{"data_kind":"retrospective_model_comparison","generated":1793841009,"input_mode":"verified_backup","input_snapshot":{"age_seconds":309,"captured_at":1793840700,"journal_mode":"delete","manifest_sha256":"6c192cb67edd7b72a06b1ea01dc0be0931942defa0b698dcc3b741072654adf3","maximum_age_seconds":5400,"sha256":"6b36d5b8f1214851905934f1552e61ee5793c54cf59cbb85e033ee4b909a9f67","state":"verified"},"protocol":{"code_sha256":{"core.py":"dc238005c8ecffa3c63622df686cbb97c3feb42fa2dd0ae72dec0b36beb9df5c","maintenance.py":"1cb7b491342acc6d2aecc4d078ec52180c8de6c6c609eb623107d2b46b324bc5","model_report.py":"1382c05d136251641015bcfe2bcde09ca33a1d8a4fde2fa5d2748dfbbb08e81b","models.py":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","sources.py":"3d87457a761102b2ae471ed2b85a0006e0593aefe45ed552e4eb2c5aca7dd096"},"comparison":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"data_kind":"retrospective_model_comparison","freshness":{"collection_interval_seconds":900,"other_rule":"max(2 * source_cadence, 2 * core.COLLECTION_INTERVAL)","seismic_seconds":345600},"input":{"manual":"Explicit --db keeps direct read-only access for tests and manual use.","production":"Latest verified maintenance backup only, in DELETE journal mode.","selection":"Latest conforming snapshot directory name first; invalid latest never falls back to an older backup.","snapshot_maximum_age_seconds":5400,"verification":"maintenance.verify_backup; complete manifest hashes and SQLite quick_check."},"runtime":{"numpy":"2.5.4","python":"3.12.3 (main, Aug 31 2026, 10:18:26) [GCC 13.3.0]","python_implementation":"CPython","scipy":"1.18.1","sqlite":"3.45.1"},"schema":"trinity.models/1","scope":{"alerts":false,"confidence_intervals":false,"cross_source_ranking":false,"model_probabilities":false,"production_detection_changed":false},"selection":{"minimum_coverage":0.9,"nonfinite_or_null":"Missing observations, explicitly counted in the common mask.","observations":"Finite observations at their exact original grid indices; no interpolation or gap compression.","off_grid_or_duplicate":"Refuse the window; no fallback to an older window.","step_seconds":3600,"window_end":"floor(latest_sample_at_or_before_report_time / core.STEP) * core.STEP","window_samples":{"3600":168,"60":360,"900":96}},"source_inventory":{"finance.btc_usdt":{"cadence_seconds":60,"domain":"finance"},"sismique.IU_ANMO":{"cadence_seconds":60,"domain":"sismique"},"sismique.IU_PAB":{"cadence_seconds":60,"domain":"sismique"},"solaire.champ_bt":{"cadence_seconds":60,"domain":"solaire"},"solaire.rayons_x":{"cadence_seconds":60,"domain":"solaire"},"solaire.vent_densite":{"cadence_seconds":60,"domain":"solaire"},"solaire.vent_vitesse":{"cadence_seconds":60,"domain":"solaire"},"web.wikipedia_en":{"cadence_seconds":3600,"domain":"web"}}},"protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","schema":"trinity.models/1","scope":{"alerts":false,"confidence_intervals":false,"cross_source_ranking":false,"model_probabilities":false,"production_detection_changed":false},"sha256":"73f678dff56a6c97e917273b023c70658f5b49dd3bbd81056a6a0e448c3c3bc6","sources":{"finance.btc_usdt":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["ar1"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":0.0,"powerlaw":-0.006913282001804966,"white":-1.0751956607342272},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"91f9218d574d8f407ebf7a35477830791ef071f217d9d01634bbdd28f64fe274","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-12.391000155916082,"mean_coefficient":11.294724089865772,"model":"ar1","observations":270,"optimization":{"evaluations":42,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.9765568053965988},"state":"ready","training_log_density":1702.8312310604588,"training_sha256":"500683c1aad02e0207a751044c67735fadb95e49364c410d895620e766a441da","trend_coefficient":0.0064697388456359224,"variance":4.155822662323026e-06},"prediction":{"log_density":557.1024644234132,"log_density_per_observation":6.190027382482368,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":-10.20058