Add deterministic benchmark splits

.gitattributes

@@ -7707,3 +7707,4 @@ run_377/volume_377.vtu filter=lfs diff=lfs merge=lfs -text
 *.vtp filter=lfs diff=lfs merge=lfs -text
 *.vtu filter=lfs diff=lfs merge=lfs -text
 *.stl filter=lfs diff=lfs merge=lfs -text
+splits/README.pdf filter=lfs diff=lfs merge=lfs -text

README.md

@@ -52,6 +52,43 @@ In addition we provide:
 * stl : folder containing stl files that were used as inputs to the OpenFOAM process
 * openfoam-casesetup.tgz : complete OpenFOAM setup that can be used to extend or reproduce the dataset
 * validation : folder containing full outputs from all four mesh levels that were used to validate the methodology
+* [`splits/`](splits/): deterministic benchmark manifests, methods documentation, derived metrics, diagnostic figures, and generation code
+
+## Recommended dataset splits
+
+AhmedML provides eight deterministic train/validation/test split families in
+[`splits/manifest.json`](splits/manifest.json). Identifiers correspond to the
+top-level `run_N` directories.
+
+| Split | Type | Train | Validation | Test | Intended evaluation |
+|---|---:|---:|---:|---:|---|
+| `full` | In-distribution | 400 | 50 | 50 | Seed-42 public baseline |
+| `medium` | In-distribution | 133 | 50 | 50 | Intermediate data efficiency |
+| `scarce` | In-distribution | 67 | 50 | 50 | Low-data evaluation |
+| `super_scarce` | In-distribution | 11 | 50 | 50 | Extreme low-data evaluation |
+| `geometry` | OOD | 350 | 50 | 100 | STL-surface geometry extrapolation |
+| `high_drag` | OOD | 350 | 50 | 100 | High-drag extrapolation |
+| `low_drag` | OOD | 350 | 50 | 100 | Low-drag extrapolation |
+| `image_wake` | OOD | 350 | 50 | 100 | Image-derived wake extrapolation |
+
+The `full` assignment follows the established seed-42 AhmedML baseline used by
+Noether. The reduced-data training sets are strictly nested and share the same
+validation and test cases. For the OOD families, validation is selected from
+the training-side population, while the held-out extreme is reserved for final
+testing.
+
+Download only the split package with:
+
+```bash
+hf download neashton/ahmedml \
+  --type dataset \
+  --include "splits/**" \
+  --local-dir ./ahmedml
+```
+
+Complete definitions, construction methods, missing-data treatment,
+diagnostic figures, and reproducibility instructions are provided in
+[`splits/README.md`](splits/README.md).
 
 Downloads
 --------------
@@ -113,6 +150,7 @@ This dataset is provided under the CC BY SA 4.0 license, please see LICENSE.txt
 
 version history:
 ---------------
+* 17/08/2026 - Added deterministic benchmark train/validation/test splits, including nested data-efficiency and out-of-distribution evaluation protocols.
 * 15/02/2025 - files uploaded to HuggingFace
 * 12/11/2024 - added validation folder that contains the full output from all four mesh levels that were used to validate the methodology used.
 * 04/08/2024 - updates to the file description and arxiv paper

splits/README.md

@@ -0,0 +1,216 @@
+# AhmedML dataset splits
+
+This directory provides deterministic train/validation/test assignments for the
+[AhmedML](https://huggingface.co/datasets/neashton/ahmedml) dataset. The
+authoritative assignments are stored in [`manifest.json`](manifest.json) as a
+flat JSON object. Keys follow the pattern `{split_name}_{train,val,test}`, and
+each value is a numerically sorted list of identifiers matching the top-level
+`run_N` directories.
+
+AhmedML contains 500 Ahmed-body variants, indexed from `run_1` to `run_500`.
+The aggregate force and moment table contains all 500 runs. The aggregate
+geometry-parameter table contains 499 rows because `run_500` is absent; the
+treatment of this missing row is described below.
+
+## Splits at a glance
+
+| Split | Type | Train | Validation | Test | Intended evaluation |
+|---|---:|---:|---:|---:|---|
+| `full` | In-distribution | 400 | 50 | 50 | Seed-42 public baseline |
+| `medium` | In-distribution | 133 | 50 | 50 | Intermediate data efficiency |
+| `scarce` | In-distribution | 67 | 50 | 50 | Low-data evaluation |
+| `super_scarce` | In-distribution | 11 | 50 | 50 | Extreme low-data evaluation |
+| `geometry` | OOD | 350 | 50 | 100 | STL-surface geometry extrapolation |
+| `high_drag` | OOD | 350 | 50 | 100 | High-drag extrapolation |
+| `low_drag` | OOD | 350 | 50 | 100 | Low-drag extrapolation |
+| `image_wake` | OOD | 350 | 50 | 100 | Image-derived wake extrapolation |
+
+The data-efficiency training sets form a strict nested sequence:
+
+`super_scarce_train ⊂ scarce_train ⊂ medium_train ⊂ full_train`
+
+They use the same validation and test assignments. For every
+out-of-distribution (OOD) family, validation is sampled from the training-side
+population; the held-out extreme is reserved for final testing.
