{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:O4LLDVR3WZGF7GVEJPAU674BR4","short_pith_number":"pith:O4LLDVR3","schema_version":"1.0","canonical_sha256":"7716b1d63bb64c5f9aa44bc14f7f818f191737cdcfd1bb7a7a309d1e4def5f7c","source":{"kind":"arxiv","id":"2209.15382","version":2},"attestation_state":"computed","paper":{"title":"Linear Convergence for Natural Policy Gradient with Log-linear Policy Parametrization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC","math.ST","stat.TH"],"primary_cat":"cs.LG","authors_text":"Carlo Alfano, Patrick Rebeschini","submitted_at":"2022-09-30T11:17:44Z","abstract_excerpt":"We analyze the convergence rate of the unregularized natural policy gradient algorithm with log-linear policy parametrizations in infinite-horizon discounted Markov decision processes. In the deterministic case, when the Q-value is known and can be approximated by a linear combination of a known feature function up to a bias error, we show that a geometrically-increasing step size yields a linear convergence rate towards an optimal policy. We then consider the sample-based case, when the best representation of the Q- value function among linear combinations of a known feature function is known"},"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":"2209.15382","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-09-30T11:17:44Z","cross_cats_sorted":["math.OC","math.ST","stat.TH"],"title_canon_sha256":"10a6bbe3300863f2ebed1337e40539cdd79356282c72f28e271b8f680510cf84","abstract_canon_sha256":"646cae619b0870276db92ee76b83dcc052a43bc2d0a59f9126ed666976900fe1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:50:50.078556Z","signature_b64":"iGA1zpgnMg51wSzaEmpBqT+92GPoQ5AxoJEgrMqlxLtOe85x1TlYTLPHbsZXu88LUy8khTMzDEGuEwKOWxlSCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7716b1d63bb64c5f9aa44bc14f7f818f191737cdcfd1bb7a7a309d1e4def5f7c","last_reissued_at":"2026-07-05T05:50:50.078016Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:50:50.078016Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Linear Convergence for Natural Policy Gradient with Log-linear Policy Parametrization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC","math.ST","stat.TH"],"primary_cat":"cs.LG","authors_text":"Carlo Alfano, Patrick Rebeschini","submitted_at":"2022-09-30T11:17:44Z","abstract_excerpt":"We analyze the convergence rate of the unregularized natural policy gradient algorithm with log-linear policy parametrizations in infinite-horizon discounted Markov decision processes. In the deterministic case, when the Q-value is known and can be approximated by a linear combination of a known feature function up to a bias error, we show that a geometrically-increasing step size yields a linear convergence rate towards an optimal policy. We then consider the sample-based case, when the best representation of the Q- value function among linear combinations of a known feature function is known"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.15382","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/2209.15382/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":"2209.15382","created_at":"2026-07-05T05:50:50.078075+00:00"},{"alias_kind":"arxiv_version","alias_value":"2209.15382v2","created_at":"2026-07-05T05:50:50.078075+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.15382","created_at":"2026-07-05T05:50:50.078075+00:00"},{"alias_kind":"pith_short_12","alias_value":"O4LLDVR3WZGF","created_at":"2026-07-05T05:50:50.078075+00:00"},{"alias_kind":"pith_short_16","alias_value":"O4LLDVR3WZGF7GVE","created_at":"2026-07-05T05:50:50.078075+00:00"},{"alias_kind":"pith_short_8","alias_value":"O4LLDVR3","created_at":"2026-07-05T05:50:50.078075+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.04406","citing_title":"Convergence and Sample Complexity of First-Order Methods for Agnostic Reinforcement Learning","ref_index":5,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4","json":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4.json","graph_json":"https://pith.science/api/pith-number/O4LLDVR3WZGF7GVEJPAU674BR4/graph.json","events_json":"https://pith.science/api/pith-number/O4LLDVR3WZGF7GVEJPAU674BR4/events.json","paper":"https://pith.science/paper/O4LLDVR3"},"agent_actions":{"view_html":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4","download_json":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4.json","view_paper":"https://pith.science/paper/O4LLDVR3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2209.15382&json=true","fetch_graph":"https://pith.science/api/pith-number/O4LLDVR3WZGF7GVEJPAU674BR4/graph.json","fetch_events":"https://pith.science/api/pith-number/O4LLDVR3WZGF7GVEJPAU674BR4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4/action/storage_attestation","attest_author":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4/action/author_attestation","sign_citation":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4/action/citation_signature","submit_replication":"https://pith.science/pith/O4LLDVR3WZGF7GVEJPAU674BR4/action/replication_record"}},"created_at":"2026-07-05T05:50:50.078075+00:00","updated_at":"2026-07-05T05:50:50.078075+00:00"}