{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:44KTV2RRA3GGMWDT2T2WGIMVIE","short_pith_number":"pith:44KTV2RR","schema_version":"1.0","canonical_sha256":"e7153aea3106cc665873d4f5632195413832e780b85a1c6ef5aba00d3eec2d82","source":{"kind":"arxiv","id":"2209.15141","version":2},"attestation_state":"computed","paper":{"title":"On Convergence of Average-Reward Off-Policy Control Algorithms in Weakly Communicating MDPs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Richard S. Sutton, Yi Wan","submitted_at":"2022-09-30T00:07:48Z","abstract_excerpt":"We show two average-reward off-policy control algorithms, Differential Q-learning (Wan, Naik, & Sutton 2021a) and RVI Q-learning (Abounadi Bertsekas & Borkar 2001), converge in weakly communicating MDPs. Weakly communicating MDPs are the most general MDPs that can be solved by a learning algorithm with a single stream of experience. The original convergence proofs of the two algorithms require that the solution set of the average-reward optimality equation only has one degree of freedom, which is not necessarily true for weakly communicating MDPs. To the best of our knowledge, our results are "},"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":"2209.15141","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-09-30T00:07:48Z","cross_cats_sorted":[],"title_canon_sha256":"47a8148a8407b967a44c9b39aee272f04986f1fe72d5f9da9a0ac42b0abe4cdc","abstract_canon_sha256":"e0e093adf2ea5afaa44098a1296dd4b4f945ea84d7a3fd3b24fe37b990238d8b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:13:38.158342Z","signature_b64":"EG6Wv3cxu7YpwdU0fRPm/dvN8mqdxwW+XTEay8Y+WTaEZmD0epc2pFC+gdMztTqMJTmzhsYeguQ/7KpcHQeIBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e7153aea3106cc665873d4f5632195413832e780b85a1c6ef5aba00d3eec2d82","last_reissued_at":"2026-07-05T05:13:38.157867Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:13:38.157867Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On Convergence of Average-Reward Off-Policy Control Algorithms in Weakly Communicating MDPs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Richard S. Sutton, Yi Wan","submitted_at":"2022-09-30T00:07:48Z","abstract_excerpt":"We show two average-reward off-policy control algorithms, Differential Q-learning (Wan, Naik, & Sutton 2021a) and RVI Q-learning (Abounadi Bertsekas & Borkar 2001), converge in weakly communicating MDPs. Weakly communicating MDPs are the most general MDPs that can be solved by a learning algorithm with a single stream of experience. The original convergence proofs of the two algorithms require that the solution set of the average-reward optimality equation only has one degree of freedom, which is not necessarily true for weakly communicating MDPs. To the best of our knowledge, our results are "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.15141","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/2209.15141/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":"2209.15141","created_at":"2026-07-05T05:13:38.157919+00:00"},{"alias_kind":"arxiv_version","alias_value":"2209.15141v2","created_at":"2026-07-05T05:13:38.157919+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.15141","created_at":"2026-07-05T05:13:38.157919+00:00"},{"alias_kind":"pith_short_12","alias_value":"44KTV2RRA3GG","created_at":"2026-07-05T05:13:38.157919+00:00"},{"alias_kind":"pith_short_16","alias_value":"44KTV2RRA3GGMWDT","created_at":"2026-07-05T05:13:38.157919+00:00"},{"alias_kind":"pith_short_8","alias_value":"44KTV2RR","created_at":"2026-07-05T05:13:38.157919+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04029","citing_title":"Position: Deployed Reinforcement Learning should be Continual","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE","json":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE.json","graph_json":"https://pith.science/api/pith-number/44KTV2RRA3GGMWDT2T2WGIMVIE/graph.json","events_json":"https://pith.science/api/pith-number/44KTV2RRA3GGMWDT2T2WGIMVIE/events.json","paper":"https://pith.science/paper/44KTV2RR"},"agent_actions":{"view_html":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE","download_json":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE.json","view_paper":"https://pith.science/paper/44KTV2RR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2209.15141&json=true","fetch_graph":"https://pith.science/api/pith-number/44KTV2RRA3GGMWDT2T2WGIMVIE/graph.json","fetch_events":"https://pith.science/api/pith-number/44KTV2RRA3GGMWDT2T2WGIMVIE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE/action/storage_attestation","attest_author":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE/action/author_attestation","sign_citation":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE/action/citation_signature","submit_replication":"https://pith.science/pith/44KTV2RRA3GGMWDT2T2WGIMVIE/action/replication_record"}},"created_at":"2026-07-05T05:13:38.157919+00:00","updated_at":"2026-07-05T05:13:38.157919+00:00"}