+
+## Using the committed manifest
+
+Normal benchmark use requires only the committed manifest. Regenerating the
+splits is not required.
+
+```python
+import json
+from pathlib import Path
+
+manifest = json.loads(Path("splits/manifest.json").read_text())
+
+train_ids = manifest["geometry_train"]
+val_ids = manifest["geometry_val"]
+test_ids = manifest["geometry_test"]
+```
+
+Change the `geometry` prefix to `full`, `medium`, `scarce`,
+`super_scarce`, `high_drag`, `low_drag`, or `image_wake` to select another
+family.
+
+Download only the split package with:
+
+```bash
+hf download neashton/ahmedml \
+  --type dataset \
+  --include "splits/**" \
+  --local-dir ./ahmedml
+```
+
+Validation data may be used for model and hyperparameter selection. Test data
+should be reserved for final evaluation and should not inform normalization,
+feature design, or repeated visual inspection during development.
+
+## Construction principles
+
+1. **Stable public baseline.** The `full` family preserves the established
+   seed-42 AhmedML assignment used by
+   [Noether](https://github.com/Emmi-AI/noether/blob/main/src/noether/data/datasets/cfd/caeml/ahmedml/split.py).
+2. **In-distribution validation.** OOD validation cases are selected from the
+   training-side population rather than the extreme test region.
+3. **Nested data-efficiency subsets.** Smaller training sets are strict
+   subsets of larger sets, with validation and test held fixed.
+4. **Direct geometry comparison.** The geometry OOD score is computed from STL
+   surfaces rather than inferred only from geometry parameters.
+5. **Dataset-defined physical quantities.** Drag families use the published
+   constant-reference-area force table.
+6. **Flow-structure information.** The image-wake family uses fixed `UxMean`
+   views rather than an integrated coefficient.
+7. **Auditability.** The manifest is distributed with the derived metrics,
+   scripts, figures, LaTeX source, and PDF methods report used to document it.
+
+## Split definitions
+
+### `full`
+
+The baseline is a seeded random partition over all 500 run identifiers. It
+constructs `torch.randperm(500)` with seed 42, maps the result to identifiers
+`1..500`, assigns 400 cases to training, 50 to validation, and 50 to testing,
+then sorts each stored list numerically. The committed identifiers match
+Noether's public `AhmedMLDefaultSplitIDs` assignment.
+
+### `medium`, `scarce`, and `super_scarce`
+
+These families retain `full_val` and `full_test` while reducing the training
+population to 133, 67, and 11 cases. A greedy max-min procedure constructs one
+nested ordering in standardized force/geometry feature space using `cd`, `cl`,
+and the available aggregate geometry parameters.
+
+`geo_parameters_all.csv` has no row for `run_500`. For this ordering only, its
+geometry features are set to the corresponding column means. The treatment is
+recorded in [`parameter_geometry_metrics.csv`](parameter_geometry_metrics.csv).
+The force, geometry OOD, and image-wake definitions do not use this imputation.
+
+### `geometry`
+
+The geometry family uses [`chamfer_metrics.csv`](chamfer_metrics.csv). Each
+`run_N/ahmed_N.stl` surface is sampled with 4,096 deterministic area-weighted
+points. Point clouds remain in the shared dataset coordinate frame and are
+scaled by the global median STL bounding-box diagonal. Pairwise surface
+difference is measured with symmetric Chamfer RMS distance.
+
+For each run, the OOD score is the mean distance to its ten nearest neighbouring
+geometries. The 100 highest-scoring cases form `geometry_test`; 50 validation
+cases are deterministically selected from the complementary population, leaving
+350 training cases.
+
+### `high_drag` and `low_drag`
+
+These families rank all cases by `cd` from the root-level `force_mom_all.csv`,
+which uses a constant reference area. `high_drag` holds out the largest 100
+values, while `low_drag` holds out the smallest 100. Validation is sampled from
+the complementary population in both cases.
+
+### `image_wake`
+
+The image-wake score uses four published `UxMean` PNGs for each run: the `Y-4`
+centreline or near-centreline image and the near-base `X-14`, `X-15`, and `X-16`
+cross-plane images. A fixed lower-flow-region colour/intensity measure is
+computed for each image and combined across the centreline and near-base views.
+The 100 largest scores form `image_wake_test`. All 500 committed scores are
+direct observations.
+
+![Split diagnostics](split_diagnostics.png)
+
+![Geometry-score examples](geometry_score_examples.png)
+
+![Image-wake-score examples](wake_score_examples.png)
+
+## Reproducibility
+
+The committed [`manifest.json`](manifest.json) is the source of truth. The
+commands below are provided to audit or rebuild the artifacts. They were
+prepared against AhmedML revision
+`29135130ea70a842ad84ae2c90248d2ba70a7a69`.