988038158,"mean_coefficient":11.294934157532984,"model":"powerlaw","observations":270,"optimization":{"evaluations":44,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":1.8645139169333216},"state":"ready","training_log_density":1700.4329238707342,"training_sha256":"500683c1aad02e0207a751044c67735fadb95e49364c410d895620e766a441da","trend_coefficient":0.0047855832631210505,"variance":3.7148399107957435e-05},"prediction":{"log_density":556.4802690432507,"log_density_per_observation":6.183114100480563,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-12.339410423619345,"mean_coefficient":11.294887448594215,"model":"white","observations":270,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":1282.7070032233498,"training_sha256":"500683c1aad02e0207a751044c67735fadb95e49364c410d895620e766a441da","trend_coefficient":0.0088989610350976,"variance":4.375847145599367e-06},"prediction":{"log_density":460.33485495733265,"log_density_per_observation":5.114831721748141,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":0,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":270},"state":"ready"},"current_window_state":"ready","domain":"finance","latest_sample_at":1793840580,"provenance":{"data_sha256":"c434add60f816193e93f00e7638707c9a6d373a7342e1c31597956a1903fa379","indices_sha256":"1c006dbc9c4b5bf97220a35786f55b1a46a5128c0403dce779d9cfad6b11de37","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"82e9452d238154b35e59301018846f7d67f03cefbc7aea86d12b9351b7ab5285","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"finance.btc_usdt","values_sha256":"e3c01259192ad62775991e1f5b662ded27aecab9eb42ac2553688cedd02903a2"},"sample_age_seconds":429,"stale_after_seconds":1800,"state":"ready","window":{"coverage":1.0,"duplicate_rows":0,"end":1793840400,"interpolated_points":0,"longest_gap_samples":0,"missing_indices":[],"missing_points":0,"nonfinite_or_null_points":0,"observed_points":360,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793818860}},"sismique.IU_ANMO":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["powerlaw"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":-0.033411512553284894,"powerlaw":0.0,"white":-0.05092853224252569},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"f6abf45138644d72fc54449510bfb72ead3b93e62d6662ea782658aed4369b00","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-4.064570181526545,"mean_coefficient":2.6070116794690987,"model":"ar1","observations":270,"optimization":{"evaluations":41,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.5498192550807431},"state":"ready","training_log_density":214.019240579252,"training_sha256":"134a77ac3135e642d717ea775c9abd6a5fa6fe735e2e5897bd1d022e6edf88ae","trend_coefficient":-0.040386735545257686,"variance":0.017170367829058297},"prediction":{"log_density":84.06777643229795,"log_density_per_observation":0.9340864048033105,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":-3.947278992189103,"mean_coefficient":2.606869121710729,"model":"powerlaw","observations":270,"optimization":{"evaluations":43,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":0.7566958863899348},"state":"ready","training_log_density":219.79939380610858,"training_sha256":"134a77ac3135e642d717ea775c9abd6a5fa6fe735e2e5897bd1d022e6edf88ae","trend_coefficient":-0.00788878500258392,"variance":0.01930716531379365},"prediction":{"log_density":87.07481256209358,"log_density_per_observation":0.9674979173565954,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-4.062363618336496,"mean_coefficient":2.6063165041572236,"model":"white","observations":270,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":165.3056845101653,"training_sha256":"134a77ac3135e642d717ea775c9abd6a5fa6fe735e2e5897bd1d022e6edf88ae","trend_coefficient":-0.04553440659438686,"variance":0.017208297162014878},"prediction":{"log_density":82.49124466026628,"log_density_per_observation":0.9165693851140697,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":0,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":270},"state":"ready"},"current_window_