{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DIUNESJXV3LCVKC5E2ABI2CX6J","short_pith_number":"pith:DIUNESJX","schema_version":"1.0","canonical_sha256":"1a28d24937aed62aa85d2680146857f2451ca749750c7fa9aed163c9aea96338","source":{"kind":"arxiv","id":"2608.10204","version":1},"attestation_state":"computed","paper":{"title":"Boundary-Seeking Policy Gradient for Safe Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chenhua Fan, Honghao Wei, Jiahui Zhu, Yuhang Zhang","submitted_at":"2026-08-10T20:20:01Z","abstract_excerpt":"Safe reinforcement learning maximizes reward subject to safety constraints. For Constrained Markov Decision Processes, the linear-programming view over occupancy measures implies that whenever the constraint is active at optimality, the optimal policy lies exactly on the constraint boundary, yet standard gradient-based methods do not exploit this structure and often settle in the feasible interior. We introduce Boundary-Seeking Policy Gradient (BSPG), a first-order method whose update combines a tangential component that improves reward while preserving cost to first order with a signed, resid"},"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":"2608.10204","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-08-10T20:20:01Z","cross_cats_sorted":[],"title_canon_sha256":"cca53521182367838dbadcb81c688b0e7581c5448163c4ffb25bcb5fb25f631b","abstract_canon_sha256":"2167452e6d880f4d503208b54e4b8fe38341aafeefb42b229225db094cc8e23e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-12T00:23:03.358882Z","signature_b64":"ShLuQu8XLRSu1lYsWxidM9RQwhoZeUgirZ2WXUBFbx1UwecfKT3dBNVzUXHeAVB3loPvZgoFypb1Zr66WsK+Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1a28d24937aed62aa85d2680146857f2451ca749750c7fa9aed163c9aea96338","last_reissued_at":"2026-08-12T00:23:03.357210Z","signature_status":"signed_v1","first_computed_at":"2026-08-12T00:23:03.357210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Boundary-Seeking Policy Gradient for Safe Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chenhua Fan, Honghao Wei, Jiahui Zhu, Yuhang Zhang","submitted_at":"2026-08-10T20:20:01Z","abstract_excerpt":"Safe reinforcement learning maximizes reward subject to safety constraints. For Constrained Markov Decision Processes, the linear-programming view over occupancy measures implies that whenever the constraint is active at optimality, the optimal policy lies exactly on the constraint boundary, yet standard gradient-based methods do not exploit this structure and often settle in the feasible interior. We introduce Boundary-Seeking Policy Gradient (BSPG), a first-order method whose update combines a tangential component that improves reward while preserving cost to first order with a signed, resid"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.10204","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/2608.10204/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":"2608.10204","created_at":"2026-08-12T00:23:03.359970+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.10204v1","created_at":"2026-08-12T00:23:03.359970+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.10204","created_at":"2026-08-12T00:23:03.359970+00:00"},{"alias_kind":"pith_short_12","alias_value":"DIUNESJXV3LC","created_at":"2026-08-12T00:23:03.359970+00:00"},{"alias_kind":"pith_short_16","alias_value":"DIUNESJXV3LCVKC5","created_at":"2026-08-12T00:23:03.359970+00:00"},{"alias_kind":"pith_short_8","alias_value":"DIUNESJX","created_at":"2026-08-12T00:23:03.359970+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J","json":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J.json","graph_json":"https://pith.science/api/pith-number/DIUNESJXV3LCVKC5E2ABI2CX6J/graph.json","events_json":"https://pith.science/api/pith-number/DIUNESJXV3LCVKC5E2ABI2CX6J/events.json","paper":"https://pith.science/paper/DIUNESJX"},"agent_actions":{"view_html":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J","download_json":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J.json","view_paper":"https://pith.science/paper/DIUNESJX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.10204&json=true","fetch_graph":"https://pith.science/api/pith-number/DIUNESJXV3LCVKC5E2ABI2CX6J/graph.json","fetch_events":"https://pith.science/api/pith-number/DIUNESJXV3LCVKC5E2ABI2CX6J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J/action/storage_attestation","attest_author":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J/action/author_attestation","sign_citation":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J/action/citation_signature","submit_replication":"https://pith.science/pith/DIUNESJXV3LCVKC5E2ABI2CX6J/action/replication_record"}},"created_at":"2026-08-12T00:23:03.359970+00:00","updated_at":"2026-08-12T00:23:03.359970+00:00"}