+
+Install the lightweight generation and plotting dependencies:
+
+```bash
+python3 -m pip install numpy matplotlib pillow
+```
+
+From the dataset repository root, download the aggregate source tables and
+regenerate the manifest and diagnostic plot:
+
+```bash
+python3 splits/download_hf_inputs.py --output-dir data
+python3 splits/generate_splits.py
+python3 splits/visualize_splits.py
+```
+
+The commands above use the committed Chamfer and image metrics. To recompute
+those metrics and recreate the example figures, keep large STL and PNG inputs
+outside the repository:
+
+```bash
+ASSET_ROOT=../ahmedml_hf_assets
+
+python3 splits/download_hf_inputs.py --output-dir "$ASSET_ROOT" \
+  --include-stls --include-wake-images --workers 6
+
+python3 splits/compute_chamfer_splits.py --data-root "$ASSET_ROOT" \
+  --output-dir /tmp/ahmedml_chamfer_4096 --samples 4096 \
+  --workers 16 --sample-workers 2 --runs all \
+  --base-manifest splits/manifest.json
+
+cp /tmp/ahmedml_chamfer_4096/chamfer_metrics.csv \
+  splits/chamfer_metrics.csv
+
+python3 splits/compute_image_metrics.py --data-root "$ASSET_ROOT" \
+  --output splits/image_metrics.csv
+
+AHMEDML_DATA_ROOT="$ASSET_ROOT" python3 splits/generate_splits.py
+AHMEDML_DATA_ROOT="$ASSET_ROOT" python3 splits/visualize_splits.py
+python3 splits/create_example_figures.py --asset-root "$ASSET_ROOT" \
+  --force-root "$ASSET_ROOT"
+```
+
+Full Chamfer recomputation additionally requires SciPy and trimesh:
+
+```bash
+python3 -m pip install scipy trimesh
+```
+
+Rebuild the PDF methods report with:
+
+```bash
+latexmk -pdf -cd splits/README.tex
+```
+
+The generated manifest remains a flat mapping such as:
+
+```json
+{
+  "full_train": ["run_1", "run_2"],
+  "full_val": ["run_24"],
+  "full_test": ["run_4"]
+}
+```
+
+The shortened lists above illustrate the format only; use the committed
+manifest for the complete assignments.

splits/README.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c76c2a37fff6118d81a1310da4a24fc751d0ed6a5c5a688d615d219933d39721
+size 1268353

splits/README.tex

@@ -0,0 +1,317 @@
+\documentclass[10pt]{article}
+
+\usepackage[margin=0.72in]{geometry}
+\usepackage{booktabs}
+\usepackage{caption}
+\usepackage{enumitem}
+\usepackage{float}
+\usepackage[T1]{fontenc}
+\usepackage{graphicx}
+\usepackage{hyperref}
+\usepackage{microtype}
+\usepackage{tabularx}
+\usepackage{xcolor}
+
+\hypersetup{
+  colorlinks=true,
+  linkcolor=blue!55!black,
+  urlcolor=blue!55!black,
+  citecolor=blue!55!black
+}
+
+\setlength{\parindent}{0pt}
+\setlength{\parskip}{0.55em}
+\setlength{\emergencystretch}{2em}
+\setlist[itemize]{leftmargin=1.35em, itemsep=0.22em, topsep=0.25em}
+\captionsetup{font=small, labelfont=bf}
+
+\newcommand{\code}[1]{\texttt{#1}}
+\newcommand{\splitkey}[1]{\texttt{#1}}
+
+\title{\vspace{-1.2em}\textbf{AhmedML Dataset Splits}}
+\author{}
+\date{}
+
+\begin{document}
+\maketitle
+\vspace{-2.0em}
+
+Deterministic train/validation/test splits for the
+\href{https://huggingface.co/datasets/neashton/ahmedml}{AhmedML} dataset
+\cite{ahmedml_dataset}. The split manifest is stored at
+\code{splits/manifest.json} as a flat JSON object with keys named
+\code{\{split\_name\}\_\{train,val,test\}} and case IDs that match the
+on-disk run directories.
+
+AhmedML contains 500 geometric variations of the Ahmed car body. The public
+force/moment table contains all 500 runs. This split package intentionally does
+not store the large per-run STL or PNG assets. The committed lightweight
+artifacts are the manifest, derived Chamfer/image metrics, scripts, and
+diagnostic figures; the aggregate source tables remain at the dataset root.
+The geometry split is generated from STL-surface Chamfer scores in
+\code{splits/chamfer\_metrics.csv}. The image-wake split is generated
+from \code{UxMean} Y-4 centreline and near-base X-slice PNG scores in
+\code{splits/image\_metrics.csv}. The geometry-parameter table contains 499 rows;
+run 500 is absent from that file, so the generator mean-imputes its geometry
+parameters only for the nested data-efficiency ordering and records this in
+\code{splits/parameter\_geometry\_metrics.csv}.
