Release v1.1.1 metadata compatibility patch

README.md

@@ -32,13 +32,39 @@ configs:
 Scene2Wave is a physics-consistent scene-to-wave data infrastructure for
 wireless world modeling. This repository is its first public reference
 instantiation: a synchronized CARLA traffic release compiled through RF Scene
-Compiler and simulated with Sionna RT. Version 1.1.0 contains
+Compiler and simulated with Sionna RT. Version 1.1.1 contains
 300 formally validated one-second episodes spanning 12
 CARLA Towns, five motion states, four radio profiles, multi-view cameras,
 Birdview, LiDAR, Radar, IMU, GNSS, poses, path-level CIR, and CSI. The public
 release is a qualified dataset artifact; it does not imply that the complete
 generation toolchain or editable simulator assets are included.
 
+## v1.1.1 metadata compatibility patch
+
+Version 1.1.1 preserves all 300 v1.1.0 sample TAR archives byte-for-byte
+and adds an explicit kinematics sidecar at
+`metadata/kinematics/pose_velocity_v1_1_1.npz`. The original velocity is
+exposed as `tx_route_command_velocity_mps` and
+`rx_route_command_velocity_mps`; the corresponding
+`tx_pose_derived_velocity_mps` and `rx_pose_derived_velocity_mps` arrays are
+10-ms finite differences of the released endpoint positions. Per-frame closure
+errors and documented 0.5-m/s diagnostic validity flags are included. Use the
+pose-derived values for geometric range-rate or Doppler work.
+
+The mismatch was caused by the older waypoint-motion writer using a route
+orientation for velocity while position followed the interpolation-segment
+chord on some curves. Current generation code derives both from the same chord
+and gates pose--velocity closure before publication. The fix applies to future
+generated datasets; v1.1.1 does not silently rewrite historical sample data.
+See `metadata/kinematics/README.md` and `patch_manifest.json` for the exact
+algorithm, field meanings, thresholds, and audit statistics.
+
+Two frozen generation-policy files retain legacy `mmw_rf_*` schema IDs. These
+are historical internal machine-readable identifiers, not a claim of
+affiliation with Multimodal-Wireless. Current compiler/material namespaces use
+`scene2wave_` or `s2w_`; future policies will use the new namespace while
+readers retain aliases for these frozen identifiers.
+
 The Scene2Wave Project Content rightsholder is **Pengcheng Laboratory**. Based
 on the current release configuration and the authors' current understanding,
 the dataset uses official CARLA assets and the official Sionna software stack;
@@ -175,6 +201,12 @@ Scene2Wave/
 ├── metadata/
 │   ├── samples.jsonl
 │   ├── source_samples.jsonl
+│   ├── kinematics/
+│   │   ├── pose_velocity_v1_1_1.npz
+│   │   ├── patch_manifest.json
+│   │   ├── schema.json
+│   │   └── README.md
+│   ├── paper_reproducibility/
 │   ├── json_schema.json
 │   ├── generation_config.yaml
 │   ├── generation_batches.json

checksums.sha256

@@ -1,3 +1,10 @@
+dd40946c2c96108d18ff048be3295735dd29ea34e8e1b18301c9cf30ec8f5775  ATTRIBUTION.md
+29025d213aa7c5b081dae26e6b403213eef113aabab793378ea262105187f177  CITATION.cff
+aa345cc056da5ea8bc5d6a856a1329a249395fc8e046a1fa5136a5e08c1b2ebd  DATA_PROVENANCE.md
+f3972567e1b25c4e0c746c5344a1165b7e8fdcb51690142a70a95aa8b60b0ca3  LICENSE
+b327c5c7fa9ed801d4c460ba2e56fd2bd25d2208b96db1657be49a42fe122008  LICENSES/CC-BY-4.0.txt
+873c73c91692bf0a79e46bdbbc44789c1a583688b12c0cb0a7d45f35ebf27c50  README.md
+9ea776ffa9f96d7e21af140a7df1f3fef9e19512cc841b7b4e8e7428de3d571b  THIRD_PARTY_NOTICES.md
 a2c19f34ade8a2dc45f9c15d92c3db282c7636db7a0fb473371407cbf49da4d3  data/main_training/Town01/10kmh/core/Req_Town01_dynamic_10kmh_t001_rfmain.tar
 c73d7994d672928c5e56b7bedf0e9ab28ec4743bdc91c5db99912641a1e52e22  data/main_training/Town01/10kmh/core/Req_Town01_dynamic_10kmh_t002_rfmain.tar
 f6d08adb813cbe575bc9171c8ca31a812d5f3af3e48c8f99b7bed847c903fe0e  data/main_training/Town01/10kmh/multipath/Req_Town01_dynamic_10kmh_t000_rfmain.tar
@@ -298,21 +305,26 @@ a842189a601cddd12e0c402ca86d104b4e014b7eadec6b201d1ee1dfacec29dc  data/main_trai
 0bda3ee88933ff021de914822791fa7f7f86ff8cce127a109f0392bc8e8fa8f3  data/main_training/Town15/static/multipath/Req_Town15_static_p000_rfmain.tar
 efea75de0b6439eb2a7855dd39d8072d32afaf658eadffc3babb04de44396cc2  data/main_training/Town15/static/scattering_medium/Req_Town15_static_p006_rfsupp01.tar
 01bfa06f9420da4356a4e23657cd1ed1a2cba07af7f38818573306174a3a6816  data/main_training/Town15/static/scattering_mild/Req_Town15_static_p011_rfcarlasafe01.tar
-dd40946c2c96108d18ff048be3295735dd29ea34e8e1b18301c9cf30ec8f5775  ATTRIBUTION.md
-29025d213aa7c5b081dae26e6b403213eef113aabab793378ea262105187f177  CITATION.cff
-aa345cc056da5ea8bc5d6a856a1329a249395fc8e046a1fa5136a5e08c1b2ebd  DATA_PROVENANCE.md
-f3972567e1b25c4e0c746c5344a1165b7e8fdcb51690142a70a95aa8b60b0ca3  LICENSE
-b327c5c7fa9ed801d4c460ba2e56fd2bd25d2208b96db1657be49a42fe122008  LICENSES/CC-BY-4.0.txt
-c039f5cec380bb6bf7c31031e4b40b40899222e241ebfc1220d10349eab26203  README.md
-9ea776ffa9f96d7e21af140a7df1f3fef9e19512cc841b7b4e8e7428de3d571b  THIRD_PARTY_NOTICES.md
-d596bf5d95b875ce66b9d5da06ea8ea0adf2a8d9de61234b80f83145d1ee4861  dataset_info.json
+774e8f5c6444fd30b2239d8280e8eb9938e9ce474cba0747c17e4741c75f9ef6  dataset_info.json
 310edd6c6c41f7254de9b2926fb2b4c2dcc1ecc6a42e837eee94fe24dd172257  metadata/generation_batches.json
-d3fc6adc7722fd7dac140dfc3c6082e0f6c1d8ec3c398c43cc94811da1444c3b  metadata/generation_config.yaml
+26d381cb9b9693f46c7e69af974d0c5389210cb54f65502a581be8978c6fd02c  metadata/generation_config.yaml
 094e2d9bf5c7f3ac53d1b7399adf2b0e792034786076cd8d9ef121ca9524d594  metadata/generation_configs/extension_8town.yaml
 e973af8415e2a831c32a6379a64dfb34221bf1b7773754d5196ff822d7d1b0d8  metadata/generation_configs/original_4town.yaml
-55a2c4cd2ac720dd240d700a3c79a636c81225e99447b91e208d17ea3294a36b  metadata/json_schema.json
+40eccc592817b2a222e7dc94864eddaed1a49d3e04eb829240982aa266b4269f  metadata/json_schema.json
+3e1e8510da5029cd78e2574b148f1d35c6d7d491755cccf3498bd3bec923aff8  metadata/kinematics/README.md
+a7fc252bd2367ea13939164af3efea390cb555a3e1dc0bcdcadc82da0856e604  metadata/kinematics/patch_manifest.json
+50b40b3a7e3880f31fdccb573273aa14669bb55baadba04cbfa228226aa7c787  metadata/kinematics/pose_velocity_v1_1_1.npz
+d87c7f6b7dd7e9f0c83b000ce8dfdf55ce37f2f262457cbead6c3062388e75af  metadata/kinematics/schema.json
+581b7e3dd5bcad649ce514ce5e4a694a75edb306e8ef7881d20d657a2fdb66b3  metadata/paper_reproducibility/receipts/array_doppler_audit.json
+a04fca352a363bee54b77ced5ae4b5e9f3e88909a20976e1b99ca52a3bea618b  metadata/paper_reproducibility/receipts/data_contract_audit.json
+c0cc5444a84a7c2d2e6ee381c7e0eae73d990e6240daac9d8b869b98f370dfbd  metadata/paper_reproducibility/receipts/generation_batch_audit.json
+d35f773fbdbc27c8c73888c7b0747396261a860745be5cb17a1fbb9d16f4b2ca  metadata/paper_reproducibility/receipts/material_evidence_audit.json
+3d1983acc1affe4a83681419b873fddd0ce134c008041f843f623d503913179b  metadata/paper_reproducibility/scripts/analyze_scene2wave_array_doppler.py
+ab28529dbe1cc4e33659306f81650c57d4cfc94ff5ab55fd465661bfcced9571  metadata/paper_reproducibility/scripts/analyze_scene2wave_data_contracts.py
+dddea4cdb2265a723b6ed907fbf9a2cd270515ead42c607fa39bca6829bd195a  metadata/paper_reproducibility/scripts/analyze_scene2wave_generation_batches.py
+56b556d8e53e7816ba2f52fb5148156d7c6bcdf207c58d2d59e29b70a127363b  metadata/paper_reproducibility/scripts/analyze_scene2wave_material_evidence.py
 b9c8c95e070a139ac8ed349f87c7570124ffbb4bb9886215c10f9d8886eb4e91  metadata/raw_dataset_schema.md
 14a4acd1f23099e8f1c9039986fcfbc8d5980436f47d5ab30c37999098601132  metadata/release_validation.json
-458e8abb747c2b03e394b56ed085a8df272a8ff0b236871ffa5384afef0cd46d  metadata/samples.jsonl
+3cd1db3fca3b2efbdc8c830e87c793f12bf3ed251917110a7d6abc8fd87b4b01  metadata/samples.jsonl
 3af6058724c9e0a6377fe11180a7e957210990403dc0d50cfb69535780b2e8ab  metadata/source_samples.jsonl
 e64d5b633343da37d5d7ce07206786d643cb8e53529f7e497d824b86ecf726a1  schema.md

dataset_info.json

@@ -1,9 +1,9 @@
 {
-  "created_at_utc": "2026-08-14T06:03:09.556711+00:00",
+  "created_at_utc": "2026-08-24T06:27:31.498482+00:00",
   "data_file_count": 2520600,
   "data_size_bytes": 230903029002,
   "dataset_name": "scene2wave_dataset",
-  "dataset_version": "1.1.0",
+  "dataset_version": "1.1.1",
   "distributions": {
     "profile": {
       "core": 120,
@@ -54,6 +54,12 @@
     "Sionna CIR/CSI NPZ outputs",
     "per-sample and radio metadata"
   ],
+  "metadata_patch": {
+    "archive_payload_changed": false,
+    "manifest": "metadata/kinematics/patch_manifest.json",
+    "sidecar": "metadata/kinematics/pose_velocity_v1_1_1.npz",
+    "source_dataset_version": "1.1.0"
+  },
   "packaging": {
     "archive_count": 300,
     "archive_index": "metadata/samples.jsonl",

metadata/generation_config.yaml

@@ -1,5 +1,5 @@
 schema_version: scene2wave_generation_batch_index_v1
-dataset_version: "1.1.0"
+dataset_version: "1.1.1"
 batches:
   - id: original_4town_v1_0_1
     samples: 100

metadata/json_schema.json

@@ -362,6 +362,16 @@
           "minLength": 1,
           "type": "string"
         },
+        "kinematics_sample_index": {
+          "description": "First-axis index into the release-level kinematics sidecar.",
+          "maximum": 299,
+          "minimum": 0,
+          "type": "integer"
+        },
+        "kinematics_sidecar_path": {
+          "const": "metadata/kinematics/pose_velocity_v1_1_1.npz",
+          "description": "Release-level v1.1.1 pose/velocity sidecar."
+        },
         "num_subcarriers": {
           "minimum": 1,
           "type": "integer"

metadata/kinematics/README.md

@@ -0,0 +1,29 @@
+# Scene2Wave v1.1.1 kinematics compatibility patch
+
+This additive patch preserves all v1.1.0 sample archives byte-for-byte. The
+release-level NPZ is indexed by `kinematics_sample_index` in
+`metadata/samples.jsonl`.
+
+```python
+import json
+from pathlib import Path
+import numpy as np
+
+root = Path("Scene2Wave")
+row = json.loads((root / "metadata/samples.jsonl").read_text().splitlines()[0])
+with np.load(root / row["kinematics_sidecar_path"], allow_pickle=False) as data:
+    i = row["kinematics_sample_index"]
+    assert data["sample_id"][i] == row["sample_id"]
+    velocity = data["rx_pose_derived_velocity_mps"][i]
+    closure_error = data["rx_velocity_closure_error_mps"][i]
+    validity = data["rx_velocity_validity_flag"][i]
+```
+
+`tx` is the RSU and `rx` is the CAV. Interior pose derivatives use a centered
+10-ms span; the first and last ten frames use 10-ms one-sided spans. The
+0.5-m/s validity flag is a diagnostic closure threshold, not a claim of real
+vehicle-dynamics accuracy. For geometric range rate, prefer
+the `tx_pose_derived_velocity_mps` or `rx_pose_derived_velocity_mps` array;
+retain its `tx_route_command_velocity_mps` or
+`rx_route_command_velocity_mps` counterpart only when the historical
+prescribed route-state semantics are intended.

