{
  "visual_id": "E78-GNSS-temporal",
  "title": "Training-selected clock filtering and a retained adequacy failure",
  "classification": "scoped_public_input_calculation",
  "run_ids": [
    "20261003T014345831882Z-W23-GNSS-temporal"
  ],
  "claim_ids": [
    "C07",
    "C14"
  ],
  "input_files": [
    {
      "path": "runs/20261003T014345831882Z-W23-GNSS-temporal/training-selection.json",
      "sha256": "12d2717ac1724f96ba5d3bbd70e08f14b6b77baa7843cb90fb678c7bc0ac653f"
    },
    {
      "path": "runs/20261003T014345831882Z-W23-GNSS-temporal/temporal-diagnostics.json",
      "sha256": "2fa2dc07b6f7849d2685fc8d3cd034151d9039cb015657d8378c3e9ccb96cdea"
    },
    {
      "path": "runs/20261003T014345831882Z-W23-GNSS-temporal/results.json",
      "sha256": "81b5dfb04bcd4809f4816ef38384b3a102220411564f174883a982a0e759b3aa"
    }
  ],
  "configurations": {
    "W23-GNSS-temporal": {
      "parent_run": "20261002T062354709266Z-W23-GNSS-clocks",
      "parent_manifest_sha256": "7427c983f57c83bca4c35d1f07b7072f622b9e603d577628008f7b4ca312e2de",
      "orders": [
        1,
        2,
        4,
        8,
        16,
        32
      ],
      "ridge": [
        0.0,
        0.001,
        0.1
      ],
      "common_trim": 32,
      "training_folds": "four contiguous half-days from the original first two days; each held out in turn; never form cross-segment lags",
      "selection": "minimum summed restricted conditional Gaussian score, fitted exclusively with training products; ties within 1e-10 choose smaller p then smaller ridge",
      "covariance": "ML residual second moment; no jitter, no claim of unbiased covariance with fitted/ridge parameters",
      "maximum_covariance_condition": 100000000,
      "bootstrap_block_samples": [
        20,
        40
      ],
      "bootstrap_replicates": 200,
      "parameter_bootstrap_replicates": 100,
      "parameter_block_samples": 120,
      "parameter_bootstrap_scope": "46 moving one-hour blocks from day-detrended training data, no repeated short-block detrending or artificial lag joins; 4048 lag rows versus 5694 original rows; conditional sensitivity, not a confidence interval",
      "prior_parameter_method_run": "20261003T014221634649Z-W23-GNSS-temporal",
      "parameter_template_indices": [
        0,
        38,
        75,
        143
      ],
      "acf_lags": [
        1,
        2,
        3,
        4,
        5,
        10,
        20
      ],
      "descriptive_max_abs_lag_correlation_gate": 0.05,
      "heldout_second_moment_gate": [
        0.8,
        1.2
      ],
      "algebra_relative_gate": 1e-08,
      "injection_relative_gate": 1e-06,
      "seed": 230603,
      "whole_job_timeout_seconds": 600,
      "maximum_RSS_bytes": 3221225472,
      "threads": 1,
      "scope": "Training-only temporal model comparison and conditional processed-product uncertainty; original third day is reused, not a new blind validation. No receiver response, Gaussian significance, event probability or physical exclusion."
    }
  },
  "assumptions": {
    "W23-GNSS-temporal": [
      "Training-only temporal model comparison and conditional processed-product uncertainty; original third day is reused, not a new blind validation. No receiver response, Gaussian significance, event probability or physical exclusion.",
      "Models estimated after segment-wise affine detrending; conditional Gaussian selection is a plug-in score, not an exact noise likelihood.",
      "Restricted score includes the filtered affine-nuisance determinant, common 32-lag trimming and no cross-segment lag pairs.",
      "Parameter resampling conditions on selected order and ridge and does not calibrate model-selection or receiver uncertainty."
