{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TZRN6NGVRFZT426ZTGDGBZ7DCE","short_pith_number":"pith:TZRN6NGV","schema_version":"1.0","canonical_sha256":"9e62df34d589733e6bd9998660e7e3112ef064be8d2b66d6fcdc12a97c476390","source":{"kind":"arxiv","id":"2602.03067","version":3},"attestation_state":"computed","paper":{"title":"FlashSinkhorn: IO-Aware Entropic Optimal Transport on GPU","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NA","math.NA"],"primary_cat":"cs.LG","authors_text":"An Yu, Davis Wertheimer, Felix X.-F. Ye, Linsong Chu, Ming-Ching Chang, Xingjie Li","submitted_at":"2026-02-03T03:52:20Z","abstract_excerpt":"Entropic optimal transport (EOT) via Sinkhorn iterations is widely used in modern machine learning, yet GPU solvers remain inefficient at scale. Tensorized implementations suffer quadratic HBM traffic from dense $n\\times m$ interactions, while existing online backends avoid storing dense matrices but still rely on generic tiled map-reduce reduction kernels with limited fusion. We present \\textbf{FlashSinkhorn}, an IO-aware EOT solver for squared Euclidean cost that rewrites stabilized log-domain Sinkhorn updates as row-wise LogSumExp reductions of biased dot-product scores, the same normalizat"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2602.03067","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-03T03:52:20Z","cross_cats_sorted":["cs.AI","cs.NA","math.NA"],"title_canon_sha256":"20dc4b1873eba0d3f9d7caab2f15e5c2518ceddd2abc2600c638233758219914","abstract_canon_sha256":"19269b113da12d101464300dbef1f4191418f0bda15c8772f2317997c65debc3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:03:56.173716Z","signature_b64":"Czg73RzUj6MDEooCJs+2U8pyynTDeM6mybMVY6XhOSl0Wk7o+D0NnvH/m03guL/tYMAWO78wYGpH/OtZjZTBCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e62df34d589733e6bd9998660e7e3112ef064be8d2b66d6fcdc12a97c476390","last_reissued_at":"2026-05-22T01:03:56.172941Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:03:56.172941Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FlashSinkhorn: IO-Aware Entropic Optimal Transport on GPU","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.NA","math.NA"],"primary_cat":"cs.LG","authors_text":"An Yu, Davis Wertheimer, Felix X.-F. Ye, Linsong Chu, Ming-Ching Chang, Xingjie Li","submitted_at":"2026-02-03T03:52:20Z","abstract_excerpt":"Entropic optimal transport (EOT) via Sinkhorn iterations is widely used in modern machine learning, yet GPU solvers remain inefficient at scale. Tensorized implementations suffer quadratic HBM traffic from dense $n\\times m$ interactions, while existing online backends avoid storing dense matrices but still rely on generic tiled map-reduce reduction kernels with limited fusion. We present \\textbf{FlashSinkhorn}, an IO-aware EOT solver for squared Euclidean cost that rewrites stabilized log-domain Sinkhorn updates as row-wise LogSumExp reductions of biased dot-product scores, the same normalizat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.03067","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2602.03067/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2602.03067","created_at":"2026-05-22T01:03:56.173042+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.03067v3","created_at":"2026-05-22T01:03:56.173042+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.03067","created_at":"2026-05-22T01:03:56.173042+00:00"},{"alias_kind":"pith_short_12","alias_value":"TZRN6NGVRFZT","created_at":"2026-05-22T01:03:56.173042+00:00"},{"alias_kind":"pith_short_16","alias_value":"TZRN6NGVRFZT426Z","created_at":"2026-05-22T01:03:56.173042+00:00"},{"alias_kind":"pith_short_8","alias_value":"TZRN6NGV","created_at":"2026-05-22T01:03:56.173042+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.18389","citing_title":"Spherical Harmonic Optimal Transport: Application to Climate Models Comparisons","ref_index":58,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12879","citing_title":"ASAP: Amortized Doubly-Stochastic Attention via Sliced Dual Projection","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE","json":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE.json","graph_json":"https://pith.science/api/pith-number/TZRN6NGVRFZT426ZTGDGBZ7DCE/graph.json","events_json":"https://pith.science/api/pith-number/TZRN6NGVRFZT426ZTGDGBZ7DCE/events.json","paper":"https://pith.science/paper/TZRN6NGV"},"agent_actions":{"view_html":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE","download_json":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE.json","view_paper":"https://pith.science/paper/TZRN6NGV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.03067&json=true","fetch_graph":"https://pith.science/api/pith-number/TZRN6NGVRFZT426ZTGDGBZ7DCE/graph.json","fetch_events":"https://pith.science/api/pith-number/TZRN6NGVRFZT426ZTGDGBZ7DCE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE/action/storage_attestation","attest_author":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE/action/author_attestation","sign_citation":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE/action/citation_signature","submit_replication":"https://pith.science/pith/TZRN6NGVRFZT426ZTGDGBZ7DCE/action/replication_record"}},"created_at":"2026-05-22T01:03:56.173042+00:00","updated_at":"2026-05-22T01:03:56.173042+00:00"}