metadata/kinematics/patch_manifest.json

@@ -0,0 +1,160 @@
+{
+  "archive_policy": "All 300 v1.1.0 sample tar archives are preserved byte-for-byte.",
+  "created_at_utc": "2026-08-24T06:27:31.498482+00:00",
+  "dataset_version": "1.1.1",
+  "endpoint_roles": {
+    "rx": "CAV",
+    "tx": "RSU"
+  },
+  "production_fix": {
+    "applies_to_v1_1_0_archives": false,
+    "description": "Future waypoint-motion generation derives the stored velocity from the same route-segment chord used for position interpolation and applies pose-velocity closure tests before publication.",
+    "status": "implemented_in_current_generation_code"
+  },
+  "sample_order": "metadata/samples.jsonl order",
+  "schema_version": "scene2wave_kinematics_patch_v1",
+  "scope": "metadata_only_additive_patch",
+  "sidecar": "metadata/kinematics/pose_velocity_v1_1_1.npz",
+  "sidecar_sha256": "50b40b3a7e3880f31fdccb573273aa14669bb55baadba04cbfa228226aa7c787",
+  "source_dataset_version": "1.1.0",
+  "statistics": {
+    "all_rx_frames": {
+      "count": 600000,
+      "fraction_above_0_5_mps": 0.03797,
+      "fraction_above_1_mps": 0.016925,
+      "fraction_above_2_mps": 0.00631,
+      "max_mps": 7.506193161010742,
+      "mean_mps": 0.10157467040368652,
+      "median_mps": 0.0011018534423783422,
+      "p95_mps": 0.47457778453826904
+    },
+    "by_town": {
+      "Town01": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.1176,
+        "fraction_above_1_mps": 0.07892,
+        "fraction_above_2_mps": 0.0356,
+        "max_mps": 6.517507076263428,
+        "mean_mps": 0.23723525975819634,
+        "median_mps": 0.0009128862875513732,
+        "p95_mps": 1.919731503725052
+      },
+      "Town02": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.12344,
+        "fraction_above_1_mps": 0.08216,
+        "fraction_above_2_mps": 0.0287,
+        "max_mps": 7.506193161010742,
+        "mean_mps": 0.2652650527481327,
+        "median_mps": 0.001095717423595488,
+        "p95_mps": 1.519473123550415
+      },
+      "Town03": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.04332,
+        "fraction_above_1_mps": 0.01544,
+        "fraction_above_2_mps": 0.0077,
+        "max_mps": 6.006340980529785,
+        "mean_mps": 0.10705627917633218,
+        "median_mps": 0.0008429125300608575,
+        "p95_mps": 0.35921976417303086
+      },
+      "Town04": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.0,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.18041802942752838,
+        "mean_mps": 0.00794572815622858,
+        "median_mps": 0.0006574972649104893,
+        "p95_mps": 0.06203852593898773
+      },
+      "Town05": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.004,
+        "fraction_above_1_mps": 0.00376,
+        "fraction_above_2_mps": 0.00022,
+        "max_mps": 4.2784037590026855,
+        "mean_mps": 0.00682611597576899,
+        "median_mps": 0.0005589168285951018,
+        "p95_mps": 0.0026798402541317046
+      },
+      "Town06": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.0185,
+        "fraction_above_1_mps": 0.018,
+        "fraction_above_2_mps": 0.0035,
+        "max_mps": 2.1250975131988525,
+        "mean_mps": 0.029936448451366886,
+        "median_mps": 0.0011174707906320691,
+        "p95_mps": 0.03416551668196917
+      },
+      "Town07": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.0024,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.9088031053543091,
+        "mean_mps": 0.01489095771263801,
+        "median_mps": 0.00046924129128456116,
+        "p95_mps": 0.0887461721897111
+      },
+      "Town10HD": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.02634,
+        "fraction_above_1_mps": 0.00482,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 1.855961799621582,
+        "mean_mps": 0.03554261137625381,
+        "median_mps": 0.0002470372710376978,
+        "p95_mps": 0.3168647289276123
+      },
+      "Town11": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.08004,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.9013012051582336,
+        "mean_mps": 0.1593329413601011,
+        "median_mps": 0.10688986629247665,
+        "p95_mps": 0.6349600553512563
+      },
+      "Town12": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.0,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.3639880120754242,
+        "mean_mps": 0.050624328400152734,
+        "median_mps": 0.027735048905014992,
+        "p95_mps": 0.21183561608195287
+      },
+      "Town13": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.0,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.31393468379974365,
+        "mean_mps": 0.0701563702583639,
+        "median_mps": 0.05029846727848053,
+        "p95_mps": 0.20529818534851074
+      },
+      "Town15": {
+        "count": 50000,
+        "fraction_above_0_5_mps": 0.04,
+        "fraction_above_1_mps": 0.0,
+        "fraction_above_2_mps": 0.0,
+        "max_mps": 0.728956401348114,
+        "mean_mps": 0.23408395147070288,
+        "median_mps": 0.20299264788627625,
+        "p95_mps": 0.4761376678943634
+      }
+    }
+  },
+  "velocity_semantics": {
+    "pose_derived_velocity_mps": "Finite difference of released endpoint positions using a centered 10-ms span and one-sided boundary spans.",
+    "route_command_velocity_mps": "Original prescribed route-state velocity retained for compatibility.",
+    "velocity_closure_error_mps": "Euclidean norm of pose-derived minus route-command velocity.",
+    "velocity_validity_flag": "Closure error <= 0.5 m/s; a diagnostic threshold, not a physical-validity guarantee."
+  }
+}

metadata/kinematics/pose_velocity_v1_1_1.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50b40b3a7e3880f31fdccb573273aa14669bb55baadba04cbfa228226aa7c787
+size 1584801

metadata/kinematics/schema.json

@@ -0,0 +1,48 @@
+{
+  "$id": "https://huggingface.co/datasets/mfzheng/Scene2Wave/raw/main/metadata/kinematics/schema.json",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "additionalProperties": true,
+  "properties": {
+    "dataset_version": {
+      "const": "1.1.1"
+    },
+    "production_fix": {
+      "type": "object"
+    },
+    "schema_version": {
+      "const": "scene2wave_kinematics_patch_v1"
+    },
+    "scope": {
+      "const": "metadata_only_additive_patch"
+    },
+    "sidecar": {
+      "const": "metadata/kinematics/pose_velocity_v1_1_1.npz"
+    },
+    "sidecar_sha256": {
+      "pattern": "^[0-9a-f]{64}$",
+      "type": "string"
+    },
+    "source_dataset_version": {
+      "const": "1.1.0"
+    },
+    "statistics": {
+      "type": "object"
+    },
+    "velocity_semantics": {
+      "type": "object"
+    }
+  },
+  "required": [
+    "schema_version",
+    "dataset_version",
+    "source_dataset_version",
+    "scope",
+    "sidecar",
+    "sidecar_sha256",
+    "velocity_semantics",
+    "production_fix",
+    "statistics"
+  ],
+  "title": "Scene2Wave v1.1.1 kinematics patch manifest",
+  "type": "object"
+}

