{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:MO7TX7BR5Y57SURKFIN7VFYR7P","short_pith_number":"pith:MO7TX7BR","schema_version":"1.0","canonical_sha256":"63bf3bfc31ee3bf9522a2a1bfa9711fbecd2c3934e764b81a7f1917109fc7f88","source":{"kind":"arxiv","id":"2103.12690","version":2},"attestation_state":"computed","paper":{"title":"An Exponential Lower Bound for Linearly-Realizable MDPs with Constant Suboptimality Gap","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ruosong Wang, Sham M. Kakade, Yuanhao Wang","submitted_at":"2021-03-23T17:05:54Z","abstract_excerpt":"A fundamental question in the theory of reinforcement learning is: suppose the optimal $Q$-function lies in the linear span of a given $d$ dimensional feature mapping, is sample-efficient reinforcement learning (RL) possible? The recent and remarkable result of Weisz et al. (2020) resolved this question in the negative, providing an exponential (in $d$) sample size lower bound, which holds even if the agent has access to a generative model of the environment. One may hope that this information theoretic barrier for RL can be circumvented by further supposing an even more favorable assumption: "},"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":"2103.12690","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-03-23T17:05:54Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"6469ea0bb39e56cf4ea8c09eb5a185c22411ebed548d7441e8270e7cf34c6f50","abstract_canon_sha256":"d477d560b448ba844397d783c2ec0b2c8bc3687a3fb0261dfe8da428a9c25c57"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:24:09.408369Z","signature_b64":"CQxgzsx+k8ViBpD3phDK06n+LSisWl+vre29gWxHnuys8VuOG3znitDM1wOrOUhwCCjgkllQWjN9WQ52l89GCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"63bf3bfc31ee3bf9522a2a1bfa9711fbecd2c3934e764b81a7f1917109fc7f88","last_reissued_at":"2026-07-05T03:24:09.407912Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:24:09.407912Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Exponential Lower Bound for Linearly-Realizable MDPs with Constant Suboptimality Gap","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ruosong Wang, Sham M. Kakade, Yuanhao Wang","submitted_at":"2021-03-23T17:05:54Z","abstract_excerpt":"A fundamental question in the theory of reinforcement learning is: suppose the optimal $Q$-function lies in the linear span of a given $d$ dimensional feature mapping, is sample-efficient reinforcement learning (RL) possible? The recent and remarkable result of Weisz et al. (2020) resolved this question in the negative, providing an exponential (in $d$) sample size lower bound, which holds even if the agent has access to a generative model of the environment. One may hope that this information theoretic barrier for RL can be circumvented by further supposing an even more favorable assumption: "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.12690","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/2103.12690/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":"2103.12690","created_at":"2026-07-05T03:24:09.407984+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.12690v2","created_at":"2026-07-05T03:24:09.407984+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.12690","created_at":"2026-07-05T03:24:09.407984+00:00"},{"alias_kind":"pith_short_12","alias_value":"MO7TX7BR5Y57","created_at":"2026-07-05T03:24:09.407984+00:00"},{"alias_kind":"pith_short_16","alias_value":"MO7TX7BR5Y57SURK","created_at":"2026-07-05T03:24:09.407984+00:00"},{"alias_kind":"pith_short_8","alias_value":"MO7TX7BR","created_at":"2026-07-05T03:24:09.407984+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.06545","citing_title":"Robust Average-Reward Markov Decision Processes: Minimax-Optimal Learning via Plug-in Reductions","ref_index":269,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P","json":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P.json","graph_json":"https://pith.science/api/pith-number/MO7TX7BR5Y57SURKFIN7VFYR7P/graph.json","events_json":"https://pith.science/api/pith-number/MO7TX7BR5Y57SURKFIN7VFYR7P/events.json","paper":"https://pith.science/paper/MO7TX7BR"},"agent_actions":{"view_html":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P","download_json":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P.json","view_paper":"https://pith.science/paper/MO7TX7BR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.12690&json=true","fetch_graph":"https://pith.science/api/pith-number/MO7TX7BR5Y57SURKFIN7VFYR7P/graph.json","fetch_events":"https://pith.science/api/pith-number/MO7TX7BR5Y57SURKFIN7VFYR7P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P/action/storage_attestation","attest_author":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P/action/author_attestation","sign_citation":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P/action/citation_signature","submit_replication":"https://pith.science/pith/MO7TX7BR5Y57SURKFIN7VFYR7P/action/replication_record"}},"created_at":"2026-07-05T03:24:09.407984+00:00","updated_at":"2026-07-05T03:24:09.407984+00:00"}