+
+\section*{Splits at a glance}
+
+\begin{tabularx}{\textwidth}{@{}l l r r r X@{}}
+\toprule
+Split & Type & Train & Val & Test & What it tests \\
+\midrule
+\splitkey{full} & In-dist & 400 & 50 & 50 & Noether-compatible seed-42 random baseline split \\
+\splitkey{medium} & In-dist & 133 & 50 & 50 & Data efficiency, 1/3 of \splitkey{full} training data \\
+\splitkey{scarce} & In-dist & 67 & 50 & 50 & Data efficiency, 1/6 of \splitkey{full} training data \\
+\splitkey{super\_scarce} & In-dist & 11 & 50 & 50 & Extreme data efficiency, 1/36 of \splitkey{full} training data \\
+\splitkey{geometry} & OOD & 350 & 50 & 100 & STL-shape extrapolation using the top 20\% local Chamfer-isolation score \\
+\splitkey{high\_drag} & OOD & 350 & 50 & 100 & High-drag extrapolation using the top 20\% \code{cd} \\
+\splitkey{low\_drag} & OOD & 350 & 50 & 100 & Low-drag extrapolation using the bottom 20\% \code{cd} \\
+\splitkey{image\_wake} & OOD & 350 & 50 & 100 & Image-derived wake extrapolation using the top 20\% \code{UxMean} Y-4 plus near-base X-slice score \\
+\bottomrule
+\end{tabularx}
+
+\textbf{Difficulty ladders:}
+\begin{itemize}
+  \item Data efficiency: \splitkey{full} < \splitkey{medium} < \splitkey{scarce} < \splitkey{super\_scarce}, with fixed validation/test sets.
+  \item OOD regimes: \splitkey{geometry}, \splitkey{high\_drag}, \splitkey{low\_drag}, and \splitkey{image\_wake}.
+\end{itemize}
+
+\section*{Which split should I use?}
+
+\begin{itemize}
+  \item \textbf{Simple baseline or literature compatibility}: \splitkey{full}
+  \item \textbf{Data efficiency study}: compare \splitkey{super\_scarce}, \splitkey{scarce}, \splitkey{medium}, and \splitkey{full}
+  \item \textbf{Geometry extrapolation}: \splitkey{geometry}
+  \item \textbf{Targeted coefficient extrapolation}: \splitkey{high\_drag} or \splitkey{low\_drag}
+  \item \textbf{Image-derived wake extrapolation}: \splitkey{image\_wake}
+\end{itemize}
+
+\section*{Using the committed splits}
+
+For standard benchmark use, consume the committed manifest at
+\code{splits/manifest.json}; no regeneration is required. The manifest is the
+source of truth for all split membership and contains one train, validation,
+and test key for each split listed above.
+
+\begin{verbatim}
+import json
+from pathlib import Path
+
+manifest = json.loads(Path("splits/manifest.json").read_text())
+
+train_ids = manifest["geometry_train"]
+val_ids = manifest["geometry_val"]
+test_ids = manifest["geometry_test"]
+\end{verbatim}
+
+Change the \code{geometry} prefix to \code{full}, \code{medium},
+\code{scarce}, \code{super\_scarce}, \code{high\_drag}, \code{low\_drag}, or
+\code{image\_wake} to select a different split. Each value is a sorted list of
+\code{run\_N} directory names that match the dataset case folders.
+
+\section*{Design principles}
+
+\begin{itemize}
+  \item \textbf{Validation is always in-distribution with train.} For OOD splits, validation is drawn from the training-side population, not from the held-out extreme test region.
+  \item \textbf{Keep the public baseline stable.} \splitkey{full} matches Noether's AhmedML seed-42 split: 400 train, 50 validation, and 50 test.
+  \item \textbf{Use the source data for each regime.} Drag-regime splits use \code{force\_mom\_all.csv}; geometry-regime splits use externally downloaded STL files and \code{splits/chamfer\_metrics.csv}; image-regime splits use externally downloaded \code{UxMean} PNG files and \code{splits/image\_metrics.csv}.
+  \item \textbf{Nested data-efficiency subsets.} \splitkey{super\_scarce\_train} is a strict subset of \splitkey{scarce\_train}, \splitkey{scarce\_train} is a strict subset of \splitkey{medium\_train}, and \splitkey{medium\_train} is a strict subset of \splitkey{full\_train}.
+  \item \textbf{Test set integrity.} Test cases should not be used for hyperparameter tuning, model selection, normalization fitting, or visual inspection-driven iteration.
+\end{itemize}
+
+\section*{Split details}
+
+\subsection*{\splitkey{full}}
+
+The default public split is a seeded random split over all 500 run IDs, not a
+physics-stratified split. It constructs \code{torch.randperm(500)} with seed
+42, shifts the IDs to \code{1..500}, then assigns the first 400 IDs to train,
+the next 50 to validation, and the final 50 to test. The case IDs are sorted
+before being stored in the manifest. These IDs match the public
+\href{https://github.com/Emmi-AI/noether/blob/main/src/noether/data/datasets/cfd/caeml/ahmedml/split.py}{AhmedMLDefaultSplitIDs}
+implementation in Noether \cite{noether_split}.
+
+\subsection*{\splitkey{medium}, \splitkey{scarce}, and \splitkey{super\_scarce}}
+
+Same validation/test as \splitkey{full}, but the training set is a nested
+force- and geometry-diverse subset of \splitkey{full\_train}: 133 cases for
+\splitkey{medium}, 67 cases for \splitkey{scarce}, and 11 cases for
+\splitkey{super\_scarce}. The subset order is a greedy max-min selection in
+standardized force/geometry feature space using \code{cd}, \code{cl}, and all
+available geometry parameters.