    ]
  },
  "plotted_values": {
    "path": "E78-GNSS-temporal-data.json",
    "sha256": "e5d63218858def29b65ae566c489cae641c952a43a8042b1db65fd142668b1a7"
  },
  "caption": "The left score includes filtered affine-nuisance determinants and identical conditioning intervals. Selection uses only the first two days; the third day was already inspected in the earlier analysis. The right curves show maxima across all channel pairs in the fixed contrast basis at seven declared lags, not an all-lag or basis-independent whitening certificate. Selected order 16 still exceeds the 0.05 correlation gate; a centered second moment of 1.20325 separately exceeds 1.2. The improved filter provides no calibrated event significance or physical exclusion.",
  "alt_text": "The training score favors order 16 with weak regularization. Its residual correlations are smaller than the order-1 baseline, but the validation threshold is still exceeded.",
  "generator": "src/metastable/public_report.py",
  "generator_sha256": "9e263878b7237d781a41d94693004412279b5cfa808ccd6e335b7b07269910d4",
  "generator_dependencies_sha256": {
    "src/metastable/render.py": "ff4837be5102588b6614aeb2b7880b10c0a0c20a155e2eb7b409014a47a3de01",
    "src/metastable/spin_report.py": "dd5c64c5c658938948672a7dc767aa02c34dde012c3fc4aedfc3642f88796847",
    "src/metastable/continuation_report.py": "f9bbce6d0e201a9f55c826d6e9c47ab901e51a71b4a1f962b52eb08a63ba5fe9",
    "src/metastable/refinement_report.py": "4539c0ec814603105ab3a2b5cc82ce291c0151113b5bd3916cdc235aadb9bfb0",
    "src/metastable/atomic_mass_report.py": "46dd1cdb56a5a02cafe001a1f6a7fb201e149461850df41e012f77a580c3694b",
    "src/metastable/closure_report.py": "b34a03a9e0058eed65722df542dbb92691dd138787e2078b43ee9f18e221f75c",
    "src/metastable/october_report.py": "f93e7f89e43165d8d25e2184b9b32e6e21c3e1ffe66020d35944f263aaecb726",
    "src/metastable/october_followup_report.py": "925d905ff3bddf97a7ace2f68a4a28e7dfea2238c4f81d891dda2a9e70c6d4f6",
    "src/metastable/october_dynamics_report.py": "6b8d84c0ce0d78f52d064c68697f0d9c885cdc736e215b54eba46821dbe59a33",
    "src/metastable/october_calibration_report.py": "f714ef7ccfaa26e847c6135054f53528b3ff1d7096e7d113cffeea333e77e3d2",
    "src/metastable/october_interface_report.py": "2fc8826a33c952d7898154f84830748c26842c9a165d14aa7b7fb9721fac00bd",
    "src/metastable/october_checkpoint8_report.py": "f3b54f23510b9bfeffc5ff04e6df54e967228f43ba25a15e12e34f898cdfe0af",
    "src/metastable/october_checkpoint9_report.py": "4bd987aac59964ccb4768a95e0cb6f7d1fdccbeffe9483860da7b431f2224ae4",
    "src/metastable/october_checkpoint10_report.py": "9da673d773b63298e041914447ffd97065c1cf2b8966c90fc4ebcbee25a0ed63",
    "src/metastable/october_checkpoint11_report.py": "5f01064f02f4c76d0059d348a763b285c92c3677e6cb0ba0bbf7faa1a4b98248",
    "src/metastable/october_checkpoint12_report.py": "46022030652e3a1e274dfd27b36dfab25d90a97fde439b1da44d6f1fdc5140b5",
    "src/metastable/october_checkpoint13_report.py": "997c59be3af1b6b7be95fd4314e0f3b98c562629901a639572284045a65980c2",
    "src/metastable/october_checkpoint14_report.py": "d84de7d8516d8a4d8e21b5f9f279476f3035f719c1c8ab03b7ac1e44bf822240",
    "src/metastable/october_checkpoint15_report.py": "1f6e17b9549dc31fbb58ba578c4d8ed8e219b4ad30810df0a9d7d2d00e63efdc",
    "src/metastable/october_checkpoint16_report.py": "74cb8aac45d1d5ce0e6eb6678e9b6cdf98cd46fd6e264a513b612ab763fcd034",
    "src/metastable/october_checkpoint17_report.py": "9daceb74e91f6f8db0f7d89929226d46c464c284d06d0694f525fecf0c736bf6",
    "src/metastable/october_checkpoint18_report.py": "f33f88e316cae93af87cef75d8df41ce82970c8f8bc466dd7a094a92c5f00f8b",
    "src/metastable/october_checkpoint19_report.py": "ce2250507145354b6478cafaaff98c2b8ae2795c3c309f0d903a920365a77090",
    "src/metastable/october_checkpoint20_report.py": "e2ea0a8a374036254008cf6de8f10352dfb2585c4ccb18c1e850c017e6128294",
    "src/metastable/october_checkpoint21_report.py": "596b873676167227b2c92d358fcdbc0614908a5a9daad1da6227b81fce203eec",
    "src/metastable/publication_files.py": "e68bd0ee4e849a72fa828f0cfcf34cb6b1b9f2a8bbdbb3d51085054915039607"
  },
  "command": "python run.py full public-report",
  "output_sha256": {
    "svg": "16e8a2a1b57824522681fa2e4d87d289354a87f8ce77a12681dbb6d2390648ad",
    "png": "82f7173280b2b30ed9a9c48547268016649198281b04e54b8d06393b748a34b9"
  }
}