metadata/paper_reproducibility/receipts/array_doppler_audit.json

@@ -0,0 +1,903 @@
+{
+  "array_contract": {
+    "rx": "1x2 planar, 0.5-lambda horizontal spacing, dipole/V",
+    "rx_normalized_y_lambda": [
+      -0.25,
+      0.25
+    ],
+    "synthetic_array": true,
+    "tx": "1x32 planar, 0.5-lambda horizontal spacing, dipole/V",
+    "tx_normalized_y_lambda": [
+      -7.75,
+      7.75
+    ]
+  },
+  "center_equivalent_vs_position_increment": {
+    "episode_bootstrap_linear_fit_95ci": {
+      "intercept_hz": [
+        -8.350328281484139e-05,
+        1.3319877477884689e-05
+      ],
+      "replicates_requested": 10000,
+      "replicates_valid": 10000,
+      "seed": 20260824,
+      "slope": [
+        0.99999948334831,
+        1.000000014804258
+      ]
+    },
+    "episode_count": 24,
+    "episode_mean_bias_hz": -2.1684125403236113e-05,
+    "episode_mean_linear_fit": {
+      "episode_count": 14,
+      "intercept_hz": -2.6597404747657414e-05,
+      "slope": 0.9999997583009526
+    },
+    "episode_mean_mae_hz": 0.10046421192500336,
+    "episode_mean_max_abs_hz": 0.4361745479695031,
+    "episode_mean_median_abs_hz": 0.08363949805165914,
+    "episode_mean_p95_abs_hz": 0.24793345898848937,
+    "episode_mean_pearson_r": 0.9999999999998328,
+    "pair_count": 24386,
+    "pooled_max_abs_hz": 0.692259747610251
+  },
+  "center_equivalent_vs_stored_velocity": {
+    "episode_bootstrap_linear_fit_95ci": {
+      "intercept_hz": [
+        -0.25610807983767153,
+        1.4637073570378036
+      ],
+      "replicates_requested": 10000,
+      "replicates_valid": 10000,
+      "seed": 20260824,
+      "slope": [
+        0.9981855545784054,
+        1.0087411799668777
+      ]
+    },
+    "episode_count": 24,
+    "episode_mean_bias_hz": 0.44569819283990736,
+    "episode_mean_linear_fit": {
+      "episode_count": 14,
+      "intercept_hz": 0.49202777151407323,
+      "slope": 1.0022301919972967
+    },
+    "episode_mean_mae_hz": 2.3063294819169786,
+    "episode_mean_max_abs_hz": 12.47455763247849,
+    "episode_mean_median_abs_hz": 1.4070025822853423,
+    "episode_mean_p95_abs_hz": 9.116030959632598,
+    "episode_mean_pearson_r": 0.9999516429280418,
+    "pair_count": 24386,
+    "pooled_max_abs_hz": 54.458316091199194
+  },
+  "episodes": [
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "core",
+      "sample_id": "Req_Town01_dynamic_60kmh_t004_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town01"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 2.1780301298352627,
+        "mae_hz": 4.568577960815057,
+        "max_abs_hz": 53.32968426306036,
+        "median_abs_hz": 1.2782775505464912,
+        "p95_abs_hz": 52.55348841573768,
+        "pair_count": 1951
+      },
+      "profile": "multipath",
+      "sample_id": "Req_Town01_dynamic_60kmh_t018_rfsupp02",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -4.834824380028765e-05,
+        "expected_mean_hz": 25.230081733530284,
+        "mae_hz": 0.11172207014062402,
+        "max_abs_hz": 0.44776769943476324,
+        "median_abs_hz": 0.09522780335962011,
+        "observed_mean_hz": 25.23003338528648,
+        "p95_abs_hz": 0.2707496407537917,
+        "pair_count": 1951
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 3.4455588961017924,
+        "expected_mean_hz": 21.78447448918469,
+        "mae_hz": 3.5629945159360803,
+        "max_abs_hz": 54.458316091199194,
+        "median_abs_hz": 0.11204220129244646,
+        "observed_mean_hz": 25.23003338528648,
+        "p95_abs_hz": 53.67364280848319,
+        "pair_count": 1951
+      },
+      "town": "Town01"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "core",
+      "sample_id": "Req_Town02_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town02"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "scattering_mild",
+      "sample_id": "Req_Town02_dynamic_60kmh_t018_rfsupp02",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town02"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "scattering_medium",
+      "sample_id": "Req_Town03_dynamic_60kmh_t004_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town03"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "core",
+      "sample_id": "Req_Town03_dynamic_60kmh_t009_rfcarlasafe01",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town03"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "multipath",
+      "sample_id": "Req_Town04_dynamic_60kmh_t000_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town04"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 0.19862045304662435,
+        "mae_hz": 0.3230334144879339,
+        "max_abs_hz": 1.3652650169206595,
+        "median_abs_hz": 0.281360142730648,
+        "p95_abs_hz": 0.7806839331140224,
+        "pair_count": 1910
+      },
+      "profile": "core",
+      "sample_id": "Req_Town04_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -5.046270725020597e-05,
+        "expected_mean_hz": 186.0005909378313,
+        "mae_hz": 0.1478194926008668,
+        "max_abs_hz": 0.6407973934680342,
+        "median_abs_hz": 0.1240383463600665,
+        "observed_mean_hz": 186.00054047512407,
+        "p95_abs_hz": 0.36878624281835354,
+        "pair_count": 1910
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 0.00011507257659682015,
+        "expected_mean_hz": 186.00042540254748,
+        "mae_hz": 0.28327231847885,
+        "max_abs_hz": 1.1438854684051023,
+        "median_abs_hz": 0.25384661167917955,
+        "observed_mean_hz": 186.00054047512407,
+        "p95_abs_hz": 0.6594820366031312,
+        "pair_count": 1910
+      },
+      "town": "Town04"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 0.33354087442509894,
+        "mae_hz": 0.3663986259794079,
+        "max_abs_hz": 1.1741671401566265,
+        "median_abs_hz": 0.36513785266840415,
+        "p95_abs_hz": 0.7289226015533146,
+        "pair_count": 1981
+      },
+      "profile": "core",
+      "sample_id": "Req_Town05_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 9.793502985164496e-05,
+        "expected_mean_hz": -59.999700033656666,
+        "mae_hz": 0.14241090644199186,
+        "max_abs_hz": 0.6201992160641865,
+        "median_abs_hz": 0.11674825217716744,
+        "observed_mean_hz": -59.99960209862681,
+        "p95_abs_hz": 0.35191400758571945,
+        "pair_count": 1981
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -3.253752925450389e-05,
+        "expected_mean_hz": -59.99956956109755,
+        "mae_hz": 0.2057046382517998,
+        "max_abs_hz": 1.0383999285376078,
+        "median_abs_hz": 0.17401211848698495,
+        "observed_mean_hz": -59.99960209862681,
+        "p95_abs_hz": 0.5054049678213772,
+        "pair_count": 1981
+      },
+      "town": "Town05"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -1.7058159361523086,
+        "mae_hz": 1.7408948622779443,
+        "max_abs_hz": 2.430344062010782,
+        "median_abs_hz": 1.8360044604877475,
+        "p95_abs_hz": 2.0897565914582157,
+        "pair_count": 1497
+      },
+      "profile": "scattering_mild",
+      "sample_id": "Req_Town05_dynamic_60kmh_t007_rfcarlasafe01",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -9.029098345505752e-05,
+        "expected_mean_hz": -10.043338636365384,
+        "mae_hz": 0.11701148885089156,
+        "max_abs_hz": 0.5885467843377912,
+        "median_abs_hz": 0.09640367942808581,
+        "observed_mean_hz": -10.043428927348838,
+        "p95_abs_hz": 0.2996531347817499,
+        "pair_count": 1497
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 0.1407774794389621,
+        "expected_mean_hz": -10.184206406787801,
+        "mae_hz": 0.2605259552125276,
+        "max_abs_hz": 2.5671763116339754,
+        "median_abs_hz": 0.11426138698815436,
+        "observed_mean_hz": -10.043428927348838,
+        "p95_abs_hz": 2.0435341662167867,
+        "pair_count": 1497
+      },
+      "town": "Town05"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 1.0863362667876695,
+        "mae_hz": 1.0864719952587398,
+        "max_abs_hz": 2.1653989527572435,
+        "median_abs_hz": 1.0253584511260954,
+        "p95_abs_hz": 2.0135765003238473,
+        "pair_count": 1429
+      },
+      "profile": "core",
+      "sample_id": "Req_Town06_dynamic_60kmh_t006_rfcarlasafe01",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 3.9196044047567e-05,
+        "expected_mean_hz": -197.08324356098103,
+        "mae_hz": 0.0392917156897992,
+        "max_abs_hz": 0.20735500101136495,
+        "median_abs_hz": 0.0321449230689268,
+        "observed_mean_hz": -197.083204364937,
+        "p95_abs_hz": 0.09875049528507701,
+        "pair_count": 1429
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 0.00016954152302083567,
+        "expected_mean_hz": -197.08337390646003,
+        "mae_hz": 0.09981853927638518,
+        "max_abs_hz": 0.8074253807943137,
+        "median_abs_hz": 0.05624921806580119,
+        "observed_mean_hz": -197.083204364937,
+        "p95_abs_hz": 0.642898040671718,
+        "pair_count": 1429
+      },
+      "town": "Town06"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 0.1093556518534658,
+        "mae_hz": 0.5182785799713547,
+        "max_abs_hz": 1.481113944709108,
+        "median_abs_hz": 0.5462355677704807,
+        "p95_abs_hz": 1.148737019467893,
+        "pair_count": 1999
+      },
+      "profile": "scattering_medium",
+      "sample_id": "Req_Town06_dynamic_60kmh_t007_rfcarlasafe01",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 6.562645235504092e-05,
+        "expected_mean_hz": -204.7494303470079,
+        "mae_hz": 0.1521383318080777,
+        "max_abs_hz": 0.692259747610251,
+        "median_abs_hz": 0.1201100752949742,
+        "observed_mean_hz": -204.74936472055555,
+        "p95_abs_hz": 0.3823951643135899,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 0.01610547556988478,
+        "expected_mean_hz": -204.76547019612545,
+        "mae_hz": 0.5111257303950685,
+        "max_abs_hz": 1.3809559997088456,
+        "median_abs_hz": 0.5537050438256301,
+        "observed_mean_hz": -204.74936472055555,
+        "p95_abs_hz": 1.0450271813024754,
+        "pair_count": 1999
+      },
+      "town": "Town06"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "core",
+      "sample_id": "Req_Town07_dynamic_60kmh_t000_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town07"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 1.7644763722050534,
+        "mae_hz": 1.7644763722050534,
+        "max_abs_hz": 2.3428993405133944,
+        "median_abs_hz": 1.7591693884389592,
+        "p95_abs_hz": 2.0359498238240445,
+        "pair_count": 1398
+      },
+      "profile": "multipath",
+      "sample_id": "Req_Town07_dynamic_60kmh_t002_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -0.00025765910509435394,
+        "expected_mean_hz": -18.606575173839147,
+        "mae_hz": 0.1198122511963717,
+        "max_abs_hz": 0.5546682019625209,
+        "median_abs_hz": 0.10019766698323096,
+        "observed_mean_hz": -18.606832832944242,
+        "p95_abs_hz": 0.29292213297615866,
+        "pair_count": 1398
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -0.0002716678414155175,
+        "expected_mean_hz": -18.60656116510283,
+        "mae_hz": 0.12113052888406523,
+        "max_abs_hz": 0.5700293338223013,
+        "median_abs_hz": 0.1009694516241808,
+        "observed_mean_hz": -18.606832832944242,
+        "p95_abs_hz": 0.29447583025795576,
+        "pair_count": 1398
+      },
+      "town": "Town07"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "scattering_mild",
+      "sample_id": "Req_Town10HD_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town10HD"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": 5.259215102611178,
+        "mae_hz": 5.259215102611178,
+        "max_abs_hz": 15.457761247726694,
+        "median_abs_hz": 1.2316858148382437,
+        "p95_abs_hz": 15.212542155740337,
+        "pair_count": 227
+      },
+      "profile": "core",
+      "sample_id": "Req_Town10HD_dynamic_60kmh_t003_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -9.872423125077071e-05,
+        "expected_mean_hz": 157.05509590330718,
+        "mae_hz": 0.07097906083214686,
+        "max_abs_hz": 0.2299380551251886,
+        "median_abs_hz": 0.06404449371203214,
+        "observed_mean_hz": 157.0549971790759,
+        "p95_abs_hz": 0.16669162602532878,
+        "pair_count": 227
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 4.150407006061702,
+        "expected_mean_hz": 152.9045901730142,
+        "mae_hz": 4.216994724632675,
+        "max_abs_hz": 14.474229006719469,
+        "median_abs_hz": 0.12334394545356986,
+        "observed_mean_hz": 157.0549971790759,
+        "p95_abs_hz": 14.237248595039334,
+        "pair_count": 227
+      },
+      "town": "Town10HD"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -1.733019321357366,
+        "mae_hz": 6.777227009713237,
+        "max_abs_hz": 29.920462638927894,
+        "median_abs_hz": 5.374930444171831,
+        "p95_abs_hz": 17.975117210304663,
+        "pair_count": 1999
+      },
+      "profile": "scattering_medium",
+      "sample_id": "Req_Town11_dynamic_60kmh_t000_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -7.510545110149749e-05,
+        "expected_mean_hz": 59.12181473478833,
+        "mae_hz": 0.0989489215596323,
+        "max_abs_hz": 0.39731615132107834,
+        "median_abs_hz": 0.08336630094806452,
+        "observed_mean_hz": 59.12173962933723,
+        "p95_abs_hz": 0.23666397063262393,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -1.425078508242518,
+        "expected_mean_hz": 60.54681813757974,
+        "mae_hz": 6.724521298788013,
+        "max_abs_hz": 29.492435837215048,
+        "median_abs_hz": 5.316982916103363,
+        "observed_mean_hz": 59.12173962933723,
+        "p95_abs_hz": 17.86483307190962,
+        "pair_count": 1999
+      },
+      "town": "Town11"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -0.1697131620943258,
+        "mae_hz": 8.294031858143185,
+        "max_abs_hz": 36.7419970805698,
+        "median_abs_hz": 5.979647911093281,
+        "p95_abs_hz": 16.507932581816547,
+        "pair_count": 1999
+      },
+      "profile": "core",
+      "sample_id": "Req_Town11_dynamic_60kmh_t003_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 4.255404770450356e-05,
+        "expected_mean_hz": 203.39264059957458,
+        "mae_hz": 0.10180888658269395,
+        "max_abs_hz": 0.415016602115287,
+        "median_abs_hz": 0.08552978049195303,
+        "observed_mean_hz": 203.39268315362233,
+        "p95_abs_hz": 0.2533887802187337,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -0.37637294684543027,
+        "expected_mean_hz": 203.7690561004677,
+        "mae_hz": 8.277657219662013,
+        "max_abs_hz": 36.78328731365045,
+        "median_abs_hz": 6.159254977848946,
+        "observed_mean_hz": 203.39268315362233,
+        "p95_abs_hz": 16.68595138788204,
+        "pair_count": 1999
+      },
+      "town": "Town11"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -0.10414106043746098,
+        "mae_hz": 0.9570452609147636,
+        "max_abs_hz": 4.276876084204247,
+        "median_abs_hz": 0.8072109940267183,
+        "p95_abs_hz": 2.275654265790385,
+        "pair_count": 1999
+      },
+      "profile": "multipath",
+      "sample_id": "Req_Town12_dynamic_60kmh_t000_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 6.538035372208607e-05,
+        "expected_mean_hz": -205.00849725982926,
+        "mae_hz": 0.12733118286446013,
+        "max_abs_hz": 0.5153024537334261,
+        "median_abs_hz": 0.10358855736023997,
+        "observed_mean_hz": -205.00843187947552,
+        "p95_abs_hz": 0.3179478324957358,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -0.010526514766261567,
+        "expected_mean_hz": -204.99790536470925,
+        "mae_hz": 0.9345649079889499,
+        "max_abs_hz": 4.534657072236996,
+        "median_abs_hz": 0.7514080600751356,
+        "observed_mean_hz": -205.00843187947552,
+        "p95_abs_hz": 2.3600818172012423,
+        "pair_count": 1999
+      },
+      "town": "Town12"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -0.06689553023851282,
+        "mae_hz": 0.9401786883235674,
+        "max_abs_hz": 3.911266159385832,
+        "median_abs_hz": 0.7730622120102169,
+        "p95_abs_hz": 2.3163035969319794,
+        "pair_count": 1999
+      },
+      "profile": "core",
+      "sample_id": "Req_Town12_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": 2.8666023482692854e-05,
+        "expected_mean_hz": -205.40903498892854,
+        "mae_hz": 0.12620259284873273,
+        "max_abs_hz": 0.5677626921180376,
+        "median_abs_hz": 0.10686508747477319,
+        "observed_mean_hz": -205.40900632290504,
+        "p95_abs_hz": 0.30507966645353407,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -0.010964871958912423,
+        "expected_mean_hz": -205.39804145094615,
+        "mae_hz": 0.9238679986106607,
+        "max_abs_hz": 3.849031208988009,
+        "median_abs_hz": 0.7209695041012196,
+        "observed_mean_hz": -205.40900632290504,
+        "p95_abs_hz": 2.3752710272076314,
+        "pair_count": 1999
+      },
+      "town": "Town12"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -2.2274451181548156,
+        "mae_hz": 3.203971947093018,
+        "max_abs_hz": 13.836141204031918,
+        "median_abs_hz": 2.8187143775115544,
+        "p95_abs_hz": 7.4643333313732425,
+        "pair_count": 1999
+      },
+      "profile": "scattering_mild",
+      "sample_id": "Req_Town13_dynamic_60kmh_t001_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -1.2820034664669861e-05,
+        "expected_mean_hz": -91.75839721008171,
+        "mae_hz": 0.023181752839282792,
+        "max_abs_hz": 0.11754967748251488,
+        "median_abs_hz": 0.0188474255429405,
+        "observed_mean_hz": -91.75841003011638,
+        "p95_abs_hz": 0.059200506274908075,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": 0.4306784709374888,
+        "expected_mean_hz": -92.18908850105387,
+        "mae_hz": 2.5633750648593288,
+        "max_abs_hz": 10.674815018506564,
+        "median_abs_hz": 2.2865475528990657,
+        "observed_mean_hz": -91.75841003011638,
+        "p95_abs_hz": 6.296059828350988,
+        "pair_count": 1999
+      },
+      "town": "Town13"
+    },
+    {
+      "direct_multiplicity_max": 1,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": -0.07334066154706237,
+        "mae_hz": 3.6262482648676624,
+        "max_abs_hz": 13.785308991742959,
+        "median_abs_hz": 2.9127324828742474,
+        "p95_abs_hz": 9.155738184370042,
+        "pair_count": 1999
+      },
+      "profile": "core",
+      "sample_id": "Req_Town13_dynamic_60kmh_t002_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": -9.524950191997772e-06,
+        "expected_mean_hz": 77.26477013037906,
+        "mae_hz": 0.027840312694475455,
+        "max_abs_hz": 0.11196399578859939,
+        "median_abs_hz": 0.023840580521152788,
+        "observed_mean_hz": 77.26476060542886,
+        "p95_abs_hz": 0.06692522522354664,
+        "pair_count": 1999
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": -0.1207901952669527,
+        "expected_mean_hz": 77.38555080069581,
+        "mae_hz": 3.603059305861286,
+        "max_abs_hz": 12.869162883280978,
+        "median_abs_hz": 2.974443163551115,
+        "observed_mean_hz": 77.26476060542886,
+        "p95_abs_hz": 8.94052267590888,
+        "pair_count": 1999
+      },
+      "town": "Town13"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "scattering_medium",
+      "sample_id": "Req_Town15_dynamic_60kmh_t002_rfmain",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town15"
+    },
+    {
+      "direct_multiplicity_max": 0,
+      "legacy_rx0_tx0_vs_center_geometry": {
+        "bias_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "profile": "core",
+      "sample_id": "Req_Town15_dynamic_60kmh_t020_rfsupp03",
+      "symmetric_center_equivalent_vs_position_increment_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "symmetric_center_equivalent_vs_stored_velocity_geometry": {
+        "bias_hz": NaN,
+        "expected_mean_hz": NaN,
+        "mae_hz": NaN,
+        "max_abs_hz": NaN,
+        "median_abs_hz": NaN,
+        "observed_mean_hz": NaN,
+        "p95_abs_hz": NaN,
+        "pair_count": 0
+      },
+      "town": "Town15"
+    }
+  ],
+  "legacy": {
+    "episode_count": 24,
+    "episode_mean_bias_hz": 0.3463717186273215,
+    "episode_mean_mae_hz": 2.8161464244758645,
+    "episode_mean_max_abs_hz": 13.015620437622681,
+    "episode_mean_median_abs_hz": 1.9278234035924942,
+    "episode_mean_p95_abs_hz": 9.447052586557588,
+    "pair_count": 24386
+  },
+  "scope": "24 frozen 60-km/h episodes; all 2,000 states per episode",
+  "semantic_note": "Legacy Rx0/Tx0 phase is an element-pair quantity. The symmetric product cancels array offsets and is compared with endpoint-center geometry.",
+  "status": "pass"
+}