+
+\subsection*{\splitkey{geometry}}
+
+The geometry split is an STL-surface OOD split. Each \code{run\_N/ahmed\_N.stl}
+is sampled with 4096 deterministic area-weighted surface points. The sampled
+point clouds are compared using symmetric Chamfer RMS distance after scaling by
+the global median STL bounding-box diagonal. The geometry score is the mean
+Chamfer distance to the 10 nearest neighboring runs:
+
+\begin{verbatim}
+geometry_score_i = mean_10_nearest_neighbors(chamfer_distance_i)
+\end{verbatim}
+
+The top 20\% of runs by this local-isolation score form the OOD test set.
+Validation is sampled from the remaining training-side pool:
+
+\begin{verbatim}
+geometry_test  = top_20_percent_i(geometry_score_i)
+geometry_pool  = all_runs - geometry_test
+geometry_val   = deterministic_random_sample(geometry_pool, round(0.125 * |geometry_pool|))
+geometry_train = geometry_pool - geometry_val
+\end{verbatim}
+
+In the current metrics, the most locally isolated STL cases include
+\splitkey{run\_52}, \splitkey{run\_153}, \splitkey{run\_244},
+\splitkey{run\_480}, \splitkey{run\_410}, \splitkey{run\_96},
+\splitkey{run\_431}, and \splitkey{run\_91}. Figure~\ref{fig:diagnostics}
+shows the geometry split against the Chamfer score used to define the holdout.
+Figure~\ref{fig:geometry_examples} shows a low-score and high-score geometry
+example, with both STLs projected in the same coordinate frame.
+
+\begin{figure}[H]
+  \centering
+  \includegraphics[width=0.96\textwidth]{geometry_score_examples.png}
+  \caption{Geometry-score examples for \splitkey{geometry}. The low-score case is \splitkey{run\_71} with Chamfer score 0.010996, \code{Cd}=0.2846, and \code{Cl}=0.4048. The high-score case is \splitkey{run\_52} with Chamfer score 0.018450, \code{Cd}=0.4319, and \code{Cl}=0.0288. The lower panels overlay the low-score and high-score STL vertex projections in a common coordinate frame.}
+  \label{fig:geometry_examples}
+\end{figure}
+
+\subsection*{\splitkey{high\_drag} and \splitkey{low\_drag}}
+
+Both splits sort all 500 cases by drag coefficient \code{cd}.
+\splitkey{high\_drag} holds out the top 20\% by \code{cd};
+\splitkey{low\_drag} holds out the bottom 20\% by \code{cd}. In both cases,
+train/validation are sampled from the complementary 80\% so validation remains
+in-distribution with training. Figure~\ref{fig:diagnostics} shows these
+holdouts across run IDs.
+
+In the current CSV, the highest-drag runs include \splitkey{run\_362},
+\splitkey{run\_205}, \splitkey{run\_284}, \splitkey{run\_280}, and
+\splitkey{run\_315}. The lowest-drag runs include \splitkey{run\_390},
+\splitkey{run\_238}, \splitkey{run\_122}, \splitkey{run\_117}, and
+\splitkey{run\_10}.
+
+\subsection*{\splitkey{image\_wake}}
+
+The image-wake split is computed from downloaded \code{UxMean} PNGs under an
+external asset directory with \code{run\_N/images/UxMean} subfolders. Each run contributes the \code{Y-4}
+centreline/near-centreline side image plus three near-base cross-plane images:
+\code{X-14}, \code{X-15}, and \code{X-16}. For each image, the script scores
+the lower flow region using a reproducible color-intensity measure, then
+combines the \code{Y-4} score with the mean near-base X score. The top 20\% of
+runs by this image score form the OOD test set, and validation is sampled from
+the complementary training-side pool. Figure~\ref{fig:diagnostics} shows both
+the score used to define \splitkey{image\_wake} and the same split colored on
+drag coefficient.
+
+In the current metrics, the highest image-wake scores include
+\splitkey{run\_280}, \splitkey{run\_205}, \splitkey{run\_284},
+\splitkey{run\_309}, \splitkey{run\_315}, \splitkey{run\_362},
+\splitkey{run\_171}, and \splitkey{run\_295}.
+Figure~\ref{fig:wake_examples} compares the lowest and highest image-wake
+score cases using the \code{Y-4} centreline image and a nearer-base \code{X-14}
+image from the same PNG source used by the score.
+
+\begin{figure}[H]
+  \centering
+  \includegraphics[width=0.96\textwidth]{wake_score_examples.png}
+  \caption{Image-wake examples for \splitkey{image\_wake}. The low-score case is \splitkey{run\_318} with image score 0.3353, \code{Cd}=0.3871, and \code{Cl}=-0.2007. The high-score case is \splitkey{run\_280} with image score 0.3889, \code{Cd}=0.4921, and \code{Cl}=0.7001. The top row shows \code{Y-4}, which captures the centreline wake, and the bottom row shows the nearer-base \code{X-14} slice.}
+  \label{fig:wake_examples}
+\end{figure}
+
+\begin{figure}[H]
+  \centering
+  \includegraphics[width=0.96\textwidth]{split_diagnostics.png}
+  \caption{AhmedML split diagnostics. The panels show the full random baseline, high-drag holdout, low-drag holdout, STL-Chamfer geometry holdout, image-wake holdout, and image-wake split colored on \code{Cd}. The x-axis is run ID, and points are colored by each split's train, validation, and test partitions.}
+  \label{fig:diagnostics}
+\end{figure}
+
+\section*{Repeatability and transparency}
+
+The committed manifest is intended for normal use. The commands below are for
+auditing the split construction, recreating the figures, or refreshing the
+artifacts after changing the source data.