state":"ready","domain":"sismique","latest_sample_at":1793750340,"provenance":{"data_sha256":"fc178ab44bd5cd45df7123a023f3426e4d3cb3667e38c2ba89315456e2b7e7c6","indices_sha256":"1c006dbc9c4b5bf97220a35786f55b1a46a5128c0403dce779d9cfad6b11de37","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"82e9452d238154b35e59301018846f7d67f03cefbc7aea86d12b9351b7ab5285","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"sismique.IU_ANMO","values_sha256":"4151ed2c7c6b94140178d1f65971991c93df8d1176bbb4cd8b823622eb846f2f"},"sample_age_seconds":90669,"stale_after_seconds":345600,"state":"ready","window":{"coverage":1.0,"duplicate_rows":0,"end":1793746800,"interpolated_points":0,"longest_gap_samples":0,"missing_indices":[],"missing_points":0,"nonfinite_or_null_points":0,"observed_points":360,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793725260}},"sismique.IU_PAB":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["ar1"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":0.0,"powerlaw":-0.0037077103565790193,"white":-0.0004988715189417903},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"73caf8860a4e43b6287a6855d2a46e63803d6696a9fab39c9b6e7e5d708125e6","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-5.642121357705469,"mean_coefficient":2.415557516844081,"model":"ar1","observations":270,"optimization":{"evaluations":41,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.011163928191878504},"state":"ready","training_log_density":378.5897435475531,"training_sha256":"cec6cf3a5466cc903481a6f8263867ffe94a5adcc12992d5e547906c570be496","trend_coefficient":-0.041443678854423245,"variance":0.0035453394900369284},"prediction":{"log_density":113.36291436145198,"log_density_per_observation":1.2595879373494665,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":-5.641889075482155,"mean_coefficient":2.415543853218342,"model":"powerlaw","observations":270,"optimization":{"evaluations":43,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":-0.05532481600386169},"state":"ready","training_log_density":378.73700775667345,"training_sha256":"cec6cf3a5466cc903481a6f8263867ffe94a5adcc12992d5e547906c570be496","trend_coefficient":-0.04139267569326533,"variance":0.0035461631050279366},"prediction":{"log_density":113.02922042935987,"log_density_per_observation":1.2558802269928875,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-5.642121302384695,"mean_coefficient":2.4155623465496383,"model":"white","observations":270,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":378.5729718566721,"training_sha256":"cec6cf3a5466cc903481a6f8263867ffe94a5adcc12992d5e547906c570be496","trend_coefficient":-0.041405344880260576,"variance":0.003545339686167861},"prediction":{"log_density":113.31801592474723,"log_density_per_observation":1.2590890658305247,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":0,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":270},"state":"ready"},"current_window_state":"ready","domain":"sismique","latest_sample_at":1793750340,"provenance":{"data_sha256":"80acf53de9055d1463ad6a650a3a9d1b1a5e048756a9e718960230b2dd6857fc","indices_sha256":"1c006dbc9c4b5bf97220a35786f55b1a46a5128c0403dce779d9cfad6b11de37","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"82e9452d238154b35e59301018846f7d67f03cefbc7aea86d12b9351b7ab5285","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"sismique.IU_PAB","values_sha256":"c56ca369f749c11ec6c7a7a3045f1614772bcf15a15ded09c07e848b6ddd0080"},"sample_age_seconds":90669,"stale_after_seconds":345600,"state":"ready","window":{"coverage":1.0,"duplicate_rows":0,"end":1793746800,"interpolated_points":0,"longest_gap_samples":0,"missing_indices":[],"missing_points":0,"nonfinite_or_null_points":0,"observed_points":360,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793725260}},"solaire.champ_bt":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["powerlaw"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":-0.2183406284586187,"powerlaw":0.0,"white":-4.619370441613336},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"46287fec