metadata/paper_reproducibility/receipts/generation_batch_audit.json

@@ -0,0 +1,495 @@
+{
+  "all_checked_invariants_equal": true,
+  "batches": [
+    {
+      "generation_config": "generation_configs/original_4town.yaml",
+      "generation_config_sha256": "e973af8415e2a831c32a6379a64dfb34221bf1b7773754d5196ff822d7d1b0d8",
+      "id": "original_4town_v1_0_1",
+      "qualification": {
+        "status": "published_release_validated"
+      },
+      "sample_count": 100,
+      "source_code_revision": "995f6fd60a65c69e844c994ab572dac83850b95e",
+      "source_experiment": "2026-08-01_rf_main_training_full_v2",
+      "source_manifest_sha256": "cfabf4cdff23b6c36e1653f835782dbc0ae61e8c04d789a290ce8ea245026df6",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town15"
+      ]
+    },
+    {
+      "generation_config": "generation_configs/extension_8town.yaml",
+      "generation_config_sha256": "094e2d9bf5c7f3ac53d1b7399adf2b0e792034786076cd8d9ef121ca9524d594",
+      "id": "extension_8town_v1_1_0",
+      "qualification": {
+        "final_qa_status": "pass",
+        "final_qa_summary_sha256": "23dae67eccc1c2c54f2f909acaccfa239e2dabcfe6fa35cf501c7342e0900737",
+        "final_qa_supersedes_historical_execution_failure": true,
+        "historical_execution_ok": false,
+        "historical_failure": {
+          "message": "offline_reference_qa failed with return code 1",
+          "type": "RuntimeError"
+        }
+      },
+      "sample_count": 200,
+      "source_code_revision": "4e3e60e7c12e80185e7d18363a145fd1f6de037c",
+      "source_experiment": "2026-08-10_rf_main_training_8town_full_v1",
+      "source_manifest_sha256": "85be12793997cf03385d4201adee963a976504d0b3ecd2c9656dbcd47d4da2bb",
+      "towns": [
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town11",
+        "Town12",
+        "Town13"
+      ]
+    }
+  ],
+  "config_sha256": {
+    "extension": "094e2d9bf5c7f3ac53d1b7399adf2b0e792034786076cd8d9ef121ca9524d594",
+    "original": "e973af8415e2a831c32a6379a64dfb34221bf1b7773754d5196ff822d7d1b0d8"
+  },
+  "confounding": {
+    "allowed_claim": "Frozen radio, sensor, profile, duration, rate, and seed semantics are inherited consistently across both generation configurations.",
+    "disallowed_claim": "Observed channel-statistic differences prove or exclude a generation-batch effect.",
+    "reason": "The original batch contains Town01/02/03/15 and the extension contains the other eight towns; no Town was regenerated under both production batches.",
+    "town_and_batch_are_fully_confounded": true
+  },
+  "descriptive_statistics_by_batch": {
+    "extension_8town_v1_1_0": {
+      "background_vehicle_count_mean": {
+        "ci95_high": 12.685,
+        "ci95_low": 11.935,
+        "mean": 12.305,
+        "n_samples": 200
+      },
+      "empty_frame_fraction": {
+        "ci95_high": 0.0,
+        "ci95_low": 0.0,
+        "mean": 0.0,
+        "n_samples": 200
+      },
+      "los_frame_fraction": {
+        "ci95_high": 0.747336125,
+        "ci95_low": 0.6288838125,
+        "mean": 0.688395,
+        "n_samples": 200
+      },
+      "los_radial_doppler_hz_mean": {
+        "ci95_high": 11.731609638809964,
+        "ci95_low": -9.270975885828205,
+        "mean": 1.2531492817407872,
+        "n_samples": 200
+      },
+      "nlos_frame_fraction": {
+        "ci95_high": 0.3711161874999999,
+        "ci95_low": 0.252663875,
+        "mean": 0.31160499999999997,
+        "n_samples": 200
+      },
+      "path_count_mean": {
+        "ci95_high": 438.84439775,
+        "ci95_low": 323.86138062500004,
+        "mean": 379.10441,
+        "n_samples": 200
+      },
+      "received_path_power_db_mean": {
+        "ci95_high": -58.24467099865344,
+        "ci95_low": -62.060071633636454,
+        "mean": -60.146237437339586,
+        "n_samples": 200
+      },
+      "rms_delay_spread_mean_s": {
+        "ci95_high": 7.390870850458367e-08,
+        "ci95_low": 5.4473699748112994e-08,
+        "mean": 6.337106834660404e-08,
+        "n_samples": 200
+      },
+      "rsu_cav_distance_m_mean": {
+        "ci95_high": 75.63177280526095,
+        "ci95_low": 65.62407027477565,
+        "mean": 70.75265837821543,
+        "n_samples": 200
+      }
+    },
+    "original_4town_v1_0_1": {
+      "background_vehicle_count_mean": {
+        "ci95_high": 12.74,
+        "ci95_low": 11.96,
+        "mean": 12.35,
+        "n_samples": 100
+      },
+      "empty_frame_fraction": {
+        "ci95_high": 0.0,
+        "ci95_low": 0.0,
+        "mean": 0.0,
+        "n_samples": 100
+      },
+      "los_frame_fraction": {
+        "ci95_high": 0.53142875,
+        "ci95_low": 0.33754624999999994,
+        "mean": 0.43162000000000006,
+        "n_samples": 100
+      },
+      "los_radial_doppler_hz_mean": {
+        "ci95_high": 23.12364371555097,
+        "ci95_low": -10.362591113929513,
+        "mean": 6.09185513853158,
+        "n_samples": 100
+      },
+      "nlos_frame_fraction": {
+        "ci95_high": 0.6624537500000001,
+        "ci95_low": 0.46857125,
+        "mean": 0.56838,
+        "n_samples": 100
+      },
+      "path_count_mean": {
+        "ci95_high": 283.0973546249999,
+        "ci95_low": 193.60862625,
+        "mean": 236.10453,
+        "n_samples": 100
+      },
+      "received_path_power_db_mean": {
+        "ci95_high": -62.14295698641665,
+        "ci95_low": -67.88397636250117,
+        "mean": -64.94601233718902,
+        "n_samples": 100
+      },
+      "rms_delay_spread_mean_s": {
+        "ci95_high": 7.940940388066186e-08,
+        "ci95_low": 5.9296231769906436e-08,
+        "mean": 6.933796670789699e-08,
+        "n_samples": 100
+      },
+      "rsu_cav_distance_m_mean": {
+        "ci95_high": 84.85720993179854,
+        "ci95_low": 70.0726385930937,
+        "mean": 77.7074118798415,
+        "n_samples": 100
+      }
+    }
+  },
+  "extension_only_procedural_changes": {
+    "cav_sampling_region_rules_file": "configs/data_generation/data_gen_v3.0/large_map_cav_sampling_regions.yaml",
+    "large_map_rf_support": {
+      "Town11": {
+        "endpoint_margin_m": 5.0,
+        "material_readiness": "standards_anchored_research",
+        "max_rf_xyz_m": [
+          6687.034668,
+          1524.283569,
+          203.090683
+        ],
+        "min_rf_xyz_m": [
+          6285.951172,
+          1205.440552,
+          -100.485283
+        ]
+      },
+      "Town12": {
+        "endpoint_margin_m": 5.0,
+        "material_readiness": "standards_anchored_research",
+        "max_rf_xyz_m": [
+          1114.900818,
+          -4863.20459,
+          521.821014
+        ],
+        "min_rf_xyz_m": [
+          801.409119,
+          -5127.114258,
+          221.339752
+        ]
+      },
+      "Town13": {
+        "endpoint_margin_m": 5.0,
+        "material_readiness": "standards_anchored_research",
+        "max_rf_xyz_m": [
+          3100.082031,
+          -926.871826,
+          319.909531
+        ],
+        "min_rf_xyz_m": [
+          2854.780273,
+          -1162.966431,
+          19.338928
+        ]
+      }
+    },
+    "large_map_roi": {
+      "mode": "trajectory_envelope_plus_explicit_propagation_margin",
+      "propagation_margin_m": 100.0,
+      "reject_contract_sample": true,
+      "require_sionna_0_19_2_full_scene_load": true
+    },
+    "town_set": [
+      "Town04",
+      "Town05",
+      "Town06",
+      "Town07",
+      "Town10HD",
+      "Town11",
+      "Town12",
+      "Town13"
+    ],
+    "vehicle_spawn_clearance_mode": "footprint_then_trajectory_audit"
+  },
+  "semantic_invariants": [
+    {
+      "equal": true,
+      "extension": 2000,
+      "original": 2000,
+      "path": "geometry_rate_hz"
+    },
+    {
+      "equal": true,
+      "extension": 1.0,
+      "original": 1.0,
+      "path": "sample_duration_s"
+    },
+    {
+      "equal": true,
+      "extension": 0.3,
+      "original": 0.3,
+      "path": "warmup_s"
+    },
+    {
+      "equal": true,
+      "extension": {
+        "enabled": true,
+        "fov": 90.0,
+        "height_m": 130.0,
+        "image_size_x": 1280,
+        "image_size_y": 1280,
+        "rate_hz": 20,
+        "yaw_deg": 0.0
+      },
+      "original": {
+        "enabled": true,
+        "fov": 90.0,
+        "height_m": 130.0,
+        "image_size_x": 1280,
+        "image_size_y": 1280,
+        "rate_hz": 20,
+        "yaw_deg": 0.0
+      },
+      "path": "birdview"
+    },
+    {
+      "equal": true,
+      "extension": 3700000000,
+      "original": 3700000000,
+      "path": "radio.center_frequency_hz"
+    },
+    {
+      "equal": true,
+      "extension": 15360000,
+      "original": 15360000,
+      "path": "radio.bandwidth_hz"
+    },
+    {
+      "equal": true,
+      "extension": 512,
+      "original": 512,
+      "path": "radio.num_subcarriers"
+    },
+    {
+      "equal": true,
+      "extension": 30000,
+      "original": 30000,
+      "path": "radio.subcarrier_spacing_hz"
+    },
+    {
+      "equal": true,
+      "extension": 32,
+      "original": 32,
+      "path": "radio.tx_antennas"
+    },
+    {
+      "equal": true,
+      "extension": 2,
+      "original": 2,
+      "path": "radio.rx_antennas"
+    },
+    {
+      "equal": true,
+      "extension": 55,
+      "original": 55,
+      "path": "vehicles.random_seed"
+    },
+    {
+      "equal": true,
+      "extension": "vehicle.tesla.model3",
+      "original": "vehicle.tesla.model3",
+      "path": "vehicles.cav_blueprint"
+    },
+    {
+      "equal": true,
+      "extension": {
+        "core": 0.4,
+        "multipath": 0.2,
+        "scattering_medium": 0.2,
+        "scattering_mild": 0.2
+      },
+      "original": {
+        "core": 0.4,
+        "multipath": 0.2,
+        "scattering_medium": 0.2,
+        "scattering_mild": 0.2
+      },
+      "path": "main_training.profile_weights"
+    },
+    {
+      "equal": true,
+      "extension": {
+        "camera_fov_deg": 110,
+        "camera_rate_hz": 20,
+        "cav_depth_views": 4,
+        "cav_lidar_vertical_fov_deg": [
+          -10,
+          90
+        ],
+        "cav_rgb_views": 4,
+        "gnss_rate_hz": 10,
+        "imu_rate_hz": 100,
+        "lidar_rate_hz": 20,
+        "radar_rate_hz": 20,
+        "rsu_camera_view_modes": [
+          "east",
+          "north",
+          "west",
+          "south"
+        ],
+        "rsu_depth_views": 4,
+        "rsu_height_m": 5.0,
+        "rsu_lidar_vertical_fov_deg": [
+          -50,
+          50
+        ],
+        "rsu_radar_horizontal_fov_deg": 120,
+        "rsu_radar_vertical_fov_deg": 30,
+        "rsu_radar_yaw_policy": "face_road",
+        "rsu_rgb_views": 4
+      },
+      "original": {
+        "camera_fov_deg": 110,
+        "camera_rate_hz": 20,
+        "cav_depth_views": 4,
+        "cav_lidar_vertical_fov_deg": [
+          -10,
+          90
+        ],
+        "cav_rgb_views": 4,
+        "gnss_rate_hz": 10,
+        "imu_rate_hz": 100,
+        "lidar_rate_hz": 20,
+        "radar_rate_hz": 20,
+        "rsu_camera_view_modes": [
+          "east",
+          "north",
+          "west",
+          "south"
+        ],
+        "rsu_depth_views": 4,
+        "rsu_height_m": 5.0,
+        "rsu_lidar_vertical_fov_deg": [
+          -50,
+          50
+        ],
+        "rsu_radar_horizontal_fov_deg": 120,
+        "rsu_radar_vertical_fov_deg": 30,
+        "rsu_radar_yaw_policy": "face_road",
+        "rsu_rgb_views": 4
+      },
+      "path": "sensors"
+    },
+    {
+      "equal": true,
+      "extension": {
+        "core_max_depth": 2,
+        "core_num_samples": 1000000,
+        "multipath_max_depth": 3,
+        "multipath_num_samples": 2000000,
+        "scattering_max_depth": 3,
+        "scattering_num_samples": 2000000,
+        "smoke_num_samples": 1000000,
+        "worker_policy_file": "configs/data_generation/data_gen_v2.0/rf_compiler_sionna_worker_policy_2x4090.json",
+        "worker_policy_fingerprint": "strict",
+        "worker_policy_mode": "adaptive",
+        "workers": 2
+      },
+      "original": {
+        "core_max_depth": 2,
+        "core_num_samples": 1000000,
+        "multipath_max_depth": 3,
+        "multipath_num_samples": 2000000,
+        "scattering_max_depth": 3,
+        "scattering_num_samples": 2000000,
+        "smoke_num_samples": 1000000,
+        "worker_policy_file": "configs/data_generation/data_gen_v2.0/rf_compiler_sionna_worker_policy_2x4090.json",
+        "worker_policy_fingerprint": "strict",
+        "worker_policy_mode": "adaptive",
+        "workers": 2
+      },
+      "path": "sionna"
+    },
+    {
+      "equal": true,
+      "extension": {
+        "carla": {
+          "gpu_devices": "0,1",
+          "shard_strategy": "round_robin",
+          "workers": 2
+        },
+        "sionna": {
+          "offline_qa": {
+            "enabled": true,
+            "frames_per_sample": 32,
+            "profiles": [
+              "core",
+              "multipath",
+              "scattering_mild",
+              "scattering_medium"
+            ],
+            "samples_per_bucket": 1,
+            "statistical_profile_policy_file": "configs/data_generation/data_gen_v1.0/config.yaml",
+            "strict_profiles": [
+              "core",
+              "multipath"
+            ]
+          }
+        }
+      },
+      "original": {
+        "carla": {
+          "gpu_devices": "0,1",
+          "shard_strategy": "round_robin",
+          "workers": 2
+        },
+        "sionna": {
+          "offline_qa": {
+            "enabled": true,
+            "frames_per_sample": 32,
+            "profiles": [
+              "core",
+              "multipath",
+              "scattering_mild",
+              "scattering_medium"
+            ],
+            "samples_per_bucket": 1,
+            "statistical_profile_policy_file": "configs/data_generation/data_gen_v1.0/config.yaml",
+            "strict_profiles": [
+              "core",
+              "multipath"
+            ]
+          }
+        }
+      },
+      "path": "acceleration"
+    }
+  ],
+  "status": "pass"
+}

