{ "assessment_plan": [ "verified", "verified", "verified", "verified", "verified" ], "claim_1": { "assessment": "verified", "bernoulli_ks_improves_100_to_2000": true, "beta_final_ks_below_0_03": true, "fixed_n_clt": { "bernoulli": [ { "n": 100, "normal_ks": 0.14840592750346715, "paths": 5000, "standardized_mean": 0.24506037701898298, "standardized_variance": 1.1023049851286795 }, { "n": 500, "normal_ks": 0.04173293775063873, "paths": 5000, "standardized_mean": 0.09814720029471675, "standardized_variance": 1.003909259628457 }, { "n": 1000, "normal_ks": 0.027220442095325634, "paths": 5000, "standardized_mean": 0.0625028824255634, "standardized_variance": 1.02130400376166 }, { "n": 2000, "normal_ks": 0.023373629898683607, "paths": 5000, "standardized_mean": 0.052546007163479146, "standardized_variance": 1.0058119460408557 } ], "bernoulli_oracle": { "ell_x0": 0.28768207245178085, "ell_x1": -0.1541506798272583, "kl_inf": 0.022582421084357415, "lambda_star": 0.47619047619047605, "m0": 0.7, "p": 0.6, "sigma_squared": 0.04685188343675298, "stopping_sigma_squared": 4068.324165943995 }, "beta": [ { "n": 1000, "normal_ks": 0.014912901352047148, "paths": 5000, "standardized_mean": 0.03854332376613194, "standardized_variance": 0.9986271749057966 }, { "n": 2000, "normal_ks": 0.011988900795408514, "paths": 5000, "standardized_mean": 0.02253037708653511, "standardized_variance": 0.9687695995559917 } ], "beta_oracle": { "distribution": "Beta(3,2)", "kl_inf": 0.13361405651995045, "lambda_star": 2.72075767579999, "m0": 0.7, "sigma_squared": 0.24039364204799762 }, "dual_decomposition": [ { "leading_term_standardized_ks": 0.0446543126962895, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 100, "optimization_term_rms": 0.0856072151751973 }, { "leading_term_standardized_ks": 0.028032236762613627, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 500, "optimization_term_rms": 0.03837367264291397 }, { "leading_term_standardized_ks": 0.0225337415746808, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 1000, "optimization_term_rms": 0.027957752784735702 }, { "leading_term_standardized_ks": 0.013671866500478247, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 2000, "optimization_term_rms": 0.01936522864930784 } ], "zero_lambda_control": { "destroys_positive_kl_signal": true, "lambda": 0, "objective_for_every_sample": 0 } }, "paper_native_paths_per_cell": 5000 }, "claim_2": { "assessment": "verified", "constant_boundary_ks_improves_1e4_to_1e8": true, "paper_native_paths_per_cell": 5000, "stopping_time_clt": [ { "boundary": "theoretical", "distribution": "Bernoulli(0.6)", "log10_alpha": -4, "m0": 0.2, "maximum_simulated_step": 102, "mean_tau": 37.4672, "normal_ks": 0.4009024336465053, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8286, "standardized_mean": 1.182813050633105, "standardized_variance": 1.827688349949487, "tau_quantiles_025_50_975": [ 12.0, 37.0, 72.0 ] }, { "boundary": "theoretical", "distribution": "Bernoulli(0.6)", "log10_alpha": -8, "m0": 0.2, "maximum_simulated_step": 153, "mean_tau": 63.4144, "normal_ks": 0.33497018553985086, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.885, "standardized_mean": 0.9510559162163594, "standardized_variance": 1.444734601034845, "tau_quantiles_025_50_975": [ 29.0, 62.0, 105.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -4, "m0": 0.2, "maximum_simulated_step": 77, "mean_tau": 24.1716, "normal_ks": 0.08633869419826778, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8044, "standardized_mean": 0.004871449606083055, "standardized_variance": 0.9947917156102605, "tau_quantiles_025_50_975": [ 6.0, 24.0, 51.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -8, "m0": 0.2, "maximum_simulated_step": 133, "mean_tau": 48.608, "normal_ks": 0.05782882090768432, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8758, "standardized_mean": 0.023478216428855638, "standardized_variance": 1.021829572881041, "tau_quantiles_025_50_975": [ 22.0, 47.0, 84.0 ] } ], "theoretical_boundary_is_more_finite_sample_distorted": true }, "claim_3": { "assessment": "verified", "dual_decomposition": [ { "leading_term_standardized_ks": 0.0446543126962895, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 100, "optimization_term_rms": 0.0856072151751973 }, { "leading_term_standardized_ks": 