Metraipolis › archive › dc_160d2a3eef29

FlashSinkhorn 2: Block-Sparse Entropic Optimal Transport — arXiv:2610.02395

Source
arXiv
Original
https://arxiv.org/abs/2610.02395
License (as archived)
CC0-1.0
Archived
2026-10-05T22:48:56.634Z
SHA-256 of the archived text
1bb9cc39734021fbfe1bcf03aef54a21b677b0951fefea2dd64164d3130b9cf1
Stable URL of this page
https://metraipolis.com/documento/dc_160d2a3eef29

How to cite

FlashSinkhorn 2: Block-Sparse Entropic Optimal Transport — arXiv:2610.02395 — arXiv, https://arxiv.org/abs/2610.02395 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_160d2a3eef29 (Metraipolis archive dc_160d2a3eef29, sha256 1bb9cc39734021fbfe1bcf03aef54a21b677b0951fefea2dd64164d3130b9cf1)
BibTeX
@misc{dc160d2a3eef29,
  title = {FlashSinkhorn 2: Block-Sparse Entropic Optimal Transport — arXiv:2610.02395},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.02395},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_160d2a3eef29; sha256 1bb9cc39734021fbfe1bcf03aef54a21b677b0951fefea2dd64164d3130b9cf1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.02395. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Streaming GPU solvers for entropic optimal transport (EOT), such as FlashSinkhorn, avoid storing the dense kernel but still evaluate all $n\times m$ point pairs in every Sinkhorn iteration. We present \textbf{FlashSinkhorn~2} (FS2), a solver for squared-Euclidean cost on low-dimensional point clouds that solves large discrete EOT problems to a prescribed marginal residual on a single GPU by coupling two stages. A coarse stage solves on cell centroids, lifts the potentials to every point and, when a sampled marginal check rejects the lift, continues on the centroids, replacing most point-level updates. A block-sparse fine stage then removes the centroid error that coarse updates cannot. Its Morton-ordered blocks support screening and fused tensor-core execution, and a threshold set by the block masses bounds each omitted tile's contribution to every row and column. On synthetic benchmarks, FS2 reaches the target residual on all 32 problems and GeomLoss multiscale on 10. On one A100, FS2 solves discrete EOT between two $1.34\times10^8$-particle measures from a cosmological $N$-body simulation, at an entropic blur equal to the mean interparticle distance, to an all-particle marginal residual below 0.01 in under 2.5 hours. To our knowledge, it is the largest discrete EOT problem solved to this accuracy within hours. For reproducibility, we release an open-source implementation at this https URL

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Instrumentation and Methods for Astrophysics (astro-ph.IM); Numerical Analysis (math.NA).
Pagina del lavoro: https://arxiv.org/abs/2610.02395 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-05T22:45:55.580Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).