+
+From a clean split-package checkout, download \code{force\_mom\_all.csv},
+\code{geo\_parameters\_all.csv}, STLs, and the \code{UxMean} PNGs needed for
+the image-wake split into a directory outside this repository, then regenerate
+the metrics, manifest, and diagnostic plot:
+
+\begin{verbatim}
+ASSET_ROOT=../ahmedml_hf_assets
+
+python3 splits/download_hf_inputs.py --output-dir "$ASSET_ROOT" \
+  --include-stls --include-wake-images --workers 6
+python3 splits/compute_chamfer_splits.py --data-root "$ASSET_ROOT" \
+  --output-dir /tmp/ahmedml_chamfer_4096 --samples 4096 \
+  --workers 16 --sample-workers 2 --runs all \
+  --base-manifest splits/manifest.json
+cp /tmp/ahmedml_chamfer_4096/chamfer_metrics.csv \
+  splits/chamfer_metrics.csv
+python3 splits/compute_image_metrics.py --data-root "$ASSET_ROOT" \
+  --output splits/image_metrics.csv
+AHMEDML_DATA_ROOT="$ASSET_ROOT" python3 splits/generate_splits.py
+AHMEDML_DATA_ROOT="$ASSET_ROOT" python3 splits/visualize_splits.py
+python3 splits/create_example_figures.py --asset-root "$ASSET_ROOT" \
+  --force-root "$ASSET_ROOT"
+\end{verbatim}
+
+Full report regeneration then uses:
+
+\begin{verbatim}
+latexmk -pdf -cd splits/README.tex
+\end{verbatim}
+
+For an existing dataset checkout, set \code{AHMEDML\_DATA\_ROOT} to the
+directory containing \code{force\_mom\_all.csv} and
+\code{geo\_parameters\_all.csv}. Set \code{AHMEDML\_ASSET\_ROOT}, or pass
+\code{--asset-root}, to the directory containing the \code{run\_*/} STL and
+PNG assets. Large reconstruction inputs should remain outside the repository.
+The split directory contains only lightweight derived artifacts.
+
+The committed source artifacts were generated against:
+
+\begin{verbatim}
+force_mom_all.csv
+geo_parameters_all.csv
+<asset-root>/run_*/ahmed_*.stl
+<asset-root>/run_*/images/UxMean/*.png
+splits/chamfer_metrics.csv
+splits/image_metrics.csv
+splits/parameter_geometry_metrics.csv
+splits/geometry_score_examples.png
+splits/wake_score_examples.png
+splits/manifest.json
+\end{verbatim}
+
+\section*{Manifest format}
+
+\begin{verbatim}
+{
+    "full_train": ["run_1", "run_2", "..."],
+    "full_val": ["run_24", "..."],
+    "full_test": ["run_4", "..."],
+    "scarce_train": ["run_10", "..."]
+}
+\end{verbatim}
+
+Case IDs match on-disk directory names and are sorted numerically by run number.
+
+{\small
+\begin{thebibliography}{9}
+\bibitem{ahmedml_dataset}
+AhmedML dataset. \url{https://huggingface.co/datasets/neashton/ahmedml}.
+
+\bibitem{ahmedml_paper}
+AhmedML paper. \url{https://arxiv.org/abs/2407.20801}.
+
+\bibitem{noether_split}
+Noether AhmedML split. \url{https://github.com/Emmi-AI/noether/blob/main/src/noether/data/datasets/cfd/caeml/ahmedml/split.py}.