653022c2e35267ecdcea42ad91d4ec5279b7576f961ceda7b6390506","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-2.38833557344889,"mean_coefficient":4.09388852944691,"model":"ar1","observations":269,"optimization":{"evaluations":41,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.8732923539942415},"state":"ready","training_log_density":131.96875938310694,"training_sha256":"79856cc20c2124e0d43e6606b0a474ec757c536e4c0c4474129d2aaae5f27802","trend_coefficient":-1.6798872510758531,"variance":0.09178232174824487},"prediction":{"log_density":9.305935094355632,"log_density_per_observation":0.1033992788261737,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":0.53826553799721,"mean_coefficient":4.303180188973314,"model":"powerlaw","observations":269,"optimization":{"evaluations":44,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":1.7061445187966544},"state":"ready","training_log_density":129.76505091803233,"training_sha256":"79856cc20c2124e0d43e6606b0a474ec757c536e4c0c4474129d2aaae5f27802","trend_coefficient":-1.4056638242562158,"variance":1.713033093176061},"prediction":{"log_density":28.956591655631314,"log_density_per_observation":0.3217399072847924,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-2.3993299942444093,"mean_coefficient":4.068653851110751,"model":"white","observations":269,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":-58.984581206183904,"training_sha256":"79856cc20c2124e0d43e6606b0a474ec757c536e4c0c4474129d2aaae5f27802","trend_coefficient":-1.758644933221351,"variance":0.09077875520679458},"prediction":{"log_density":-386.78674808956896,"log_density_per_observation":-4.297630534328544,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":1,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":269},"state":"ready"},"current_window_state":"ready","domain":"solaire","latest_sample_at":1793840280,"provenance":{"data_sha256":"78a8b125ea094e9069189270f9f25cbdcc592fbf82f78cbffae1c20955d9c746","indices_sha256":"8570766c3b7de2dd2a9ff570cba35a7a8f103c0b1b48ba57f0ac2654ddb23abd","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"30fa894d8a6e121380cfd2405523ab5a7dbc617d6c636eae448918b62e1bdf40","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"solaire.champ_bt","values_sha256":"599056fac4dbca361a6f87046b45241cd650c6bbe64283e39fa841f4d57636ea"},"sample_age_seconds":729,"stale_after_seconds":1800,"state":"ready","window":{"coverage":0.9972222222222222,"duplicate_rows":0,"end":1793836800,"interpolated_points":0,"longest_gap_samples":1,"missing_indices":[220],"missing_points":1,"nonfinite_or_null_points":0,"observed_points":359,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793815260}},"solaire.rayons_x":{"cadence_seconds":60,"comparison":null,"current_window_state":"insufficient_data","domain":"solaire","latest_sample_at":1793840520,"provenance":{"data_sha256":"9c77b840c59286e1e732a77bc05c507ba52f0a38f5dce8d8675040dce567bc46","indices_sha256":"fefd29c5b1c70848a450c00beec2f747efef735cf11abe76e4abdcd546db327a","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"dcd4b470fd4f05e7499087ebec016691b48d368f9cd737f67d9a047f9835a3b1","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"solaire.rayons_x","values_sha256":"2076f98aa890a16d88eba97d1406923368d0d7b2bc76ca17e92dd1609546acb2"},"sample_age_seconds":489,"stale_after_seconds":1800,"state":"insufficient_data","window":{"coverage":0.7222222222222222,"duplicate_rows":0,"end":1793840400,"interpolated_points":0,"longest_gap_samples":100,"missing_indices":[146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245],"missing_points":100,"nonfinite_or_null_points":0,"observed_points":260,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793818860}},"solaire.vent_densite":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["powerlaw"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":-0.03838731858593203,"powerlaw":0.0,"white":-0.6952659098171882},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"145948d462e75f3a0f9cdb572b322677157d1c4887c21e129daf18eb0d41086c","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