{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EWDRT6YF363GPSY3P3TMIFR7KL","short_pith_number":"pith:EWDRT6YF","schema_version":"1.0","canonical_sha256":"258719fb05dfb667cb1b7ee6c4163f52c9456a61bd48afc978663a271e463eb7","source":{"kind":"arxiv","id":"2509.08483","version":1},"attestation_state":"computed","paper":{"title":"Modified Loss of Momentum Gradient Descent: Fine-Grained Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.NA","math.NA","math.OC","stat.CO","stat.ML"],"primary_cat":"cs.LG","authors_text":"Boris Shigida, Matias D. Cattaneo","submitted_at":"2025-09-10T10:47:54Z","abstract_excerpt":"We analyze gradient descent with Polyak heavy-ball momentum (HB) whose fixed momentum parameter $\\beta \\in (0, 1)$ provides exponential decay of memory. Building on Kovachki and Stuart (2021), we prove that on an exponentially attractive invariant manifold the algorithm is exactly plain gradient descent with a modified loss, provided that the step size $h$ is small enough. Although the modified loss does not admit a closed-form expression, we describe it with arbitrary precision and prove global (finite \"time\" horizon) approximation bounds $O(h^{R})$ for any finite order $R \\geq 2$. We then co"},"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":"2509.08483","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-10T10:47:54Z","cross_cats_sorted":["cs.NA","math.NA","math.OC","stat.CO","stat.ML"],"title_canon_sha256":"eb0d6fb2417fce4c0f16877a743e3ede92b4f0bdd9990ad955f9b9bf50e3c442","abstract_canon_sha256":"d9c948ccd61092e7eda8157b627394ba31a6b022c1469521cb687d68713711c3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:08:30.770000Z","signature_b64":"lOvsyYg6V7HuKnZ7N9GbHdSrBeouFYyn8OHtHfAOCV4kfuygqikLXZkOqvpXF1KZVUeXhDSCNg0g3e65eXzECg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"258719fb05dfb667cb1b7ee6c4163f52c9456a61bd48afc978663a271e463eb7","last_reissued_at":"2026-07-05T12:08:30.769476Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:08:30.769476Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Modified Loss of Momentum Gradient Descent: Fine-Grained Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.NA","math.NA","math.OC","stat.CO","stat.ML"],"primary_cat":"cs.LG","authors_text":"Boris Shigida, Matias D. Cattaneo","submitted_at":"2025-09-10T10:47:54Z","abstract_excerpt":"We analyze gradient descent with Polyak heavy-ball momentum (HB) whose fixed momentum parameter $\\beta \\in (0, 1)$ provides exponential decay of memory. Building on Kovachki and Stuart (2021), we prove that on an exponentially attractive invariant manifold the algorithm is exactly plain gradient descent with a modified loss, provided that the step size $h$ is small enough. Although the modified loss does not admit a closed-form expression, we describe it with arbitrary precision and prove global (finite \"time\" horizon) approximation bounds $O(h^{R})$ for any finite order $R \\geq 2$. We then co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.08483","kind":"arxiv","version":1},"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/2509.08483/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":"2509.08483","created_at":"2026-07-05T12:08:30.769549+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.08483v1","created_at":"2026-07-05T12:08:30.769549+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.08483","created_at":"2026-07-05T12:08:30.769549+00:00"},{"alias_kind":"pith_short_12","alias_value":"EWDRT6YF363G","created_at":"2026-07-05T12:08:30.769549+00:00"},{"alias_kind":"pith_short_16","alias_value":"EWDRT6YF363GPSY3","created_at":"2026-07-05T12:08:30.769549+00:00"},{"alias_kind":"pith_short_8","alias_value":"EWDRT6YF","created_at":"2026-07-05T12:08:30.769549+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05350","citing_title":"Approximate Minimax Estimation of a Bounded Normal Mean via Stochastic Mirror Ascent","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.29152","citing_title":"Do Deep Networks Forget Initialization? A Forgetting-Time View of Practical Inductive Bias","ref_index":107,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01642","citing_title":"The Effect of Mini-Batch Noise on the Implicit Bias of Adam","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL","json":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL.json","graph_json":"https://pith.science/api/pith-number/EWDRT6YF363GPSY3P3TMIFR7KL/graph.json","events_json":"https://pith.science/api/pith-number/EWDRT6YF363GPSY3P3TMIFR7KL/events.json","paper":"https://pith.science/paper/EWDRT6YF"},"agent_actions":{"view_html":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL","download_json":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL.json","view_paper":"https://pith.science/paper/EWDRT6YF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.08483&json=true","fetch_graph":"https://pith.science/api/pith-number/EWDRT6YF363GPSY3P3TMIFR7KL/graph.json","fetch_events":"https://pith.science/api/pith-number/EWDRT6YF363GPSY3P3TMIFR7KL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL/action/storage_attestation","attest_author":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL/action/author_attestation","sign_citation":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL/action/citation_signature","submit_replication":"https://pith.science/pith/EWDRT6YF363GPSY3P3TMIFR7KL/action/replication_record"}},"created_at":"2026-07-05T12:08:30.769549+00:00","updated_at":"2026-07-05T12:08:30.769549+00:00"}