+\end{thebibliography}
+}
+
+\end{document}

splits/chamfer_metrics.csv

@@ -0,0 +1,501 @@
+run,nearest_neighbor_chamfer,mean_10_nn_chamfer,mean_all_chamfer,medoid_chamfer,medoid_run,ood_score
+1,0.010357541032135487,0.013690590858459473,0.04865063726902008,0.041591912508010864,169,0.013690590858459473
+2,0.010572250001132488,0.01343235932290554,0.03794974461197853,0.029901612550020218,169,0.01343235932290554
+3,0.010504302568733692,0.012163753621280193,0.037171244621276855,0.022878078743815422,169,0.012163753621280193
+4,0.010876658372581005,0.013760733418166637,0.03628207743167877,0.025077423080801964,169,0.013760733418166637
+5,0.010261986404657364,0.012191282585263252,0.04276628792285919,0.030164038762450218,169,0.012191282585263252
+6,0.010317728854715824,0.0132074523717165,0.038106419146060944,0.023702824488282204,169,0.0132074523717165
+7,0.010882643051445484,0.01363408099859953,0.03732367604970932,0.024635054171085358,169,0.01363408099859953
+8,0.012281513772904873,0.01408781111240387,0.051625486463308334,0.048639047890901566,169,0.01408781111240387
+9,0.010721694678068161,0.014059314504265785,0.037872955203056335,0.028307748958468437,169,0.014059314504265785
+10,0.010494369082152843,0.012107549235224724,0.04571130499243736,0.03461304306983948,169,0.012107549235224724
+11,0.012996241450309753,0.01513952948153019,0.05261416733264923,0.04217882454395294,169,0.01513952948153019
+12,0.013580425642430782,0.015355152077972889,0.05337259918451309,0.04756633937358856,169,0.015355152077972889
+13,0.0105053149163723,0.012279432266950607,0.03849048539996147,0.024683142080903053,169,0.012279432266950607
+14,0.01269200723618269,0.01443398930132389,0.043210823088884354,0.03342215344309807,169,0.01443398930132389
+15,0.011775338090956211,0.012955335900187492,0.04627624526619911,0.041541747748851776,169,0.012955335900187492
+16,0.01204174105077982,0.013836881145834923,0.0331878699362278,0.015999268740415573,169,0.013836881145834923
+17,0.0135778384283185,0.014551326632499695,0.05271439254283905,0.0488915741443634,169,0.014551326632499695
+18,0.01009734719991684,0.013269489631056786,0.04005815088748932,0.028368905186653137,169,0.013269489631056786
+19,0.011944156140089035,0.013226459734141827,0.043656084686517715,0.034411072731018066,169,0.013226459734141827
+20,0.008670798502862453,0.01183275319635868,0.04752139747142792,0.03644845634698868,169,0.01183275319635868
+21,0.011288318783044815,0.012375659309327602,0.04465587064623833,0.03644150495529175,169,0.012375659309327602
+22,0.01161975972354412,0.014294112101197243,0.052143555134534836,0.04341302439570427,169,0.014294112101197243
+23,0.01237957552075386,0.013178272172808647,0.033675894141197205,0.02162639982998371,169,0.013178272172808647
+24,0.012088504619896412,0.013226171024143696,0.03498808294534683,0.017937999218702316,169,0.013226171024143696
+25,0.011386222206056118,0.014233790338039398,0.03792916238307953,0.028757574036717415,169,0.014233790338039398
+26,0.011593397706747055,0.01263403333723545,0.03614781051874161,0.026515241712331772,169,0.01263403333723545
+27,0.011937460862100124,0.013041362166404724,0.03602125123143196,0.021866142749786377,169,0.013041362166404724
+28,0.011014118790626526,0.012982293963432312,0.039214491844177246,0.0297342948615551,169,0.012982293963432312
+29,0.011058664880692959,0.012772219255566597,0.031558580696582794,0.016012709587812424,169,0.012772219255566597
+30,0.009448019787669182,0.013106688857078552,0.04876336827874184,0.0371667742729187,169,0.013106688857078552
+31,0.011207286268472672,0.013107895851135254,0.03260891139507294,0.017815928906202316,169,0.013107895851135254
+32,0.01158224418759346,0.013211511075496674,0.04152649641036987,0.028362911194562912,169,0.013211511075496674
+33,0.010825266130268574,0.012831151485443115,0.0444299541413784,0.03455822914838791,169,0.012831151485443115
+34,0.01304931752383709,0.014198176562786102,0.04586717113852501,0.03888543322682381,169,0.014198176562786102
+35,0.010504302568733692,0.012299222871661186,0.036144278943538666,0.02077951282262802,169,0.012299222871661186
+36,0.008670798502862453,0.01176187302917242,0.04794199764728546,0.03677923232316971,169,0.01176187302917242
+37,0.010840436443686485,0.012705137953162193,0.033628467470407486,0.0189947709441185,169,0.012705137953162193
+38,0.012929731979966164,0.015548719093203545,0.045399293303489685,0.04049326106905937,169,0.015548719093203545
+39,0.011465747840702534,0.013341772370040417,0.03276221454143524,0.02040562406182289,169,0.013341772370040417
+40,0.011984668672084808,0.013245761394500732,0.04245888814330101,0.03140959143638611,169,0.013245761394500732
+41,0.010640677064657211,0.013269199058413506,0.03302765637636185,0.019320117309689522,169,0.013269199058413506
+42,0.012296306900680065,0.013638103380799294,0.033628910779953,0.021866878494620323,169,0.013638103380799294
+43,0.010353878140449524,0.011631451547145844,0.05218033865094185,0.0411733016371727,169,0.011631451547145844
+44,0.010414063930511475,0.013264410197734833,0.03247341141104698,0.01936945505440235,169,0.013264410197734833
+45,0.009908991865813732,0.011614302173256874,0.04611575976014137,0.03783326968550682,169,0.011614302173256874
