{"id":"W4414969206","doi":"10.48550/arxiv.2510.03494","title":"Trajectory Data Suffices for Statistically Efficient Policy Evaluation in Fixed-Horizon Offline RL with Linear $q^π$-Realizability and Concentrability","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Nuclear reactor physics and engineering","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Reinforcement learning; Trajectory; Function (biology); Sample (material); Value (mathematics); Sample complexity; Work (physics); Q-learning","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0007306362,0.0003430507,0.0004615206,0.0001134392,0.00003955952,0.00004079746,0.0003486295,0.000184213,0.00001232887],"category_scores_gemma":[0.0003392884,0.0003338933,0.00004118499,0.0002050398,0.00009853668,0.00007592621,0.0002797114,0.000474004,0.000001889453],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003604426,"about_ca_system_score_gemma":0.0003031846,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000552324,"about_ca_topic_score_gemma":0.000258195,"domain_scores_codex":[0.9982075,0.00005664391,0.0004412193,0.0006872957,0.0002509157,0.0003564295],"domain_scores_gemma":[0.9984479,0.0002690117,0.00006375987,0.0009649905,0.0001474852,0.0001068706],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001698883,0.000353234,0.04450396,0.004940459,0.0002574214,0.000003556052,0.0006497851,0.9212272,0.001464394,0.001769448,0.0001424275,0.02451825],"study_design_scores_gemma":[0.0006889205,0.00007257437,0.1424695,0.0002604453,0.0001009187,4.320232e-7,0.00003424008,0.8546069,0.0001524099,0.0001846493,0.001081462,0.000347523],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.974304,0.0003187578,0.02164751,0.00007359538,0.0003323804,0.001209803,0.001557067,0.0001927281,0.000364151],"genre_scores_gemma":[0.9965439,0.00008939028,0.001980558,0.00000928918,0.0002228616,0.00008960255,0.001008136,0.00005146614,0.000004810056],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09796557,"threshold_uncertainty_score":0.9999113,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05665222199401565,"score_gpt":0.3138556402433033,"score_spread":0.2572034182492877,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}