{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7MONRX5XZETBG3C6GMGVNE3SAG","short_pith_number":"pith:7MONRX5X","schema_version":"1.0","canonical_sha256":"fb1cd8dfb7c926136c5e330d569372019af8941c71552abb331f822b4f20f640","source":{"kind":"arxiv","id":"2503.10537","version":2},"attestation_state":"computed","paper":{"title":"Structured Preconditioners in Adaptive Optimization: A Unified Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Sanjiv Kumar, Sashank Reddi, Shuo Xie, Tianhao Wang, Zhiyuan Li","submitted_at":"2025-03-13T16:51:59Z","abstract_excerpt":"We present a novel unified analysis for a broad class of adaptive optimization algorithms with structured (e.g., layerwise, diagonal, and kronecker-factored) preconditioners for both online regret minimization and offline convex optimization. Our analysis not only provides matching rate to several important structured preconditioned algorithms including diagonal AdaGrad, full-matrix AdaGrad, and AdaGrad-Norm, but also gives an improved convergence rate for a one-sided variant of Shampoo over that of original Shampoo. Interestingly, more structured preconditioners (e.g., diagonal Adagrad, AdaGr"},"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":"2503.10537","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-13T16:51:59Z","cross_cats_sorted":[],"title_canon_sha256":"88885c34d1dc26780f2d11002b76f6cbfc8a298e3c72b462d114208b1032678f","abstract_canon_sha256":"d25914ad4f885007dcb740961dad6073826b486fc9cd15ccb1d4e186f5fe1faa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:37:16.879746Z","signature_b64":"WnlW8PPr81Dc7Ez47u8+gI4JN2lNrCdBsZKBa9chCnrmqmSOK+lZr+H/pu8R1qnuC5FoXB9CVYEPEb8/JlIiBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fb1cd8dfb7c926136c5e330d569372019af8941c71552abb331f822b4f20f640","last_reissued_at":"2026-07-05T11:37:16.879248Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:37:16.879248Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Structured Preconditioners in Adaptive Optimization: A Unified Analysis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Sanjiv Kumar, Sashank Reddi, Shuo Xie, Tianhao Wang, Zhiyuan Li","submitted_at":"2025-03-13T16:51:59Z","abstract_excerpt":"We present a novel unified analysis for a broad class of adaptive optimization algorithms with structured (e.g., layerwise, diagonal, and kronecker-factored) preconditioners for both online regret minimization and offline convex optimization. Our analysis not only provides matching rate to several important structured preconditioned algorithms including diagonal AdaGrad, full-matrix AdaGrad, and AdaGrad-Norm, but also gives an improved convergence rate for a one-sided variant of Shampoo over that of original Shampoo. Interestingly, more structured preconditioners (e.g., diagonal Adagrad, AdaGr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.10537","kind":"arxiv","version":2},"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/2503.10537/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":"2503.10537","created_at":"2026-07-05T11:37:16.879305+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.10537v2","created_at":"2026-07-05T11:37:16.879305+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.10537","created_at":"2026-07-05T11:37:16.879305+00:00"},{"alias_kind":"pith_short_12","alias_value":"7MONRX5XZETB","created_at":"2026-07-05T11:37:16.879305+00:00"},{"alias_kind":"pith_short_16","alias_value":"7MONRX5XZETBG3C6","created_at":"2026-07-05T11:37:16.879305+00:00"},{"alias_kind":"pith_short_8","alias_value":"7MONRX5X","created_at":"2026-07-05T11:37:16.879305+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.10777","citing_title":"Preconditioned Norms: A Unified Framework for Steepest Descent, Quasi-Newton and Adaptive Methods","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2505.23737","citing_title":"On the Convergence Analysis of Muon","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2509.11983","citing_title":"Low-rank Orthogonalization for Large-scale Matrix Optimization with Applications to Foundation Model Training","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2603.26554","citing_title":"Sharp Capacity Scaling of Spectral Optimizers in Learning Associative Memory","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02505","citing_title":"Optimal Projection-Free Adaptive SGD for Matrix Optimization","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06654","citing_title":"Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06316","citing_title":"Pro-KLShampoo: Projected KL-Shampoo with Whitening Recovered by Orthogonalization","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08742","citing_title":"Adam-HNAG: A Convergent Reformulation of Adam with Accelerated Rate","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17423","citing_title":"A unified convergence theory for adaptive first-order methods in the nonconvex case, including AdaNorm, full and diagonal AdaGrad, Shampoo and Muo","ref_index":54,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG","json":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG.json","graph_json":"https://pith.science/api/pith-number/7MONRX5XZETBG3C6GMGVNE3SAG/graph.json","events_json":"https://pith.science/api/pith-number/7MONRX5XZETBG3C6GMGVNE3SAG/events.json","paper":"https://pith.science/paper/7MONRX5X"},"agent_actions":{"view_html":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG","download_json":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG.json","view_paper":"https://pith.science/paper/7MONRX5X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.10537&json=true","fetch_graph":"https://pith.science/api/pith-number/7MONRX5XZETBG3C6GMGVNE3SAG/graph.json","fetch_events":"https://pith.science/api/pith-number/7MONRX5XZETBG3C6GMGVNE3SAG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG/action/storage_attestation","attest_author":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG/action/author_attestation","sign_citation":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG/action/citation_signature","submit_replication":"https://pith.science/pith/7MONRX5XZETBG3C6GMGVNE3SAG/action/replication_record"}},"created_at":"2026-07-05T11:37:16.879305+00:00","updated_at":"2026-07-05T11:37:16.879305+00:00"}