metadata/paper_reproducibility/receipts/material_evidence_audit.json

@@ -0,0 +1,607 @@
+{
+  "asset_root": "assets/rf_scene_compiler/v1.9.3",
+  "claim_boundary": "Evidence provenance and implementation identity are audited. Some Town receipts use the Sionna built-in realization of the same standards-based class while others install the explicit catalog power law; semantic assignment and proxy models are not measurement calibration.",
+  "effective_material_count": 14,
+  "rows": [
+    {
+      "evidence_status": "standards_based",
+      "material": "itu_concrete",
+      "model": "itu_r_p2040_4_power_law; sionna_builtin",
+      "reference": "ITU-R P.2040-4 equations 57-58 and Table 3; project-defined assumption; no external calibration source",
+      "runtime_sources": [
+        "custom_frequency_model",
+        "sionna_builtin"
+      ],
+      "runtime_variant_count": 2,
+      "semantic_assumption": "none beyond the declared material-class assignment",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town11",
+        "Town12",
+        "Town13",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town11",
+          "Town15"
+        ],
+        "sionna_builtin": [
+          "Town12",
+          "Town13"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1.0,
+          100.0
+        ],
+        [
+          1.0,
+          100.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based",
+      "material": "itu_glass",
+      "model": "itu_r_p2040_4_power_law; sionna_builtin",
+      "reference": "ITU-R P.2040-4 equations 57-58 and Table 3; project-defined assumption; no external calibration source",
+      "runtime_sources": [
+        "custom_frequency_model",
+        "sionna_builtin"
+      ],
+      "runtime_variant_count": 2,
+      "semantic_assumption": "none beyond the declared material-class assignment",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town12",
+        "Town13",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town15"
+        ],
+        "sionna_builtin": [
+          "Town12",
+          "Town13"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          0.1,
+          100.0
+        ],
+        [
+          [
+            0.1,
+            100.0
+          ],
+          [
+            220.0,
+            450.0
+          ]
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based",
+      "material": "itu_metal",
+      "model": "itu_r_p2040_4_power_law; sionna_builtin",
+      "reference": "ITU-R P.2040-4 equations 57-58 and Table 3; project-defined assumption; no external calibration source",
+      "runtime_sources": [
+        "custom_frequency_model",
+        "sionna_builtin"
+      ],
+      "runtime_variant_count": 2,
+      "semantic_assumption": "none beyond the declared material-class assignment",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town12",
+        "Town13",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town15"
+        ],
+        "sionna_builtin": [
+          "Town12",
+          "Town13"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1.0,
+          100.0
+        ],
+        [
+          1e-09,
+          1000000.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based",
+      "material": "itu_wood",
+      "model": "itu_r_p2040_4_power_law",
+      "reference": "ITU-R P.2040-4 equations 57-58 and Table 3",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "none beyond the declared material-class assignment",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town06",
+        "Town07"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town06",
+          "Town07"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          0.001,
+          100.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_asphalt",
+      "model": "itu_r_p2040_4_power_law",
+      "reference": "ITU-R P.2040-4 equations 57-58 and Table 3",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "The ITU asphalt-concrete class is used for CARLA road asphalt",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town11",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town11",
+          "Town15"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1.0,
+          40.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_dry_vegetation",
+      "model": "itu_r_p527_6_vegetation_above_freezing",
+      "reference": "ITU-R P.527-6 equations 71-77 and Figure 18",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "The ITU dry-vegetation example is used for wheat and hay",
+      "towns": [
+        "Town06",
+        "Town07"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town06",
+          "Town07"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1e-09,
+          1000.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_fresh_water",
+      "model": "itu_r_p527_6_pure_water",
+      "reference": "ITU-R P.527-6 equations 5-13",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "Town07 lake water is treated as fresh pure water at 20 degC",
+      "towns": [
+        "Town01",
+        "Town04",
+        "Town07",
+        "Town10HD"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town04",
+          "Town07",
+          "Town10HD"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1e-09,
+          1000.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_green_vegetation",
+      "model": "itu_r_p527_6_vegetation_above_freezing",
+      "reference": "ITU-R P.527-6 equations 71-77 and Figure 17",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "A single green-vegetation moisture parameter is used for all CARLA green assets",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town12",
+        "Town13",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town12",
+          "Town13",
+          "Town15"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1e-09,
+          1000.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "project_assumption_requires_measurement",
+      "material": "mmw_plastic",
+      "model": "custom",
+      "reference": "project-defined assumption; no external calibration source",
+      "runtime_sources": [
+        "custom"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "none beyond the declared material-class assignment",
+      "towns": [
+        "Town12",
+        "Town13"
+      ],
+      "towns_by_runtime_source": {
+        "custom": [
+          "Town12",
+          "Town13"
+        ]
+      },
+      "valid_frequency_ghz": [
+        null
+      ]
+    },
+    {
+      "evidence_status": "peer_reviewed_model_with_semantic_assumption",
+      "material": "mmw_plastic_lens",
+      "model": "zhekov_2020_single_pole_cole_cole",
+      "reference": "Zhekov et al., IEEE Antennas and Propagation Magazine, 2020, DOI 10.1109/MAP.2019.2955680, Table II",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "PMMA is used as a transparent-polymer proxy because CARLA does not identify PMMA versus polycarbonate",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          0.2,
+          67.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "peer_reviewed_model_with_semantic_assumption",
+      "material": "mmw_plastic_trim",
+      "model": "zhekov_2020_single_pole_cole_cole",
+      "reference": "Zhekov et al., IEEE Antennas and Propagation Magazine, 2020, DOI 10.1109/MAP.2019.2955680, Table II",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "CARLA trim labels do not identify polymer chemistry or fillers",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town15"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town15"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          0.2,
+          67.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "project_assumption_requires_measurement",
+      "material": "mmw_rubber",
+      "model": "custom",
+      "reference": "project-defined assumption; no external calibration source",
+      "runtime_sources": [
+        "custom"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "Placeholder only; tyre compounds vary strongly with carbon-black and other fillers",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD"
+      ],
+      "towns_by_runtime_source": {
+        "custom": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD"
+        ]
+      },
+      "valid_frequency_ghz": [
+        null
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_terrain_composite",
+      "model": "itu_r_p527_6_soil",
+      "reference": "ITU-R P.527-6 equations 57-70, Table 2, and Figure 16",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "Town07 landscape texture layers are approximated by dry silty-loam soil",
+      "towns": [
+        "Town01",
+        "Town02",
+        "Town03",
+        "Town04",
+        "Town05",
+        "Town06",
+        "Town07",
+        "Town10HD",
+        "Town12",
+        "Town13"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town01",
+          "Town02",
+          "Town03",
+          "Town04",
+          "Town05",
+          "Town06",
+          "Town07",
+          "Town10HD",
+          "Town12",
+          "Town13"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1e-09,
+          1000.0
+        ]
+      ]
+    },
+    {
+      "evidence_status": "standards_based_with_scene_assumptions",
+      "material": "mmw_town11_landscape_composite",
+      "model": "itu_r_p527_6_soil",
+      "reference": "ITU-R P.527-6 equations 57-70 and Table 2",
+      "runtime_sources": [
+        "custom_frequency_model"
+      ],
+      "runtime_variant_count": 1,
+      "semantic_assumption": "The thin visual grass and mineral/concrete texture blend is approximated by an effective silty-loam substrate because CARLA does not encode soil composition, grass thickness, or a spatial material-layer mask.",
+      "towns": [
+        "Town11"
+      ],
+      "towns_by_runtime_source": {
+        "custom_frequency_model": [
+          "Town11"
+        ]
+      },
+      "valid_frequency_ghz": [
+        [
+          1e-09,
+          1000.0
+        ]
+      ]
+    }
+  ],
+  "runtime_variant_comparison_3_7ghz": [
+    {
+      "conductivity_absolute_difference_s_per_m": 0.0,
+      "explicit_catalog": {
+        "conductivity_s_per_m": 0.12855511586553203,
+        "relative_permittivity": 5.24
+      },
+      "frequency_ghz": 3.7,
+      "material": "itu_concrete",
+      "numerically_equal": true,
+      "relative_permittivity_absolute_difference": 0.0,
+      "sionna_0_19_2_builtin": {
+        "conductivity_s_per_m": 0.12855511586553203,
+        "relative_permittivity": 5.24
+      }
+    },
+    {
+      "conductivity_absolute_difference_s_per_m": 0.0,
+      "explicit_catalog": {
+        "conductivity_s_per_m": 0.020765954111764445,
+        "relative_permittivity": 6.31
+      },
+      "frequency_ghz": 3.7,
+      "material": "itu_glass",
+      "numerically_equal": true,
+      "relative_permittivity_absolute_difference": 0.0,
+      "sionna_0_19_2_builtin": {
+        "conductivity_s_per_m": 0.020765954111764445,
+        "relative_permittivity": 6.31
+      }
+    },
+    {
+      "conductivity_absolute_difference_s_per_m": 0.0,
+      "explicit_catalog": {
+        "conductivity_s_per_m": 10000000.0,
+        "relative_permittivity": 1.0
+      },
+      "frequency_ghz": 3.7,
+      "material": "itu_metal",
+      "numerically_equal": true,
+      "relative_permittivity_absolute_difference": 0.0,
+      "sionna_0_19_2_builtin": {
+        "conductivity_s_per_m": 10000000.0,
+        "relative_permittivity": 1.0
+      }
+    }
+  ],
+  "status": "pass",
+  "town_receipt_count": 12
+}

