Add manifest for scalar_sc1_256_fp64

manifests/scalar_sc1_256_fp64.json

@@ -0,0 +1,147 @@
+{
+  "case": "scalar_sc1_256_fp64",
+  "compute_precision": "fp64",
+  "storage_precision": "fp32",
+  "channels": [
+    "u",
+    "v",
+    "w",
+    "p",
+    "theta"
+  ],
+  "pressure_source": "operators.pressure_hat (D4-certified)",
+  "regime": {
+    "note": "k_f/tau are confounded with Re; only pure-nu Re axis is single-variable",
+    "input": {
+      "error": "config unreadable: [Errno 2] No such file or directory: '/home/ydai17/turbulence/experiments/phase0/configs/scalar_sc1_256_fp64.yaml'"
+    },
+    "measured": {
+      "Re_lambda_mean": null,
+      "eps_mean": null,
+      "k_max_eta_window_avg": null
+    }
+  },
+  "n_seeds_accepted": 8,
+  "n_seeds_rejected": 1,
+  "accepted_seeds": [
+    0,
+    1,
+    2,
+    3,
+    4,
+    5,
+    7,
+    8
+  ],
+  "rejected_seeds": [
+    6
+  ],
+  "total_frames": 1089,
+  "frame_keep_frac_min": null,
+  "frame_keep_frac_max": null,
+  "pooled_A10": null,
+  "per_seed": [
+    {
+      "seed": 0,
+      "run": "scalar_sc1_256_fp64_corpus",
+      "frames_in_corpus": 150,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 1,
+      "run": "scalar_sc1_256_fp64_corpus_pool1",
+      "frames_in_corpus": 150,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 2,
+      "run": "scalar_sc1_256_fp64_corpus_pool2",
+      "frames_in_corpus": 140,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 3,
+      "run": "scalar_sc1_256_fp64_corpus_pool3",
+      "frames_in_corpus": 143,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 4,
+      "run": "scalar_sc1_256_fp64_corpus_pool4",
+      "frames_in_corpus": 150,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 5,
+      "run": "scalar_sc1_256_fp64_corpus_pool5",
+      "frames_in_corpus": 102,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 7,
+      "run": "scalar_sc1_256_fp64_corpus_pool7",
+      "frames_in_corpus": 104,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    },
+    {
+      "seed": 8,
+      "run": "scalar_sc1_256_fp64_corpus_pool8",
+      "frames_in_corpus": 150,
+      "frame_keep_frac": null,
+      "frames_seen": null,
+      "frames_skipped_underresolved": null,
+      "k_max_eta_window_avg": null,
+      "Re_lambda": null,
+      "eps_mean": null,
+      "K_drift_last5TL_pct": null
+    }
+  ],
+  "frame_filter": "instantaneous k_maxη>=1.5 (per-frame Class-I resolution gate)",
+  "CAVEATS": [
+    "Frames are filtered by instantaneous k_maxeta>=1.5, so they are NOT uniformly spaced in time (dropped slots are eps-peak instants). Each frame stores its true 't' AND its 'k_max_eta' -- loaders MUST use the stored 't' and not assume a uniform dt between consecutive frames.",
+    "INTERMITTENCY BIAS (physical consequence, not just timing): the dropped frames are the eps-PEAK instants -- the most intermittent, strongest-small-scale moments. They are dropped because at those instants k_maxeta<1.5, i.e. the GRID cannot faithfully resolve that burst (spectral pileup at the cutoff); keeping them would feed under-resolved numerical artefacts to a model. The consequence is that this corpus SYSTEMATICALLY UNDER-SAMPLES the strong-dissipation tail: the eps distribution is right-truncated and high-order intermittency statistics (derivative flatness/kurtosis B3, high-order structure functions) are a LOWER BOUND, not unbiased. For Re_lambda~86 the truncation is mild (dropped eps ~1.3-1.4x the mean, not extreme tails), but downstream users must NOT treat this as an unbiased sample of real turbulence's extreme events.",
+    "A seed with low frame_keep_frac (<0.80) kept fewer frames because its instantaneous k_maxeta dipped below 1.5 more often; such a trajectory may be intrinsically closer to the resolution edge -- inspect before publishing."
+  ]
+}