{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:ZS2SKAOOPR7PRYBOJLEYQNVX2I","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"1e3b5b5f0d85cc6d5838d159f8430a92acd19e59916cfe0e1b08dc4e9ac8fc9f","cross_cats_sorted":["math.DS","math.OC","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-16T06:23:18Z","title_canon_sha256":"579a22f9f3d9dd0870be92dac9c81190f376974fd9f771bf37ac8126983d8a4f"},"schema_version":"1.0","source":{"id":"2604.14669","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.14669","created_at":"2026-07-02T01:17:31Z"},{"alias_kind":"arxiv_version","alias_value":"2604.14669v2","created_at":"2026-07-02T01:17:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.14669","created_at":"2026-07-02T01:17:31Z"},{"alias_kind":"pith_short_12","alias_value":"ZS2SKAOOPR7P","created_at":"2026-07-02T01:17:31Z"},{"alias_kind":"pith_short_16","alias_value":"ZS2SKAOOPR7PRYBO","created_at":"2026-07-02T01:17:31Z"},{"alias_kind":"pith_short_8","alias_value":"ZS2SKAOO","created_at":"2026-07-02T01:17:31Z"}],"graph_snapshots":[{"event_id":"sha256:dbd1ac5c3cb7e246cd701a92b55ee844abbee9f05fc9e1425f815e4e6be3e884","target":"graph","created_at":"2026-07-02T01:17:31Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"we provide an explicit step size condition that exactly captures the (mean-square) linear stability of a family of ZO methods based on the standard two-point estimator... mean-square stability of ZO methods depends on the entire Hessian spectrum."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The linear stability analysis assumes the loss can be locally approximated by a quadratic form whose Hessian is constant over the relevant trajectory, and that the two-point estimator is used without additional smoothing or variance reduction."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Zeroth-order methods achieve mean-square stability when the step size satisfies a condition involving the entire Hessian spectrum, with full-batch ZO optimizers operating at the edge of stability and large steps regularizing the Hessian trace."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Zeroth-order methods remain mean-square stable only when their step size satisfies a bound that depends on the full Hessian spectrum rather than its largest eigenvalue alone."}],"snapshot_sha256":"532643a344f2c4571bf4ea935d01ee819ff3b696b8d96b7c29d30a87f1eb086c"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.14669/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Zeroth-order (ZO) methods are widely used when gradients are unavailable or prohibitively expensive, including black-box learning and memory-efficient fine-tuning of large models, yet their optimization dynamics in deep learning remain underexplored. In this work, we provide an explicit step size condition that exactly captures the (mean-square) linear stability of a family of ZO methods based on the standard two-point estimator. Our characterization reveals a sharp contrast with first-order (FO) methods: whereas FO stability is governed solely by the largest Hessian eigenvalue, mean-square st","authors_text":"Bingcong Li, Liang Zhang, Michael Muehlebach, Minhak Song, Niao He, Sewoong Oh","cross_cats":["math.DS","math.OC","stat.ML"],"headline":"Zeroth-order methods remain mean-square stable only when their step size satisfies a bound that depends on the full Hessian spectrum rather than its largest eigenvalue alone.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-16T06:23:18Z","title":"Zeroth-Order Optimization at the Edge of Stability"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.14669","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T11:15:27.370755Z","id":"64c66983-492d-41d4-bcd9-fbb0eec76e7d","model_set":{"reader":"grok-4.3"},"one_line_summary":"Zeroth-order methods achieve mean-square stability when the step size satisfies a condition involving the entire Hessian spectrum, with full-batch ZO optimizers operating at the edge of stability and large steps regularizing the Hessian trace.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Zeroth-order methods remain mean-square stable only when their step size satisfies a bound that depends on the full Hessian spectrum rather than its largest eigenvalue alone.","strongest_claim":"we provide an explicit step size condition that exactly captures the (mean-square) linear stability of a family of ZO methods based on the standard two-point estimator... mean-square stability of ZO methods depends on the entire Hessian spectrum.","weakest_assumption":"The linear stability analysis assumes the loss can be locally approximated by a quadratic form whose Hessian is constant over the relevant trajectory, and that the two-point estimator is used without additional smoothing or variance reduction."}},"verdict_id":"64c66983-492d-41d4-bcd9-fbb0eec76e7d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e6431ab9d749cce605f593f688d52e01440a337d567261cdffb8b988a3eb7dcf","target":"record","created_at":"2026-07-02T01:17:31Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"1e3b5b5f0d85cc6d5838d159f8430a92acd19e59916cfe0e1b08dc4e9ac8fc9f","cross_cats_sorted":["math.DS","math.OC","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-16T06:23:18Z","title_canon_sha256":"579a22f9f3d9dd0870be92dac9c81190f376974fd9f771bf37ac8126983d8a4f"},"schema_version":"1.0","source":{"id":"2604.14669","kind":"arxiv","version":2}},"canonical_sha256":"ccb52501ce7c7ef8e02e4ac98836b7d209df2144765ef2366c12dc9fb3ae21d1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ccb52501ce7c7ef8e02e4ac98836b7d209df2144765ef2366c12dc9fb3ae21d1","first_computed_at":"2026-07-02T01:17:31.259570Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-02T01:17:31.259570Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"rKPFTcflGMyZ8SM351XvgJFvZAdW8YQU1wEdZA57rR9ry20NAkyVq7q4wmizj3pfjcRdIzAwskYiAU6sDd0CBQ==","signature_status":"signed_v1","signed_at":"2026-07-02T01:17:31.260011Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.14669","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e6431ab9d749cce605f593f688d52e01440a337d567261cdffb8b988a3eb7dcf","sha256:dbd1ac5c3cb7e246cd701a92b55ee844abbee9f05fc9e1425f815e4e6be3e884"],"state_sha256":"1b539e5308621bae831dae93482357c20f215c653ca56d736f1314ac2bb6b7ca"}