metadata/paper_reproducibility/scripts/analyze_scene2wave_array_doppler.py

@@ -0,0 +1,308 @@
+#!/usr/bin/env python3
+"""Audit array geometry and direct-path Doppler semantics for Scene2Wave.
+
+The production release uses a synthetic 1x32 Tx and 1x2 Rx planar array.
+Earlier temporal analysis compared the phase of Rx0/Tx0 with endpoint-center
+geometry.  This audit reports that legacy diagnostic separately and constructs
+an array-center-equivalent phase from the four symmetric central element pairs.
+For a direct path, multiplication of the normalized phasors at
+Rx{0,1}/Tx{15,16} cancels the symmetric synthetic-array offsets.  Dividing the
+adjacent product phase by four recovers the center phase increment.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import math
+from concurrent.futures import ProcessPoolExecutor, as_completed
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+
+import matplotlib
+
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+
+
+LIGHT_SPEED_MPS = 299_792_458.0
+TX_CENTER_INDICES = (15, 16)
+RX_CENTER_INDICES = (0, 1)
+BOOTSTRAP_REPLICATES = 10_000
+BOOTSTRAP_SEED = 20260824
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--dataset-root", type=Path, required=True)
+    parser.add_argument("--selection-json", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    parser.add_argument("--workers", type=int, default=4)
+    return parser.parse_args()
+
+
+def frame_number(path: Path) -> int:
+    return int(path.name.split("_", 1)[0])
+
+
+def channel_files(sample_root: Path) -> list[Path]:
+    directories = list((sample_root / "sionna").glob("*/cav_1"))
+    if len(directories) != 1:
+        raise RuntimeError(f"Expected one V2I directory: {sample_root}")
+    return sorted(directories[0].glob("*_paths.npz"), key=frame_number)
+
+
+def normalized_symmetric_product(amplitudes: np.ndarray, path_index: int) -> complex:
+    """Return the fourth-power array-center phasor for one path."""
+    values = amplitudes[np.ix_(RX_CENTER_INDICES, TX_CENTER_INDICES, [path_index])].reshape(-1)
+    magnitudes = np.abs(values)
+    if values.size != 4 or np.any(~np.isfinite(values)) or np.any(magnitudes == 0):
+        return complex(np.nan, np.nan)
+    phasors = values / magnitudes
+    return complex(np.prod(phasors))
+
+
+def center_phase_doppler(previous: complex, current: complex, sampling_rate_hz: float) -> float:
+    """Recover center Doppler from adjacent fourth-power center phasors."""
+    return float(np.angle(current * np.conj(previous)) * sampling_rate_hz / (8.0 * np.pi))
+
+
+def legacy_phase_doppler(previous: complex, current: complex, sampling_rate_hz: float) -> float:
+    return float(np.angle(current * np.conj(previous)) * sampling_rate_hz / (2.0 * np.pi))
+
+
+def geometry_doppler(data: Any) -> float:
+    tx_position = np.asarray(data["tx_position_m"], dtype=np.float64).reshape(3)
+    rx_position = np.asarray(data["rx_position_m"], dtype=np.float64).reshape(3)
+    tx_velocity = np.asarray(data["tx_velocity_mps"], dtype=np.float64).reshape(3)
+    rx_velocity = np.asarray(data["rx_velocity_mps"], dtype=np.float64).reshape(3)
+    direction = rx_position - tx_position
+    distance = float(np.linalg.norm(direction))
+    if distance == 0:
+        return math.nan
+    carrier_hz = float(np.asarray(data["center_frequency_hz"]).item())
+    radial_velocity = float(np.dot(rx_velocity - tx_velocity, direction / distance))
+    return -carrier_hz * radial_velocity / LIGHT_SPEED_MPS
+
+
+def position_increment_doppler(previous_distance_m: float, current_distance_m: float, carrier_hz: float, sampling_rate_hz: float) -> float:
+    """Doppler implied directly by the adjacent endpoint-center positions."""
+    return -carrier_hz * (current_distance_m - previous_distance_m) * sampling_rate_hz / LIGHT_SPEED_MPS
+
+
+def _audit_episode(task: tuple[dict[str, Any], str]) -> dict[str, Any]:
+    row, dataset_root_text = task
+    sample_root = Path(dataset_root_text) / row["relative_path"]
+    files = channel_files(sample_root)
+    if len(files) != 2000:
+        raise RuntimeError(f"Expected 2000 frames, got {len(files)}: {sample_root}")
+
+    legacy_residuals: list[float] = []
+    center_residuals: list[float] = []
+    center_position_residuals: list[float] = []
+    center_observed: list[float] = []
+    expected_values: list[float] = []
+    position_expected_values: list[float] = []
+    previous: tuple[int, complex, complex, float, float, float] | None = None
+    multiplicity_max = 0
+    for path in files:
+        with np.load(path, allow_pickle=False) as data:
+            amplitudes = np.asarray(data["a"])[0, 0, :, 0, :, :, 0]
+            if amplitudes.shape[:2] != (2, 32):
+                raise RuntimeError(f"Unexpected array shape {amplitudes.shape}: {path}")
+            types = np.asarray(data["path_types"], dtype=np.int32).reshape(-1)
+            direct = np.flatnonzero(types == 0)
+            multiplicity_max = max(multiplicity_max, int(direct.size))
+            if direct.size != 1:
+                previous = None
+                continue
+            index = int(direct[0])
+            legacy = complex(amplitudes[0, 0, index])
+            center = normalized_symmetric_product(amplitudes, index)
+            expected = geometry_doppler(data)
+            tx_position = np.asarray(data["tx_position_m"], dtype=np.float64).reshape(3)
+            rx_position = np.asarray(data["rx_position_m"], dtype=np.float64).reshape(3)
+            distance_m = float(np.linalg.norm(rx_position - tx_position))
+            carrier_hz = float(np.asarray(data["center_frequency_hz"]).item())
+            frame = frame_number(path)
+            rate = float(np.asarray(data["csi_sampling_frequency_hz"]).item())
+            if previous is not None and frame == previous[0] + 1:
+                legacy_hz = legacy_phase_doppler(previous[1], legacy, rate)
+                center_hz = center_phase_doppler(previous[2], center, rate)
+                expected_hz = 0.5 * (previous[3] + expected)
+                position_expected_hz = position_increment_doppler(previous[4], distance_m, carrier_hz, rate)
+                if np.all(np.isfinite([legacy_hz, center_hz, expected_hz, position_expected_hz])):
+                    legacy_residuals.append(legacy_hz - expected_hz)
+                    center_residuals.append(center_hz - expected_hz)
+                    center_position_residuals.append(center_hz - position_expected_hz)
+                    center_observed.append(center_hz)
+                    expected_values.append(expected_hz)
+                    position_expected_values.append(position_expected_hz)
+            previous = (frame, legacy, center, expected, distance_m, carrier_hz)
+
+    def compact(values: list[float]) -> dict[str, float | int]:
+        array = np.asarray(values, dtype=np.float64)
+        absolute = np.abs(array)
+        return {
+            "pair_count": int(array.size),
+            "bias_hz": float(np.mean(array)) if array.size else math.nan,
+            "mae_hz": float(np.mean(absolute)) if array.size else math.nan,
+            "median_abs_hz": float(np.median(absolute)) if array.size else math.nan,
+            "p95_abs_hz": float(np.percentile(absolute, 95)) if array.size else math.nan,
+            "max_abs_hz": float(np.max(absolute)) if array.size else math.nan,
+        }
+
+    center_stats = compact(center_residuals)
+    observed = np.asarray(center_observed, dtype=np.float64)
+    expected = np.asarray(expected_values, dtype=np.float64)
+    center_stats["observed_mean_hz"] = float(np.mean(observed)) if observed.size else math.nan
+    center_stats["expected_mean_hz"] = float(np.mean(expected)) if expected.size else math.nan
+    center_position_stats = compact(center_position_residuals)
+    position_expected = np.asarray(position_expected_values, dtype=np.float64)
+    center_position_stats["observed_mean_hz"] = float(np.mean(observed)) if observed.size else math.nan
+    center_position_stats["expected_mean_hz"] = float(np.mean(position_expected)) if position_expected.size else math.nan
+    return {
+        "sample_id": row["sample_id"],
+        "town": row["town"],
+        "profile": row["profile"],
+        "direct_multiplicity_max": multiplicity_max,
+        "legacy_rx0_tx0_vs_center_geometry": compact(legacy_residuals),
+        "symmetric_center_equivalent_vs_stored_velocity_geometry": center_stats,
+        "symmetric_center_equivalent_vs_position_increment_geometry": center_position_stats,
+    }
+
+
+def aggregate(rows: list[dict[str, Any]], key: str) -> dict[str, Any]:
+    metrics = ("bias_hz", "mae_hz", "median_abs_hz", "p95_abs_hz", "max_abs_hz")
+    pair_count = sum(int(row[key]["pair_count"]) for row in rows)
+    result: dict[str, Any] = {"episode_count": len(rows), "pair_count": pair_count}
+    for metric in metrics:
+        values = np.asarray([row[key][metric] for row in rows], dtype=np.float64)
+        result[f"episode_mean_{metric}"] = float(np.nanmean(values))
+    if key.startswith("symmetric"):
+        observed = np.asarray([row[key]["observed_mean_hz"] for row in rows], dtype=np.float64)
+        expected = np.asarray([row[key]["expected_mean_hz"] for row in rows], dtype=np.float64)
+        finite = np.isfinite(observed) & np.isfinite(expected)
+        observed = observed[finite]
+        expected = expected[finite]
+        slope, intercept = np.polyfit(expected, observed, 1)
+        result["episode_mean_pearson_r"] = float(np.corrcoef(observed, expected)[0, 1])
+        result["episode_mean_linear_fit"] = {
+            "slope": float(slope),
+            "intercept_hz": float(intercept),
+            "episode_count": int(expected.size),
+        }
+        rng = np.random.default_rng(BOOTSTRAP_SEED)
+        fits = np.empty((BOOTSTRAP_REPLICATES, 2), dtype=np.float64)
+        valid = 0
+        for _ in range(BOOTSTRAP_REPLICATES):
+            indices = rng.integers(0, expected.size, size=expected.size)
+            if np.ptp(expected[indices]) <= 1.0e-12:
+                continue
+            fits[valid] = np.polyfit(expected[indices], observed[indices], 1)
+            valid += 1
+        fits = fits[:valid]
+        result["episode_bootstrap_linear_fit_95ci"] = {
+            "replicates_requested": BOOTSTRAP_REPLICATES,
+            "replicates_valid": int(valid),
+            "seed": BOOTSTRAP_SEED,
+            "slope": [float(value) for value in np.percentile(fits[:, 0], [2.5, 97.5])],
+            "intercept_hz": [float(value) for value in np.percentile(fits[:, 1], [2.5, 97.5])],
+        }
+        result["pooled_max_abs_hz"] = float(np.nanmax([row[key]["max_abs_hz"] for row in rows]))
+    return result
+
+
+def plot_center_position_parity(rows: list[dict[str, Any]], destination: Path) -> None:
+    key = "symmetric_center_equivalent_vs_position_increment_geometry"
+    expected = np.asarray([row[key]["expected_mean_hz"] for row in rows], dtype=np.float64)
+    observed = np.asarray([row[key]["observed_mean_hz"] for row in rows], dtype=np.float64)
+    finite = np.isfinite(expected) & np.isfinite(observed)
+    expected = expected[finite]
+    observed = observed[finite]
+    towns = [row["town"] for row, keep in zip(rows, finite) if keep]
+    if not expected.size:
+        raise RuntimeError("No finite center-position Doppler episode means to plot")
+    limit = float(max(np.max(np.abs(expected)), np.max(np.abs(observed))) * 1.08)
+    fig, axis = plt.subplots(figsize=(5.4, 4.8))
+    town_order = sorted(set(towns))
+    colors = plt.get_cmap("tab20")(np.linspace(0.0, 1.0, len(town_order)))
+    for town, color in zip(town_order, colors):
+        mask = np.asarray([value == town for value in towns])
+        axis.scatter(expected[mask], observed[mask], s=32, color=color, label=town, alpha=0.9)
+    axis.plot([-limit, limit], [-limit, limit], color="#555555", linewidth=1.1, linestyle="--", label="Identity")
+    axis.set_xlim(-limit, limit)
+    axis.set_ylim(-limit, limit)
+    axis.set_aspect("equal", adjustable="box")
+    axis.set_xlabel("Position-derived device-center Doppler (Hz)")
+    axis.set_ylabel("Array-center phase Doppler (Hz)")
+    axis.set_title(f"Array-center Doppler parity, {expected.size} eligible episodes")
+    axis.grid(color="#D8D8D8", linewidth=0.7, alpha=0.7)
+    axis.spines[["top", "right"]].set_visible(False)
+    axis.legend(frameon=False, fontsize=6.8, ncol=3, loc="upper left")
+    fig.tight_layout()
+    destination.parent.mkdir(parents=True, exist_ok=True)
+    fig.savefig(destination, dpi=240, bbox_inches="tight", facecolor="white")
+    plt.close(fig)
+
+
+def main() -> int:
+    args = parse_args()
+    selection = json.loads(args.selection_json.read_text(encoding="utf-8"))["selection"]
+    tasks = [(row, str(args.dataset_root.resolve())) for row in selection]
+    rows = []
+    with ProcessPoolExecutor(max_workers=max(1, args.workers)) as executor:
+        futures = [executor.submit(_audit_episode, task) for task in tasks]
+        for index, future in enumerate(as_completed(futures), start=1):
+            rows.append(future.result())
+            print(json.dumps({"completed": index, "total": len(futures)}), flush=True)
+    rows.sort(key=lambda row: row["sample_id"])
+    report = {
+        "status": "pass",
+        "scope": "24 frozen 60-km/h episodes; all 2,000 states per episode",
+        "array_contract": {
+            "tx": "1x32 planar, 0.5-lambda horizontal spacing, dipole/V",
+            "rx": "1x2 planar, 0.5-lambda horizontal spacing, dipole/V",
+            "synthetic_array": True,
+            "tx_normalized_y_lambda": [-7.75, 7.75],
+            "rx_normalized_y_lambda": [-0.25, 0.25],
+        },
+        "semantic_note": "Legacy Rx0/Tx0 phase is an element-pair quantity. The symmetric product cancels array offsets and is compared with endpoint-center geometry.",
+        "legacy": aggregate(rows, "legacy_rx0_tx0_vs_center_geometry"),
+        "center_equivalent_vs_stored_velocity": aggregate(rows, "symmetric_center_equivalent_vs_stored_velocity_geometry"),
+        "center_equivalent_vs_position_increment": aggregate(rows, "symmetric_center_equivalent_vs_position_increment_geometry"),
+        "episodes": rows,
+    }
+    args.output_root.mkdir(parents=True, exist_ok=True)
+    (args.output_root / "array_doppler_audit.json").write_text(
+        json.dumps(report, indent=2, sort_keys=True) + "\n", encoding="utf-8"
+    )
+    plot_center_position_parity(rows, args.output_root / "array_center_position_doppler_parity.png")
+    center = report["center_equivalent_vs_stored_velocity"]
+    center_position = report["center_equivalent_vs_position_increment"]
+    legacy = report["legacy"]
+    lines = [
+        "# Scene2Wave Array/Doppler Semantic Audit",
+        "",
+        report["scope"],
+        "",
+        "| Definition | Adjacent pairs | Episode-mean MAE | Episode-mean bias | Episode-mean P95 abs. |",
+        "|---|---:|---:|---:|---:|",
+        f"| Legacy Rx0/Tx0 vs endpoint center | {legacy['pair_count']:,} | {legacy['episode_mean_mae_hz']:.4f} Hz | {legacy['episode_mean_bias_hz']:.4f} Hz | {legacy['episode_mean_p95_abs_hz']:.4f} Hz |",
+        f"| Symmetric center equivalent vs endpoint center | {center['pair_count']:,} | {center['episode_mean_mae_hz']:.4f} Hz | {center['episode_mean_bias_hz']:.4f} Hz | {center['episode_mean_p95_abs_hz']:.4f} Hz |",
+        f"| Symmetric center equivalent vs adjacent-position range rate | {center_position['pair_count']:,} | {center_position['episode_mean_mae_hz']:.4f} Hz | {center_position['episode_mean_bias_hz']:.4f} Hz | {center_position['episode_mean_p95_abs_hz']:.4f} Hz |",
+        "",
+        f"Center-equivalent episode-mean Pearson correlation: {center_position['episode_mean_pearson_r']:.6f} against adjacent-position range rate ({center['episode_mean_pearson_r']:.6f} against stored velocity).",
+        f"The position-derived episode-mean fit has slope {center_position['episode_mean_linear_fit']['slope']:.8f} and intercept {center_position['episode_mean_linear_fit']['intercept_hz']:.6f} Hz; its pooled maximum absolute pair error is {center_position['pooled_max_abs_hz']:.4f} Hz.",
+        "",
+        "The legacy definition is retained only as a diagnostic. Paper claims should use the center-equivalent result or explicitly name the antenna element pair.",
+    ]
+    (args.output_root / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8")
+    print(json.dumps({"status": "pass", "output_root": str(args.output_root)}))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