-4.181677091486328,"mean_coefficient":1.554282698305645,"model":"ar1","observations":266,"optimization":{"evaluations":41,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.6739620010990602},"state":"ready","training_log_density":258.33247687475307,"training_sha256":"03af703f5fd6cd6a2a7b81567a653d47961b84c9345a7d27c90fd989c97fd048","trend_coefficient":0.2992711028262481,"variance":0.015272872072475015},"prediction":{"log_density":55.144748944524615,"log_density_per_observation":0.6127194327169402,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":-3.493897869236532,"mean_coefficient":1.5311067632966937,"model":"powerlaw","observations":266,"optimization":{"evaluations":42,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":1.0444044433970183},"state":"ready","training_log_density":254.29548606522712,"training_sha256":"03af703f5fd6cd6a2a7b81567a653d47961b84c9345a7d27c90fd989c97fd048","trend_coefficient":0.2196669961474422,"variance":0.030382215164874684},"prediction":{"log_density":58.5996076172585,"log_density_per_observation":0.6511067513028722,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-4.221069190775146,"mean_coefficient":1.561474002348694,"model":"white","observations":266,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":183.96455254065145,"training_sha256":"03af703f5fd6cd6a2a7b81567a653d47961b84c9345a7d27c90fd989c97fd048","trend_coefficient":0.32911484916433864,"variance":0.01468293724831325},"prediction":{"log_density":-3.974324266288434,"log_density_per_observation":-0.044159158514315934,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":4,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":266},"state":"ready"},"current_window_state":"ready","domain":"solaire","latest_sample_at":1793840280,"provenance":{"data_sha256":"e4378e9408c655bc2f24bad28b697fc0d2f419316acba6d290916a28b5f98b21","indices_sha256":"9431db350ba26b6c45b732b117a9c4b3353c8faf83556ef8cccfa12354ac64b8","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,false,true,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"ff2e878241114b7972a892b20e4b6b7acc4dc50ebc90cde648c315a5d12ecd01","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"solaire.vent_densite","values_sha256":"8b7544d2cf756f00a4f6c1b7a4767f475d58ba555a990f07c8b1c88742bc8977"},"sample_age_seconds":729,"stale_after_seconds":1800,"state":"ready","window":{"coverage":0.9888888888888889,"duplicate_rows":0,"end":1793836800,"interpolated_points":0,"longest_gap_samples":2,"missing_indices":[37,220,221,223],"missing_points":4,"nonfinite_or_null_points":0,"observed_points":356,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793815260}},"solaire.vent_vitesse":{"cadence_seconds":60,"comparison":{"comparison":{"highest_score_models":["powerlaw"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":-0.3120288687125248,"powerlaw":0.0,"white":-2.540360824892331},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"8978aa6cdba3f894281abd7897222af05266b819b4cfd5895170aa2cd885f7be","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":2.728944452180425,"mean_coefficient":387.1592659251501,"model":"ar1","observations":266,"optimization":{"evaluations":41,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.7650310672147868},"state":"ready","training_log_density":-624.5584993879254,"training_sha256":"c392d03782a340c550489c44c8d166563996cc30653b7a8b27c0bb4dcd806a9c","trend_coefficient":-31.14624481430798,"variance":15.31671096324813},"prediction":{"log_density":-259.15714203133103,"log_density_per_observation":-2.8795238003481227,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":3.8793605499323327,"mean_coefficient":385.9170227640958,"model":"powerlaw","observations":266,"optimization":{"evaluations":43,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":1.239015120673437},"state":"ready","training_log_density":-627.5119292911409,"training_sha256":"c392d03782a340c550489c44c8d166563996cc30653b7a8b27c0bb4dcd806a9c","trend_coefficient":-31.950671601629807,"variance":48.39326010187451},"prediction":{"log