangent cones with isolated singularity for almost energy minimizing connections and $ω$-ASD connections (where $ω$ is not necessarily closed) satisfying some suitable regularity and non-concentration assumptions. En route to this we establish a Luckhaus type lemma for connections to exclude curvature concentration along blow-up sequences.

math.DG↗

Fine properties of nonlinear potentials and a unified perspective on monotonicity formulas

We rigorously show that a large family of monotone quantities along the weak inverse mean curvature flow is the limit case of the corresponding ones along the level sets of $p$-capacitary potentials. Such monotone quantities include Willmore and Minkowski-type functionals on Riemannian manifolds with nonnegative Ricci curvature. In $3$-dimensional manifolds with nonnegative scalar curvature, we also recover the monotonicity of the Hawking mass and its nonlinear potential theoretic counterparts. This unified view is built on a refined analysis of $p$-capacitary potentials. We prove that they strongly converge in $W^{1,q}_{\mathrm{loc}}$ as $p\to 1^+$ to the inverse mean curvature flow and their level sets are curvature varifolds. Finally, we also deduce a Gauss-Bonnet-type theorem for level sets of $p$-capacitary potentials.

math.DG↗

A remark on the reach and upper bounds on some extrinsic geometry invariants of submanifolds

We consider a compact submanifold $M$ of a Riemannian manifold $N$ and we use the second variation formula as a tool to drive some geometric results on reach$(M, N)$ the reach of $M$ in $N$, including some useful relations between the extrinsic geometry of $M$ in $N$ and reach$(M, N)$. Our results generalize some theorems previously proved for the special case where $N$ is Euclidean space.

math.DG↗