{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:SAVWVMXB5GDUKV56RAIA5MD5JL","short_pith_number":"pith:SAVWVMXB","schema_version":"1.0","canonical_sha256":"902b6ab2e1e9874557be88100eb07d4aeddfbe377297b7f8d1814b7bcbeaba43","source":{"kind":"arxiv","id":"2406.05033","version":2},"attestation_state":"computed","paper":{"title":"Gradient Descent on Logistic Regression with Non-Separable Data and Large Step Sizes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Antonio Orvieto, Christopher De Sa, Daniel Yiming Cao, Si Yi Meng","submitted_at":"2024-06-07T15:53:06Z","abstract_excerpt":"We study gradient descent (GD) dynamics on logistic regression problems with large, constant step sizes. For linearly-separable data, it is known that GD converges to the minimizer with arbitrarily large step sizes, a property which no longer holds when the problem is not separable. In fact, the behaviour can be much more complex -- a sequence of period-doubling bifurcations begins at the critical step size $2/\\lambda$, where $\\lambda$ is the largest eigenvalue of the Hessian at the solution. Using a smaller-than-critical step size guarantees convergence if initialized nearby the solution: but"},"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":"2406.05033","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-07T15:53:06Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"5523a04ac09d1c5750f78b969dff3ba55f961247e60ceb433440fde4b23d3c71","abstract_canon_sha256":"c07b251bf2e4f667c65db9aa244160403bfe12350e509b433c4880d6812e47f2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:30:16.189472Z","signature_b64":"84U4IB0XuNwblKcRrrsgm1ve6+PpC+Zdkh6OAFD7iB1RGGI2zl/fqN93gDR0rE70iXi1KFjPjCLzPjNSoafDDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"902b6ab2e1e9874557be88100eb07d4aeddfbe377297b7f8d1814b7bcbeaba43","last_reissued_at":"2026-07-05T09:30:16.188962Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:30:16.188962Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradient Descent on Logistic Regression with Non-Separable Data and Large Step Sizes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Antonio Orvieto, Christopher De Sa, Daniel Yiming Cao, Si Yi Meng","submitted_at":"2024-06-07T15:53:06Z","abstract_excerpt":"We study gradient descent (GD) dynamics on logistic regression problems with large, constant step sizes. For linearly-separable data, it is known that GD converges to the minimizer with arbitrarily large step sizes, a property which no longer holds when the problem is not separable. In fact, the behaviour can be much more complex -- a sequence of period-doubling bifurcations begins at the critical step size $2/\\lambda$, where $\\lambda$ is the largest eigenvalue of the Hessian at the solution. Using a smaller-than-critical step size guarantees convergence if initialized nearby the solution: but"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.05033","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/2406.05033/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":"2406.05033","created_at":"2026-07-05T09:30:16.189023+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.05033v2","created_at":"2026-07-05T09:30:16.189023+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.05033","created_at":"2026-07-05T09:30:16.189023+00:00"},{"alias_kind":"pith_short_12","alias_value":"SAVWVMXB5GDU","created_at":"2026-07-05T09:30:16.189023+00:00"},{"alias_kind":"pith_short_16","alias_value":"SAVWVMXB5GDUKV56","created_at":"2026-07-05T09:30:16.189023+00:00"},{"alias_kind":"pith_short_8","alias_value":"SAVWVMXB","created_at":"2026-07-05T09:30:16.189023+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21683","citing_title":"Finite-Sample Performance of Gradient Descent in Logistic Regression with Gaussian Design","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31386","citing_title":"Stepsize Hedging: an Alternative Mechanism for Accelerating Gradient Descent","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL","json":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL.json","graph_json":"https://pith.science/api/pith-number/SAVWVMXB5GDUKV56RAIA5MD5JL/graph.json","events_json":"https://pith.science/api/pith-number/SAVWVMXB5GDUKV56RAIA5MD5JL/events.json","paper":"https://pith.science/paper/SAVWVMXB"},"agent_actions":{"view_html":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL","download_json":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL.json","view_paper":"https://pith.science/paper/SAVWVMXB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.05033&json=true","fetch_graph":"https://pith.science/api/pith-number/SAVWVMXB5GDUKV56RAIA5MD5JL/graph.json","fetch_events":"https://pith.science/api/pith-number/SAVWVMXB5GDUKV56RAIA5MD5JL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL/action/storage_attestation","attest_author":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL/action/author_attestation","sign_citation":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL/action/citation_signature","submit_replication":"https://pith.science/pith/SAVWVMXB5GDUKV56RAIA5MD5JL/action/replication_record"}},"created_at":"2026-07-05T09:30:16.189023+00:00","updated_at":"2026-07-05T09:30:16.189023+00:00"}