_density":-231.0745438472038,"log_density_per_observation":-2.567494931635598,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":2.7301439217540833,"mean_coefficient":387.2243879782226,"model":"white","observations":266,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":-740.5467914257359,"training_sha256":"c392d03782a340c550489c44c8d166563996cc30653b7a8b27c0bb4dcd806a9c","trend_coefficient":-31.23222675392792,"variance":15.335093914708537},"prediction":{"log_density":-459.7070180875136,"log_density_per_observation":-5.107855756527929,"observations":90,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":270,"interpolated_observations":0,"missing_observations":4,"nominal_length":360,"test_nominal_points":90,"test_observations":90,"train_nominal_points":270,"train_observations":266},"state":"ready"},"current_window_state":"ready","domain":"solaire","latest_sample_at":1793840280,"provenance":{"data_sha256":"df371ffd283378d6beb5ce90b627eae78b9b4eb0f4fa5384fa47b5d8647d27fe","indices_sha256":"9431db350ba26b6c45b732b117a9c4b3353c8faf83556ef8cccfa12354ac64b8","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,false,false,true,false,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"ff2e878241114b7972a892b20e4b6b7acc4dc50ebc90cde648c315a5d12ecd01","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"solaire.vent_vitesse","values_sha256":"05f654e379a1cdc78f41a8eae3982f49f9dd18138f304587f47939a403460ba0"},"sample_age_seconds":729,"stale_after_seconds":1800,"state":"ready","window":{"coverage":0.9888888888888889,"duplicate_rows":0,"end":1793836800,"interpolated_points":0,"longest_gap_samples":2,"missing_indices":[37,220,221,223],"missing_points":4,"nonfinite_or_null_points":0,"observed_points":356,"off_grid_rows":0,"row_limit_exceeded":false,"samples":360,"start":1793815260}},"web.wikipedia_en":{"cadence_seconds":3600,"comparison":{"comparison":{"highest_score_models":["powerlaw"],"interpretation":"Difference from the largest conditional log density per observed suffix point; relative prediction only.","relative_log_score_per_observation":{"ar1":-0.4674303422106396,"powerlaw":0.0,"white":-1.5188678850423794},"scored_models":["white","ar1","powerlaw"],"state":"ready"},"input_sha256":"6e17d2922a58c39e6705858b352f250773ca663035fac279db1598ba52262ed3","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"fit":{"boundary":false,"log_variance":-3.2198930145239952,"mean_coefficient":15.887962078012023,"model":"ar1","observations":126,"optimization":{"evaluations":42,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-0.99,0.99],"name":"rho","value":0.9446440716127695},"state":"ready","training_log_density":163.37278939195153,"training_sha256":"a2c1c7cfbfdd3eb5101034dae9fe98ee966499ed231ac8b05f9ab3121a6afca4","trend_coefficient":-0.025953834041351084,"variance":0.03995933310024952},"prediction":{"log_density":53.63351416864481,"log_density_per_observation":1.276988432586781,"observations":42,"parameter_uncertainty":"not_integrated","state":"ready"}},"powerlaw":{"fit":{"boundary":false,"log_variance":5.704612126490583,"mean_coefficient":17.267450710179535,"model":"powerlaw","observations":126,"optimization":{"evaluations":58,"failed_refinements":0,"numerical_failures":0},"parameter":{"bounds":[-2.0,4.0],"name":"beta","value":3.267937411888269},"state":"ready","training_log_density":211.05150668621104,"training_sha256":"a2c1c7cfbfdd3eb5101034dae9fe98ee966499ed231ac8b05f9ab3121a6afca4","trend_coefficient":-1.1120668715290847,"variance":300.24899882719893},"prediction":{"log_density":73.26558854149167,"log_density_per_observation":1.7444187747974207,"observations":42,"parameter_uncertainty":"not_integrated","state":"ready"}},"white":{"fit":{"boundary":false,"log_variance":-3.1809742905732117,"mean_coefficient":15.869035274982597,"model":"white","observations":126,"optimization":{"evaluations":1,"failed_refinements":0,"numerical_failures":0},"parameter":null,"state":"ready","training_log_density":21.61512512232359,"training_sha256":"a2c1c7cfbfdd3eb5101034dae9fe98ee966499ed231ac8b05f9ab3121a6afca4","trend_coefficient":0.08172473218533408,"variance":0.04154515834042708},"prediction":{"log_density":9.47313736971174,"log_density_per_observation":0.22555088975504142,"observations":42,"parameter_uncertainty":"not_integrated","state":"ready"}}},"protocol":{"code_sha256":"0150312907715b04a5b36d236487969e440dd4835e056691e260faa9fe835f5b","limitations":["Conditional