{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:GT3JOTKWUBMGA5EVDWDAYZ5GCH","short_pith_number":"pith:GT3JOTKW","schema_version":"1.0","canonical_sha256":"34f6974d56a0586074951d860c67a611d0381f68ec6d541ba8c3e0e579dace4b","source":{"kind":"arxiv","id":"2111.00552","version":2},"attestation_state":"computed","paper":{"title":"Policy Optimization for Constrained MDPs with Provable Fast Global Convergence","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Chao Tian, Dileep Kalathil, P. R. Kumar, Ruida Zhou, Tao Liu","submitted_at":"2021-10-31T17:46:26Z","abstract_excerpt":"We address the problem of finding the optimal policy of a constrained Markov decision process (CMDP) using a gradient descent-based algorithm. Previous results have shown that a primal-dual approach can achieve an $\\mathcal{O}(1/\\sqrt{T})$ global convergence rate for both the optimality gap and the constraint violation. We propose a new algorithm called policy mirror descent-primal dual (PMD-PD) algorithm that can provably achieve a faster $\\mathcal{O}(\\log(T)/T)$ convergence rate for both the optimality gap and the constraint violation. For the primal (policy) update, the PMD-PD algorithm uti"},"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":"2111.00552","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-10-31T17:46:26Z","cross_cats_sorted":["cs.AI","math.OC"],"title_canon_sha256":"c6f6f7f46225da5645238b63147b00b5d445600ca369bbd9ffe174a6a9ed005c","abstract_canon_sha256":"a582b72c4320af28d247386e8e8ec271d65dbc8fa83d7f14ade77668a4786007"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:54:03.354103Z","signature_b64":"1bje0IzsvIWfekPg2pizA25ilHc4Yuvgr+MdFDR5oyZORU8yJADsOrUVQnpZK49EbpCyoXQ6I/8FK/U/cp2/Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"34f6974d56a0586074951d860c67a611d0381f68ec6d541ba8c3e0e579dace4b","last_reissued_at":"2026-07-05T03:54:03.353688Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:54:03.353688Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Policy Optimization for Constrained MDPs with Provable Fast Global Convergence","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Chao Tian, Dileep Kalathil, P. R. Kumar, Ruida Zhou, Tao Liu","submitted_at":"2021-10-31T17:46:26Z","abstract_excerpt":"We address the problem of finding the optimal policy of a constrained Markov decision process (CMDP) using a gradient descent-based algorithm. Previous results have shown that a primal-dual approach can achieve an $\\mathcal{O}(1/\\sqrt{T})$ global convergence rate for both the optimality gap and the constraint violation. We propose a new algorithm called policy mirror descent-primal dual (PMD-PD) algorithm that can provably achieve a faster $\\mathcal{O}(\\log(T)/T)$ convergence rate for both the optimality gap and the constraint violation. For the primal (policy) update, the PMD-PD algorithm uti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.00552","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/2111.00552/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":"2111.00552","created_at":"2026-07-05T03:54:03.353743+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.00552v2","created_at":"2026-07-05T03:54:03.353743+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.00552","created_at":"2026-07-05T03:54:03.353743+00:00"},{"alias_kind":"pith_short_12","alias_value":"GT3JOTKWUBMG","created_at":"2026-07-05T03:54:03.353743+00:00"},{"alias_kind":"pith_short_16","alias_value":"GT3JOTKWUBMGA5EV","created_at":"2026-07-05T03:54:03.353743+00:00"},{"alias_kind":"pith_short_8","alias_value":"GT3JOTKW","created_at":"2026-07-05T03:54:03.353743+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2408.11513","citing_title":"Last-Iterate Convergence of General Parameterized Policies in Constrained MDPs","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18841","citing_title":"From Cumulative Constraints to Adaptive Runtime Safety Control for Nonstationary Reinforcement Learning","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18842","citing_title":"Safe Continual Reinforcement Learning under Nonstationarity via Adaptive Safety Constraints","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH","json":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH.json","graph_json":"https://pith.science/api/pith-number/GT3JOTKWUBMGA5EVDWDAYZ5GCH/graph.json","events_json":"https://pith.science/api/pith-number/GT3JOTKWUBMGA5EVDWDAYZ5GCH/events.json","paper":"https://pith.science/paper/GT3JOTKW"},"agent_actions":{"view_html":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH","download_json":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH.json","view_paper":"https://pith.science/paper/GT3JOTKW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.00552&json=true","fetch_graph":"https://pith.science/api/pith-number/GT3JOTKWUBMGA5EVDWDAYZ5GCH/graph.json","fetch_events":"https://pith.science/api/pith-number/GT3JOTKWUBMGA5EVDWDAYZ5GCH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH/action/storage_attestation","attest_author":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH/action/author_attestation","sign_citation":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH/action/citation_signature","submit_replication":"https://pith.science/pith/GT3JOTKWUBMGA5EVDWDAYZ5GCH/action/replication_record"}},"created_at":"2026-07-05T03:54:03.353743+00:00","updated_at":"2026-07-05T03:54:03.353743+00:00"}