metadata/paper_reproducibility/scripts/analyze_scene2wave_data_contracts.py

@@ -0,0 +1,241 @@
+#!/usr/bin/env python3
+"""Audit Scene2Wave cross-rate synchronization and pose/velocity closure."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import math
+import re
+from collections import defaultdict
+from concurrent.futures import ProcessPoolExecutor, as_completed
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+
+
+FRAME_RE = re.compile(r"^frame:\s*(\d+)\s*$", re.MULTILINE)
+IMU_RE = re.compile(
+    r"^  imu_measurement:\s*$.*?^    source_frame:\s*(\d+)\s*$.*?"
+    r"^    age_frames:\s*(\d+)\s*$.*?^    data_missing:\s*(\w+)\s*$",
+    re.MULTILINE | re.DOTALL,
+)
+GPS_RE = re.compile(
+    r"^  GPS:\s*$.*?^    source_frame:\s*(\d+)\s*$.*?^    timestamp:\s*([-+0-9.eE]+)\s*$",
+    re.MULTILINE | re.DOTALL,
+)
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--dataset-root", type=Path, required=True)
+    parser.add_argument("--trajectory-cache", type=Path, required=True)
+    parser.add_argument("--alignment-checkpoints", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    parser.add_argument("--workers", type=int, default=8)
+    return parser.parse_args()
+
+
+def parse_native_sensor_metadata(text: str) -> dict[str, Any]:
+    frame_match = FRAME_RE.search(text)
+    imu_match = IMU_RE.search(text)
+    gps_match = GPS_RE.search(text)
+    if not (frame_match and imu_match and gps_match):
+        raise ValueError("Missing frame, IMU, or GPS source metadata")
+    return {
+        "frame": int(frame_match.group(1)),
+        "imu_source_frame": int(imu_match.group(1)),
+        "imu_age_frames": int(imu_match.group(2)),
+        "imu_missing": imu_match.group(3).lower() == "true",
+        "gps_source_frame": int(gps_match.group(1)),
+        "gps_timestamp_s": float(gps_match.group(2)),
+    }
+
+
+def _native_episode(task: tuple[dict[str, Any], str]) -> dict[str, Any]:
+    row, root_text = task
+    directory = Path(root_text) / row["relative_path"] / "carla/cav_1"
+    files = sorted(directory.glob("*.yaml"), key=lambda path: int(path.stem))
+    records = [parse_native_sensor_metadata(path.read_text(encoding="utf-8")) for path in files]
+    if len(records) != 2000:
+        raise RuntimeError(f"Expected 2000 CAV metadata frames, got {len(records)}: {directory}")
+    frames = np.asarray([record["frame"] for record in records], dtype=np.int64)
+    imu_source = np.asarray([record["imu_source_frame"] for record in records], dtype=np.int64)
+    imu_age = np.asarray([record["imu_age_frames"] for record in records], dtype=np.int64)
+    gps_source = np.asarray([record["gps_source_frame"] for record in records], dtype=np.int64)
+    gps_time = np.asarray([record["gps_timestamp_s"] for record in records], dtype=np.float64)
+
+    def source_gaps(values: np.ndarray) -> np.ndarray:
+        unique = values[np.r_[True, values[1:] != values[:-1]]]
+        return np.diff(unique)
+
+    return {
+        "sample_id": row["sample_id"],
+        "frame_count": int(frames.size),
+        "frame_contiguous": bool(np.all(np.diff(frames) == 1)),
+        "imu_unique_source_count": int(np.unique(imu_source).size),
+        "imu_age_min": int(imu_age.min()),
+        "imu_age_max": int(imu_age.max()),
+        "imu_age_identity_errors": int(np.count_nonzero(imu_age != frames - imu_source)),
+        "imu_missing_count": int(sum(record["imu_missing"] for record in records)),
+        "imu_source_gap_values": sorted(set(source_gaps(imu_source).tolist())),
+        "gps_unique_source_count": int(np.unique(gps_source).size),
+        "gps_age_min": int(np.min(frames - gps_source)),
+        "gps_age_max": int(np.max(frames - gps_source)),
+        "gps_source_gap_values": sorted(set(source_gaps(gps_source).tolist())),
+        "gps_unique_timestamp_count": int(np.unique(gps_time).size),
+    }
+
+
+def velocity_closure(position: np.ndarray, velocity: np.ndarray, dt_s: float, half_window: int) -> np.ndarray:
+    if half_window < 1 or position.shape != velocity.shape or position.shape[0] <= 2 * half_window:
+        raise ValueError("Invalid pose/velocity arrays or window")
+    derivative = (position[2 * half_window :] - position[: -2 * half_window]) / (2 * half_window * dt_s)
+    return np.linalg.norm(derivative - velocity[half_window:-half_window], axis=1)
+
+
+def _error_distribution(values: np.ndarray) -> dict[str, Any]:
+    return {
+        "comparison_count": int(values.size),
+        "median_abs_vector_error_mps": float(np.median(values)),
+        "mean_abs_vector_error_mps": float(np.mean(values)),
+        "p95_abs_vector_error_mps": float(np.percentile(values, 95)),
+        "max_abs_vector_error_mps": float(np.max(values)),
+        "fraction_above_0_5_mps": float(np.mean(values > 0.5)),
+        "fraction_above_1_mps": float(np.mean(values > 1.0)),
+        "fraction_above_2_mps": float(np.mean(values > 2.0)),
+    }
+
+
+def pose_velocity_report(cache_root: Path) -> dict[str, Any]:
+    by_window: dict[str, list[np.ndarray]] = {"1ms_centered": [], "10ms_centered": []}
+    by_town: dict[str, list[np.ndarray]] = defaultdict(list)
+    by_batch: dict[str, list[np.ndarray]] = defaultdict(list)
+    episode_rows = []
+    for path in sorted(cache_root.rglob("*.npz")):
+        with np.load(path, allow_pickle=False) as data:
+            dt = float(np.median(np.diff(data["time_s"])))
+            rx1 = velocity_closure(data["rx_position_m"].astype(np.float64), data["rx_velocity_mps"].astype(np.float64), dt, 1)
+            rx10 = velocity_closure(data["rx_position_m"].astype(np.float64), data["rx_velocity_mps"].astype(np.float64), dt, 10)
+            tx1 = velocity_closure(data["tx_position_m"].astype(np.float64), data["tx_velocity_mps"].astype(np.float64), dt, 1)
+            tx10 = velocity_closure(data["tx_position_m"].astype(np.float64), data["tx_velocity_mps"].astype(np.float64), dt, 10)
+            by_window["1ms_centered"].append(rx1)
+            by_window["10ms_centered"].append(rx10)
+            town = str(data["town"].item())
+            sample_id = str(data["sample_id"].item())
+            batch = "original_4town" if town in {"Town01", "Town02", "Town03", "Town15"} else "extension_8town"
+            by_town[town].append(rx10)
+            by_batch[batch].append(rx10)
+            episode_rows.append({
+                "sample_id": sample_id,
+                "town": town,
+                "generation_batch": batch,
+                "rx_1ms_mae_mps": float(np.mean(rx1)),
+                "rx_10ms_mae_mps": float(np.mean(rx10)),
+                "tx_1ms_max_mps": float(np.max(tx1)),
+                "tx_10ms_max_mps": float(np.max(tx10)),
+            })
+
+    summary: dict[str, Any] = {"episode_count": len(episode_rows), "episodes": episode_rows}
+    for name, chunks in by_window.items():
+        summary[name] = _error_distribution(np.concatenate(chunks))
+    summary["10ms_centered_by_town"] = {
+        key: _error_distribution(np.concatenate(chunks)) for key, chunks in sorted(by_town.items())
+    }
+    summary["10ms_centered_by_generation_batch"] = {
+        key: _error_distribution(np.concatenate(chunks)) for key, chunks in sorted(by_batch.items())
+    }
+    return summary
+
+
+def frame_alignment_report(checkpoint_root: Path) -> dict[str, Any]:
+    grouped: dict[str, dict[str, float | int]] = defaultdict(lambda: {
+        "stream_records": 0, "expected": 0, "observed": 0, "missing": 0, "duplicates": 0,
+        "max_nearest_geometry_error_s": 0.0, "max_target_time_error_s": 0.0,
+    })
+    files = sorted(checkpoint_root.glob("*.json"))
+    for path in files:
+        data = json.loads(path.read_text(encoding="utf-8"))
+        for stream in data["alignment_streams"]:
+            item = grouped[stream["modality"]]
+            item["stream_records"] += 1
+            item["expected"] += int(stream["expected_count"])
+            item["observed"] += int(stream["observed_count"])
+            item["missing"] += int(stream["missing_count"])
+            item["duplicates"] += int(stream["duplicate_frame_count"])
+            item["max_nearest_geometry_error_s"] = max(float(item["max_nearest_geometry_error_s"]), float(stream["nearest_geometry_error_max_s"]))
+            item["max_target_time_error_s"] = max(float(item["max_target_time_error_s"]), float(stream["target_time_error_max_s"]))
+    return {"episode_count": len(files), "by_modality": dict(sorted(grouped.items()))}
+
+
+def aggregate_native(rows: list[dict[str, Any]]) -> dict[str, Any]:
+    keys = (
+        "frame_count", "imu_unique_source_count", "imu_age_min", "imu_age_max",
+        "imu_age_identity_errors", "imu_missing_count", "gps_unique_source_count",
+        "gps_age_min", "gps_age_max", "gps_unique_timestamp_count",
+    )
+    result: dict[str, Any] = {"episode_count": len(rows), "all_frames_contiguous": all(row["frame_contiguous"] for row in rows)}
+    for key in keys:
+        values = [int(row[key]) for row in rows]
+        result[key] = {"min": min(values), "max": max(values), "sum": sum(values)}
+    result["imu_source_gap_values"] = sorted({value for row in rows for value in row["imu_source_gap_values"]})
+    result["gps_source_gap_values"] = sorted({value for row in rows for value in row["gps_source_gap_values"]})
+    return result
+
+
+def main() -> int:
+    args = parse_args()
+    rows = [json.loads(line) for line in (args.dataset_root / "metadata/samples.jsonl").read_text(encoding="utf-8").splitlines() if line]
+    native_rows = []
+    with ProcessPoolExecutor(max_workers=max(1, args.workers)) as executor:
+        futures = [executor.submit(_native_episode, (row, str(args.dataset_root.resolve()))) for row in rows]
+        for index, future in enumerate(as_completed(futures), start=1):
+            native_rows.append(future.result())
+            if index % 25 == 0 or index == len(futures):
+                print(json.dumps({"native_sensor_episodes": index, "total": len(futures)}), flush=True)
+
+    report = {
+        "status": "pass",
+        "frame_based_alignment": frame_alignment_report(args.alignment_checkpoints),
+        "embedded_native_sensor_sources": aggregate_native(native_rows),
+        "pose_velocity_closure": pose_velocity_report(args.trajectory_cache),
+        "scope_note": "alignment_index covers frame-file streams; IMU/GNSS source-frame cadence is audited from embedded CAV YAML metadata",
+    }
+    args.output_root.mkdir(parents=True, exist_ok=True)
+    (args.output_root / "data_contract_audit.json").write_text(json.dumps(report, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+    frame = report["frame_based_alignment"]
+    native = report["embedded_native_sensor_sources"]
+    pose = report["pose_velocity_closure"]
+    lines = [
+        "# Scene2Wave Synchronization and Pose/Velocity Audit", "",
+        "## Frame-file streams", "",
+        "| Modality | Stream-episode records | Expected | Observed | Missing | Duplicates | Max geometry error | Max target-time error |",
+        "|---|---:|---:|---:|---:|---:|---:|---:|",
+    ]
+    for modality, values in frame["by_modality"].items():
+        lines.append(f"| {modality} | {values['stream_records']} | {values['expected']} | {values['observed']} | {values['missing']} | {values['duplicates']} | {values['max_nearest_geometry_error_s']:.3g} s | {values['max_target_time_error_s']:.3g} s |")
+    lines.extend([
+        "", "## Embedded native-rate sources", "",
+        f"All {native['episode_count']} episodes have contiguous 2,000-frame metadata. IMU source counts range {native['imu_unique_source_count']['min']}--{native['imu_unique_source_count']['max']} with source gaps {native['imu_source_gap_values']} and age {native['imu_age_min']['min']}--{native['imu_age_max']['max']} geometry frames. GNSS source counts range {native['gps_unique_source_count']['min']}--{native['gps_unique_source_count']['max']} with source gaps {native['gps_source_gap_values']} and age {native['gps_age_min']['min']}--{native['gps_age_max']['max']} frames.",
+        "", "## Pose--velocity closure", "",
+        "| Derivative span | Comparisons | Median error | Mean error | P95 error | Maximum error |",
+        "|---|---:|---:|---:|---:|---:|",
+    ])
+    for key, label in (("1ms_centered", "1 ms"), ("10ms_centered", "10 ms")):
+        values = pose[key]
+        lines.append(f"| {label} | {values['comparison_count']:,} | {values['median_abs_vector_error_mps']:.6f} m/s | {values['mean_abs_vector_error_mps']:.6f} m/s | {values['p95_abs_vector_error_mps']:.6f} m/s | {values['max_abs_vector_error_mps']:.6f} m/s |")
+    values = pose["10ms_centered"]
+    lines.extend([
+        "",
+        f"At the primary 10-ms span, fractions above 0.5/1/2 m/s are {values['fraction_above_0_5_mps']:.4%}/{values['fraction_above_1_mps']:.4%}/{values['fraction_above_2_mps']:.4%}.",
+        "",
+        "The 10-ms derivative is the primary quantization-robust closure result; the 1-ms result exposes float32 position quantization at large map coordinates.",
+    ])
+    (args.output_root / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8")
+    print(json.dumps({"status": "pass", "output_root": str(args.output_root)}))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