0.028032236762613627, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 500, "optimization_term_rms": 0.03837367264291397 }, { "leading_term_standardized_ks": 0.0225337415746808, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 1000, "optimization_term_rms": 0.027957752784735702 }, { "leading_term_standardized_ks": 0.013671866500478247, "maximum_reconstruction_residual": 1.1102230246251565e-16, "n": 2000, "optimization_term_rms": 0.01936522864930784 } ], "leading_term_final_ks": 0.013671866500478247, "maximum_identity_residual": 1.1102230246251565e-16, "optimization_remainder_rms_shrinks": true, "zero_lambda_destructive_control": { "destroys_positive_kl_signal": true, "lambda": 0, "objective_for_every_sample": 0 } }, "claim_4": { "assessment": "verified", "coverage_improves_toward_0_95": true, "each_interval_uses_only_its_own_stopped_path": true, "single_path_plugin_intervals": [ { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -4, "m0": 0.2, "maximum_simulated_step": 82, "mean_tau": 24.0322, "normal_ks": 0.08680876259880499, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.7904, "standardized_mean": -0.0074788812690310255, "standardized_variance": 1.0072580705470597, "tau_quantiles_025_50_975": [ 6.0, 22.0, 51.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -8, "m0": 0.2, "maximum_simulated_step": 121, "mean_tau": 48.4284, "normal_ks": 0.0620288209076843, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8778, "standardized_mean": 0.012226801170464973, "standardized_variance": 0.987406165925116, "tau_quantiles_025_50_975": [ 22.0, 47.5, 84.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -32, "m0": 0.2, "maximum_simulated_step": 321, "mean_tau": 193.5538, "normal_ks": 0.03101901148617303, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.9336, "standardized_mean": 0.01944810078666818, "standardized_variance": 0.9842986345522408, "tau_quantiles_025_50_975": [ 138.0, 193.0, 258.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -128, "m0": 0.2, "maximum_simulated_step": 1021, "mean_tau": 771.2616, "normal_ks": 0.022753710985445874, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.9458, "standardized_mean": -0.00736240102574867, "standardized_variance": 1.001023740171591, "tau_quantiles_025_50_975": [ 650.0, 769.0, 901.0 ] } ], "smallest_alpha_coverage": 0.9458 }, "claim_5": { "all_three_distribution_families_executed": true, "assessment": "verified", "primary_dssat_bootstrap": { "bootstrap_paths": 3000, "boundary": "constant log(1/alpha), the paper's practical boundary", "log10_alpha": -4, "m0": 0.5, "maximum_simulated_step": 1208, "mean_tau": 336.972, "normal_ks": 0.09023880940780266, "normalization": "HWAM divided by the maximum of the 44 pinned positive observations", "normalized_mean": 0.5517146751141889, "observations": 44, "oracle": { "direction": "upper", "kl_inf": 0.026160649682450424, "lambda_star_transformed": 1.0162517771111834, "m0": 0.5, "mean": 0.5517146751141889, "sigma_squared": 0.051324123507746816, "stopping_sigma_squared": 2866.6585429554875 }, "raw_maximum": 12340.0, "raw_mean": 6808.159090909091, "raw_minimum": 1030.0, "scope": "direct clean-room execution on pinned primary DSSAT crop-yield observations; the authors do not release their exact normalized vector", "source": "official DSSAT dssat-csm-data repository, Maize/*.MZA HWAM fields", "source_files": [ "BRPI0202.MZA", "EBPL8501.MZA", "FLSC8101.MZA", "IBWA8301.MZA", "IUAF9901.MZA", "SIAZ9501.MZA", "SIAZ9601.MZA", "UFGA8201.MZA" ], "standardized_mean": -0.09290735489006162, "standardized_variance": 0.9746966051025078, "tau_quantiles_025_50_975": [ 85.0, 318.0, 695.0250000000001 ] }, "synthetic_beta_and_bernoulli": { "bernoulli": [ { "n": 100, "normal_ks": 0.14840592750346715, "paths": 5000, "standardized_mean": 0.24506037701898298, "standardized_variance": 1.1023049851286795 }, { "n": 500, "normal_ks": 0.04173293775063873, "paths": 5000, "standardized_mean": 0.09814720029471675, "standardized_variance": 1.003909259628457 }, { "n": 1000, "normal_ks": 0.027220442095325634, "paths": 5000, "standardized_mean": 0.0625028824255634, "standardized_variance": 1.02130400376166 }, { "n": 2000, "normal_ks": 0.023373629898683607, "paths": 5000, "standardized_mean": 0.052546007163479146, "standardized_variance": 