Gaussian prediction with parameters fitted to the prefix only; parameter uncertainty is not integrated.","Relative predictive scores are neither goodness-of-fit tests nor model probabilities; all three models may be inadequate.","No confidence interval, p-value or claim of a physical 1/f regime is produced.","Stationary covariance around a common fitted mean and linear trend is an assumption, not an observed fact.","Missing observations are marginalized at their real grid positions; missingness is treated as fixed and ignorable.","The power-law model has a fixed finite extension and positive low-frequency plateau; its cutoff is not estimated.","Overlapping windows and correlated sources are not independent replications."],"models":{"ar1":{"covariance":"rho ** abs(real_grid_lag)","rho_bounds":[-0.99,0.99]},"powerlaw":{"beta_bounds":[-2.0,4.0],"covariance":"irfft(weights)[abs(real_grid_lag)] / irfft(weights)[0]","dc":"positive, equal to weight at f_low","extension_factor":8,"frequency_unit":"cycles_per_grid_sample","low_frequency":"1 / (8 * nominal_length)","weights":"(max(abs(f), f_low) / f_low) ** (-beta)"},"white":{"covariance":"identity"}},"nuisance":{"design":"[1, (grid_index - (nominal_length-1)/2) / nominal_length]","fit":"Gaussian maximum likelihood, GLS mean and linear trend; variance=Q/n_train","same_design_for_all_models":true},"optimization":{"boundary_tolerance":1e-07,"bounds_included":true,"grid_points":33,"jitter":false,"maxiter_per_refinement":128,"method":"evaluate fixed global grid, refine every grid-local minimum using bounded minimization, retain best","xatol":1e-07},"prediction":"exact conditional joint Gaussian log density on the observed suffix, per observed point; plug-in parameters","references":[{"title":"Sykulski et al. (2019), The debiased Whittle likelihood","url":"https://doi.org/10.1093/biomet/asy071"},{"title":"SciPy: Cholesky factorization","url":"https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.cho_factor.html"}],"runtime":{"implementation":"CPython","machine":"aarch64","numpy":"2.5.4","python":"3.12.3","scipy":"1.18.1","system":"Linux"},"schema":"trinity.models/1","split":{"first_test_index":"(3 * nominal_length) // 4","minimum_test_observations":16,"minimum_train_observations":64,"nominal_length_maximum":512,"test":"index >= first_test_index","train":"index < first_test_index"}},"schema":"trinity.models/1","split":{"first_test_index":126,"interpolated_observations":0,"missing_observations":0,"nominal_length":168,"test_nominal_points":42,"test_observations":42,"train_nominal_points":126,"train_observations":126},"state":"ready"},"current_window_state":"ready","domain":"web","latest_sample_at":1793822400,"provenance":{"data_sha256":"fd99639971dc598098037f2632d311f7b2e06a3b3e7a74ded7e4da2878baf175","indices_sha256":"b0e0da1d0bfd54a15aafcce9e48a97847a756056fe4969208fcd2eed17e2db58","interpolated_points":0,"mask":[true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true,true],"mask_sha256":"6444f941a030015f38ef4c2f96969cecda887fd86c40cff641de35e56f9fd73e","protocol_sha256":"6e511c8e0e662c36d2abe3bc8a0dae0cdefe1a09040218170c96e1281e674dfc","source":"web.wikipedia_en","values_sha256":"2e7ba0eb276c89cee98ce98290fb1e8ff195028982be8c934aa6ce20670fbeee"},"sample_age_seconds":18609,"stale_after_seconds":7200,"state":"stale","window":{"coverage":1.0,"duplicate_rows":0,"end":1793822400,"interpolated_points":0,"longest_gap_samples":0,"missing_indices":[],"missing_points":0,"nonfinite_or_null_points":0,"observed_points":168,"off_grid_rows":0,"row_limit_exceeded":false,"samples":168,"start":1793221200}}},"state":"partial","work":{"cache_hits":2,"cache_misses":5,"elapsed_seconds":0.792}}