One-Shot Generative Flows: Existence and Obstructions
Panos Tsimpos, Daniel Sharp, Youssef Marzouk

TL;DR
This paper investigates the existence of straight-line flows in measure transport for generative modeling, revealing conditions under which such flows are possible or obstructed, especially for Gaussian versus multi-modal targets.
Contribution
It provides a structural theory characterizing when straight-line generative flows can exist, including explicit constructions for Gaussian endpoints and impossibility results for multi-modal targets.
Findings
Explicit straight-line processes for Gaussian endpoints.
Impossibility of straight-line flows for targets with well-separated modes.
Fundamental relationship between process behavior and flow geometry.
Abstract
We study dynamic measure transport for generative modeling, focusing on transport maps that connect a source measure to a target measure by integrating a velocity field of the form , where is a stochastic process satisfying and is its time derivative. We investigate when induces a \emph{straight-line flow}: a flow whose pointwise acceleration vanishes and is therefore exactly integrable by any first-order method. First, we develop multiple characterizations of straight-line flows in terms of PDEs involving the conditional statistics of the process. Then, we prove that straight-line flows under endpoint independence exhibit a sharp dichotomy. On the one hand, we construct explicit, computable straight-line processes for arbitrary Gaussian endpoints. On the…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