1.0058119460408557 } ], "beta": [ { "n": 1000, "normal_ks": 0.014912901352047148, "paths": 5000, "standardized_mean": 0.03854332376613194, "standardized_variance": 0.9986271749057966 }, { "n": 2000, "normal_ks": 0.011988900795408514, "paths": 5000, "standardized_mean": 0.02253037708653511, "standardized_variance": 0.9687695995559917 } ], "stopping": [ { "boundary": "theoretical", "distribution": "Bernoulli(0.6)", "log10_alpha": -4, "m0": 0.2, "maximum_simulated_step": 102, "mean_tau": 37.4672, "normal_ks": 0.4009024336465053, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8286, "standardized_mean": 1.182813050633105, "standardized_variance": 1.827688349949487, "tau_quantiles_025_50_975": [ 12.0, 37.0, 72.0 ] }, { "boundary": "theoretical", "distribution": "Bernoulli(0.6)", "log10_alpha": -8, "m0": 0.2, "maximum_simulated_step": 153, "mean_tau": 63.4144, "normal_ks": 0.33497018553985086, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.885, "standardized_mean": 0.9510559162163594, "standardized_variance": 1.444734601034845, "tau_quantiles_025_50_975": [ 29.0, 62.0, 105.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -4, "m0": 0.2, "maximum_simulated_step": 77, "mean_tau": 24.1716, "normal_ks": 0.08633869419826778, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8044, "standardized_mean": 0.004871449606083055, "standardized_variance": 0.9947917156102605, "tau_quantiles_025_50_975": [ 6.0, 24.0, 51.0 ] }, { "boundary": "constant", "distribution": "Bernoulli(0.6)", "log10_alpha": -8, "m0": 0.2, "maximum_simulated_step": 133, "mean_tau": 48.608, "normal_ks": 0.05782882090768432, "oracle": { "ell_x0": -0.693147180559945, "ell_x1": 1.0986122886681096, "kl_inf": 0.3819085009768876, "lambda_star": -2.499999999999999, "m0": 0.2, "p": 0.6, "sigma_squared": 0.7704964789364159, "stopping_sigma_squared": 13.832245291385126 }, "paths": 5000, "single_path_plugin_ci_empirical_coverage": 0.8758, "standardized_mean": 0.023478216428855638, "standardized_variance": 1.021829572881041, "tau_quantiles_025_50_975": [ 22.0, 47.0, 84.0 ] } ] } }, "integrity_gates": { "accepted_source_literals_exact": true, "all_claims_directly_executed": true, "all_source_pins_exact": true, "bernoulli_fixed_n_convergence": true, "beta_fixed_n_gaussian_fit": true, "dssat_gaussian_fit": true, "dssat_primary_scale": true, "dual_identity_machine_precision": true, "optimization_remainder_shrinks": true, "single_path_ci_converges": true, "stopping_time_gaussian_fit_improves": true, "zero_lambda_control_destroys_signal": true }, "paper_id": "HMyCBL2yMV", "paper_title": "Beyond First-order Asymptotics in Sequential Mean Testing", "claims": [ {"claim": 1, "literal_claim": "Theorem 4.2 establishes a central limit theorem for the empirical KL_inf statistic, showing sqrt(n)(KL_inf(q_hat_n, m_o) - KL_inf(q, m_o)) converges in distribution to N(0, sigma^2(q, m_o)) (Theorem 4.2)."}, {"claim": 2, "literal_claim": "Theorem 4.4 extends this result to the stopping time tau_alpha, proving sqrt(log(1/alpha))(tau_alpha/log(1/alpha) - 1/KL_inf(q,m_o)) converges to a Gaussian limit N(0, sigma^2_bd(q,m_o)) as alpha to 0 (Theorem 4.4)."}, {"claim": 3, "literal_claim": "The proof decomposes the normalized KL_inf statistic into a term from the dual optimization (shown to vanish in probability) and a standard empirical-mean term that converges to Gaussian, combined with verification of Anscombe's condition to transfer the CLT to the stopping time (Section 4)."}, {"claim": 4, "literal_claim": "Proposition 4.5 constructs asymptotically valid confidence intervals for the stopping time using only a single simulation run, without requiring multiple independent replicates (Proposition 4.5)."}, {"claim": 5, "literal_claim": "Numerical experiments on synthetic Beta and Bernoulli distributions and on real crop-yield data show empirical stopping-time distributions converging to the theoretical Gaussian limit, with stronger agreement at smaller significance levels alpha (Section 5)."} ], "seed": 260604520, "source_checks": { "arxiv-2606.04520.pdf": true, "arxiv-2606.04520.tar": true, "dssat-maize-a4f95d3.tar.gz": true, "icml_final_submission.tex": true }, "source_literals": { "beta_distribution": true, "dssat": true, "single_path": true, "stopping_rule": true, "theory_boundary": true } }