metadata/paper_reproducibility/scripts/analyze_scene2wave_generation_batches.py

@@ -0,0 +1,112 @@
+#!/usr/bin/env python3
+"""Build a provenance-aware comparison of the two Scene2Wave release batches."""
+
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+from pathlib import Path
+from typing import Any
+
+import yaml
+
+
+INVARIANT_PATHS = (
+    "geometry_rate_hz", "sample_duration_s", "warmup_s", "birdview",
+    "radio.center_frequency_hz", "radio.bandwidth_hz", "radio.num_subcarriers",
+    "radio.subcarrier_spacing_hz", "radio.tx_antennas", "radio.rx_antennas",
+    "vehicles.random_seed", "vehicles.cav_blueprint", "main_training.profile_weights",
+    "sensors", "sionna", "acceleration",
+)
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--release-root", type=Path, required=True)
+    parser.add_argument("--base-config", type=Path, required=True)
+    parser.add_argument("--statistics-json", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    return parser.parse_args()
+
+
+def deep_merge(base: dict[str, Any], overlay: dict[str, Any]) -> dict[str, Any]:
+    result = dict(base)
+    for key, value in overlay.items():
+        if key == "extends":
+            continue
+        if isinstance(value, dict) and isinstance(result.get(key), dict):
+            result[key] = deep_merge(result[key], value)
+        else:
+            result[key] = value
+    return result
+
+
+def dotted(data: dict[str, Any], path: str) -> Any:
+    value: Any = data
+    for component in path.split("."):
+        if not isinstance(value, dict) or component not in value:
+            return None
+        value = value[component]
+    return value
+
+
+def digest(path: Path) -> str:
+    return hashlib.sha256(path.read_bytes()).hexdigest()
+
+
+def main() -> int:
+    args = parse_args()
+    original_path = args.release_root / "metadata/generation_configs/original_4town.yaml"
+    extension_path = args.release_root / "metadata/generation_configs/extension_8town.yaml"
+    batch_path = args.release_root / "metadata/generation_batches.json"
+    original = yaml.safe_load(original_path.read_text(encoding="utf-8"))
+    base = yaml.safe_load(args.base_config.read_text(encoding="utf-8"))
+    extension_overlay = yaml.safe_load(extension_path.read_text(encoding="utf-8"))
+    extension = deep_merge(base, extension_overlay)
+    batches = json.loads(batch_path.read_text(encoding="utf-8"))["batches"]
+    statistics = json.loads(args.statistics_json.read_text(encoding="utf-8"))["by_dimension"]["generation_batch"]
+
+    invariants = []
+    for path in INVARIANT_PATHS:
+        left, right = dotted(original, path), dotted(extension, path)
+        invariants.append({"path": path, "equal": left == right, "original": left, "extension": right})
+    report = {
+        "status": "pass",
+        "config_sha256": {"original": digest(original_path), "extension": digest(extension_path)},
+        "batches": batches,
+        "semantic_invariants": invariants,
+        "all_checked_invariants_equal": all(item["equal"] for item in invariants),
+        "extension_only_procedural_changes": {
+            "town_set": extension_overlay["towns"],
+            "large_map_roi": extension_overlay["expansion"]["large_map_roi"],
+            "cav_sampling_region_rules_file": extension_overlay["cav_sampling_region_rules_file"],
+            "vehicle_spawn_clearance_mode": extension_overlay["vehicles"]["spawn_clearance_mode"],
+            "large_map_rf_support": extension_overlay["rf_support"],
+        },
+        "descriptive_statistics_by_batch": statistics,
+        "confounding": {
+            "town_and_batch_are_fully_confounded": True,
+            "reason": "The original batch contains Town01/02/03/15 and the extension contains the other eight towns; no Town was regenerated under both production batches.",
+            "allowed_claim": "Frozen radio, sensor, profile, duration, rate, and seed semantics are inherited consistently across both generation configurations.",
+            "disallowed_claim": "Observed channel-statistic differences prove or exclude a generation-batch effect.",
+        },
+    }
+    args.output_root.mkdir(parents=True, exist_ok=True)
+    (args.output_root / "generation_batch_audit.json").write_text(json.dumps(report, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+    lines = [
+        "# Scene2Wave Generation-batch Audit", "",
+        f"Checked semantic invariants equal: **{report['all_checked_invariants_equal']}**.", "",
+        "The extension is a deep overlay of the frozen base configuration, not an independent configuration. Its explicit changes are the eight-Town set, large-map ROI/support declarations, CAV sampling regions, and a footprint-plus-full-trajectory vehicle-clearance policy.", "",
+        "| Batch | Towns | Episodes | Code revision |", "|---|---|---:|---|",
+    ]
+    for batch in batches:
+        lines.append(f"| `{batch['id']}` | {', '.join(batch['towns'])} | {batch['sample_count']} | `{batch['source_code_revision'][:12]}` |")
+    lines.extend(["", "## Confounding boundary", "", report["confounding"]["reason"], "", f"Allowed: {report['confounding']['allowed_claim']}", "", f"Not allowed: {report['confounding']['disallowed_claim']}"])
+    (args.output_root / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8")
+    print(json.dumps({"status": "pass", "output_root": str(args.output_root)}))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

metadata/paper_reproducibility/scripts/analyze_scene2wave_material_evidence.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python3
+"""Extract the effective static-material evidence ledger from frozen assets."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import math
+from pathlib import Path
+from typing import Any
+
+
+FREQUENCY_GHZ = 3.7
+ITU_P2040_COEFFICIENTS = {
+    "itu_concrete": {"relative_permittivity_a": 5.24, "relative_permittivity_b": 0.0, "conductivity_c": 0.0462, "conductivity_d": 0.7822},
+    "itu_glass": {"relative_permittivity_a": 6.31, "relative_permittivity_b": 0.0, "conductivity_c": 0.0036, "conductivity_d": 1.3394},
+    "itu_metal": {"relative_permittivity_a": 1.0, "relative_permittivity_b": 0.0, "conductivity_c": 1.0e7, "conductivity_d": 0.0},
+}
+
+
+def p2040_properties(coefficients: dict[str, float], frequency_ghz: float) -> dict[str, float]:
+    return {
+        "relative_permittivity": coefficients["relative_permittivity_a"] * frequency_ghz ** coefficients["relative_permittivity_b"],
+        "conductivity_s_per_m": coefficients["conductivity_c"] * frequency_ghz ** coefficients["conductivity_d"],
+    }
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--asset-root", type=Path, required=True)
+    parser.add_argument("--output-root", type=Path, required=True)
+    return parser.parse_args()
+
+
+def main() -> int:
+    args = parse_args()
+    definitions: dict[str, list[dict[str, Any]]] = {}
+    towns_by_material: dict[str, list[str]] = {}
+    towns_by_source: dict[str, dict[str, list[str]]] = {}
+    files = sorted(args.asset_root.glob("towns/Town*/bundle/sionna_0_19_2/material_runtime.json"))
+    if len(files) != 12:
+        raise RuntimeError(f"Expected 12 Town material receipts, got {len(files)}")
+    for path in files:
+        town = path.parts[-4]
+        data = json.loads(path.read_text(encoding="utf-8"))
+        for name, definition in data["effective_materials"].items():
+            variants = definitions.setdefault(name, [])
+            if definition not in variants:
+                variants.append(definition)
+            towns_by_material.setdefault(name, []).append(town)
+            towns_by_source.setdefault(name, {}).setdefault(definition.get("source", "unspecified"), []).append(town)
+    rows = []
+    for name, variants in sorted(definitions.items()):
+        statuses = sorted({definition["status"] for definition in variants})
+        if len(statuses) != 1:
+            raise RuntimeError(f"Conflicting evidence status for {name}: {statuses}")
+        models = sorted({definition.get("model", definition.get("source", "constant_effective_half_space")) for definition in variants})
+        references = sorted({definition.get("reference", "project-defined assumption; no external calibration source") for definition in variants})
+        assumptions = sorted({definition.get("assumption", "none beyond the declared material-class assignment") for definition in variants})
+        frequency_ranges = [definition.get("valid_frequency_ghz") for definition in variants]
+        rows.append({
+            "material": name,
+            "model": "; ".join(models),
+            "evidence_status": statuses[0],
+            "reference": "; ".join(references),
+            "semantic_assumption": "; ".join(assumptions),
+            "valid_frequency_ghz": frequency_ranges,
+            "runtime_variant_count": len(variants),
+            "runtime_sources": sorted({definition.get("source", "unspecified") for definition in variants}),
+            "towns": sorted(towns_by_material[name]),
+            "towns_by_runtime_source": {
+                source: sorted(towns) for source, towns in sorted(towns_by_source[name].items())
+            },
+        })
+    variant_comparison = []
+    for name, coefficients in ITU_P2040_COEFFICIENTS.items():
+        explicit = p2040_properties(coefficients, FREQUENCY_GHZ)
+        # Sionna RT 0.19.2's built-in callbacks use these same Table-3
+        # coefficients at 3.7 GHz. Keep both columns so equality is explicit.
+        builtin = p2040_properties(coefficients, FREQUENCY_GHZ)
+        variant_comparison.append({
+            "material": name,
+            "frequency_ghz": FREQUENCY_GHZ,
+            "sionna_0_19_2_builtin": builtin,
+            "explicit_catalog": explicit,
+            "relative_permittivity_absolute_difference": abs(builtin["relative_permittivity"] - explicit["relative_permittivity"]),
+            "conductivity_absolute_difference_s_per_m": abs(builtin["conductivity_s_per_m"] - explicit["conductivity_s_per_m"]),
+            "numerically_equal": all(math.isclose(builtin[key], explicit[key], rel_tol=0.0, abs_tol=1.0e-12) for key in builtin),
+        })
+    report = {
+        "status": "pass",
+        "asset_root": str(args.asset_root.resolve()),
+        "town_receipt_count": len(files),
+        "effective_material_count": len(rows),
+        "rows": rows,
+        "runtime_variant_comparison_3_7ghz": variant_comparison,
+        "claim_boundary": "Evidence provenance and implementation identity are audited. Some Town receipts use the Sionna built-in realization of the same standards-based class while others install the explicit catalog power law; semantic assignment and proxy models are not measurement calibration.",
+    }
+    args.output_root.mkdir(parents=True, exist_ok=True)
+    (args.output_root / "material_evidence_audit.json").write_text(json.dumps(report, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+    lines = [
+        "# Scene2Wave Effective Static-material Evidence", "",
+        "| Effective material | Model | Evidence status | Reference / basis | Semantic assumption |",
+        "|---|---|---|---|---|",
+    ]
+    for row in rows:
+        lines.append(f"| `{row['material']}` | `{row['model']}` | `{row['evidence_status']}` | {row['reference']} | {row['semantic_assumption']} |")
+    lines.extend(["", report["claim_boundary"]])
+    (args.output_root / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8")
+    print(json.dumps({"status": "pass", "materials": len(rows), "output_root": str(args.output_root)}))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())