{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:Y5NNMHSPIIYMU435RFPTI2ACMP","short_pith_number":"pith:Y5NNMHSP","schema_version":"1.0","canonical_sha256":"c75ad61e4f4230ca737d895f34680263c3664407a604c4e2def9fd2f76b93c83","source":{"kind":"arxiv","id":"2506.23803","version":1},"attestation_state":"computed","paper":{"title":"SGD with Adaptive Preconditioning: Unified Analysis and Momentum Acceleration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Dmitry Kovalev","submitted_at":"2025-06-30T12:47:10Z","abstract_excerpt":"In this paper, we revisit stochastic gradient descent (SGD) with AdaGrad-type preconditioning. Our contributions are twofold. First, we develop a unified convergence analysis of SGD with adaptive preconditioning under anisotropic or matrix smoothness and noise assumptions. This allows us to recover state-of-the-art convergence results for several popular adaptive gradient methods, including AdaGrad-Norm, AdaGrad, and ASGO/One-sided Shampoo. In addition, we establish the fundamental connection between two recently proposed algorithms, Scion and DASGO, and provide the first theoretical guarantee"},"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":"2506.23803","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-30T12:47:10Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"190586bad58f7c3f934fa3e4921bf81054b8f87059b3b1375e52ef914b376c6b","abstract_canon_sha256":"718f8591f4c06a3230c6398045e80097ddd87d6d8a9d9acc1ddcf60446a1a5d9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:28.147806Z","signature_b64":"vY1ngSr9xqwy0AgJig0Y8f5FB+009Ijxyv40MtmAEaZeSpCg/S+KLcHtaYhjONSRKampWe4nkyCdRrl5mEqAAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c75ad61e4f4230ca737d895f34680263c3664407a604c4e2def9fd2f76b93c83","last_reissued_at":"2026-07-05T11:29:28.147291Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:28.147291Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SGD with Adaptive Preconditioning: Unified Analysis and Momentum Acceleration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Dmitry Kovalev","submitted_at":"2025-06-30T12:47:10Z","abstract_excerpt":"In this paper, we revisit stochastic gradient descent (SGD) with AdaGrad-type preconditioning. Our contributions are twofold. First, we develop a unified convergence analysis of SGD with adaptive preconditioning under anisotropic or matrix smoothness and noise assumptions. This allows us to recover state-of-the-art convergence results for several popular adaptive gradient methods, including AdaGrad-Norm, AdaGrad, and ASGO/One-sided Shampoo. In addition, we establish the fundamental connection between two recently proposed algorithms, Scion and DASGO, and provide the first theoretical guarantee"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.23803","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/2506.23803/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":"2506.23803","created_at":"2026-07-05T11:29:28.147368+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.23803v1","created_at":"2026-07-05T11:29:28.147368+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.23803","created_at":"2026-07-05T11:29:28.147368+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y5NNMHSPIIYM","created_at":"2026-07-05T11:29:28.147368+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y5NNMHSPIIYMU435","created_at":"2026-07-05T11:29:28.147368+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y5NNMHSP","created_at":"2026-07-05T11:29:28.147368+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17364","citing_title":"A Polyak-Ruppert Central Limit Theorem for SA-Adam with Momentum and Non-Convergent Adaptive Preconditioning","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29304","citing_title":"On subspace-constrained preconditioning for randomized iterative methods","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15522","citing_title":"Stochastic Non-Smooth Convex Optimization with Unbounded Gradients","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02505","citing_title":"Optimal Projection-Free Adaptive SGD for Matrix Optimization","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP","json":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP.json","graph_json":"https://pith.science/api/pith-number/Y5NNMHSPIIYMU435RFPTI2ACMP/graph.json","events_json":"https://pith.science/api/pith-number/Y5NNMHSPIIYMU435RFPTI2ACMP/events.json","paper":"https://pith.science/paper/Y5NNMHSP"},"agent_actions":{"view_html":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP","download_json":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP.json","view_paper":"https://pith.science/paper/Y5NNMHSP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.23803&json=true","fetch_graph":"https://pith.science/api/pith-number/Y5NNMHSPIIYMU435RFPTI2ACMP/graph.json","fetch_events":"https://pith.science/api/pith-number/Y5NNMHSPIIYMU435RFPTI2ACMP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP/action/storage_attestation","attest_author":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP/action/author_attestation","sign_citation":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP/action/citation_signature","submit_replication":"https://pith.science/pith/Y5NNMHSPIIYMU435RFPTI2ACMP/action/replication_record"}},"created_at":"2026-07-05T11:29:28.147368+00:00","updated_at":"2026-07-05T11:29:28.147368+00:00"}