+46,0.01060036476701498,0.012924512848258018,0.03914544731378555,0.02927315980195999,169,0.012924512848258018
+47,0.011697300709784031,0.013774638064205647,0.04154588282108307,0.028770368546247482,169,0.013774638064205647
+48,0.010871578939259052,0.011911588720977306,0.04977167770266533,0.03849468752741814,169,0.011911588720977306
+49,0.011416839435696602,0.013558918610215187,0.033581025898456573,0.016955574974417686,169,0.013558918610215187
+50,0.013092942535877228,0.014531808905303478,0.05154316499829292,0.04892740026116371,169,0.014531808905303478
+51,0.012127816677093506,0.014281360432505608,0.0461912676692009,0.03805425763130188,169,0.014281360432505608
+52,0.0153756532818079,0.018450217321515083,0.05257749930024147,0.04406584054231644,169,0.018450217321515083
+53,0.01052606850862503,0.013883814215660095,0.03767555207014084,0.02358272112905979,169,0.013883814215660095
+54,0.014489120803773403,0.01624922826886177,0.05832618474960327,0.05610630288720131,169,0.01624922826886177
+55,0.01131670642644167,0.014787177555263042,0.04689197614789009,0.03541342914104462,169,0.014787177555263042
+56,0.011664236895740032,0.014596186578273773,0.03776883706450462,0.023870833218097687,169,0.014596186578273773
+57,0.01068926602602005,0.012761947698891163,0.033345483243465424,0.019629035145044327,169,0.012761947698891163
+58,0.010844903998076916,0.013804329559206963,0.03768566995859146,0.028450194746255875,169,0.013804329559206963
+59,0.01230448205024004,0.014481755904853344,0.04620562493801117,0.04204600676894188,169,0.014481755904853344
+60,0.013804234564304352,0.01571488007903099,0.04984140396118164,0.046769287437200546,169,0.01571488007903099
+61,0.013420493341982365,0.014750729314982891,0.03528792783617973,0.022745948284864426,169,0.014750729314982891
+62,0.010008202865719795,0.014413413591682911,0.03926493227481842,0.028023121878504753,169,0.014413413591682911
+63,0.013189896941184998,0.0159463994204998,0.04469246417284012,0.03353248909115791,169,0.0159463994204998
+64,0.01202941033989191,0.013417975977063179,0.04617421701550484,0.04235498234629631,169,0.013417975977063179
+65,0.013628000393509865,0.01529519073665142,0.04180057346820831,0.03489040210843086,169,0.01529519073665142
+66,0.01146139856427908,0.013638284988701344,0.04044078290462494,0.029659828171133995,169,0.013638284988701344
+67,0.012673024088144302,0.013638140633702278,0.03510800004005432,0.019506491720676422,169,0.013638140633702278
+68,0.010334083810448647,0.013120383024215698,0.034227512776851654,0.018188098445534706,169,0.013120383024215698
+69,0.01121639832854271,0.01239126455038786,0.04262060672044754,0.030516376718878746,169,0.01239126455038786
+70,0.011909627355635166,0.013376638293266296,0.03207401931285858,0.016526056453585625,169,0.013376638293266296
+71,0.009886077605187893,0.010995633900165558,0.040852680802345276,0.028874173760414124,169,0.010995633900165558
+72,0.012350335717201233,0.013219550251960754,0.04106033220887184,0.03495529666543007,169,0.013219550251960754
+73,0.01166579220443964,0.013555760495364666,0.032679907977581024,0.014466953463852406,169,0.013555760495364666
+74,0.012409050017595291,0.01435908954590559,0.03573538735508919,0.02096012979745865,169,0.01435908954590559
+75,0.012423854321241379,0.014121102169156075,0.04910758510231972,0.04462150111794472,169,0.014121102169156075
+76,0.010733268223702908,0.01399337686598301,0.03870641440153122,0.024843085557222366,169,0.01399337686598301
+77,0.012551328167319298,0.014366689138114452,0.04228929802775383,0.034665945917367935,169,0.014366689138114452
+78,0.01060737669467926,0.013147721998393536,0.05095382779836655,0.04326262325048447,169,0.013147721998393536
+79,0.01148789469152689,0.013093968853354454,0.033699095249176025,0.01663936860859394,169,0.013093968853354454
+80,0.011426065117120743,0.01334718894213438,0.03650535270571709,0.02748291566967964,169,0.01334718894213438
+81,0.009900936856865883,0.013265612535178661,0.04821401089429855,0.03683061897754669,169,0.013265612535178661
+82,0.009966660290956497,0.013194538652896881,0.050037894397974014,0.03987065702676773,169,0.013194538652896881
+83,0.010662689805030823,0.0127017293125391,0.04164646565914154,0.030833769589662552,169,0.0127017293125391
+84,0.0093757975846529,0.012615008279681206,0.04807910695672035,0.03666431084275246,169,0.012615008279681206
+85,0.010912708938121796,0.013369726948440075,0.041920870542526245,0.03141314908862114,169,0.013369726948440075
+86,0.012249528430402279,0.013400924392044544,0.035127848386764526,0.022340185940265656,169,0.013400924392044544
+87,0.011106611229479313,0.012902927584946156,0.04596449062228203,0.04080753028392792,169,0.012902927584946156
+88,0.011770924553275108,0.012861068360507488,0.037088971585035324,0.026686634868383408,169,0.012861068360507488
+89,0.00986604206264019,0.011088532395660877,0.0389346107840538,0.02632840909063816,169,0.011088532395660877
+90,0.011495009064674377,0.013715717010200024,0.04029598832130432,0.02929885871708393,169,0.013715717010200024
+91,0.015154455788433552,0.01645401492714882,0.049716390669345856,0.04483016952872276,169,0.01645401492714882
+92,0.012690131552517414,0.015601078979671001,0.03642088547348976,0.022994497790932655,169,0.015601078979671001
+93,0.010127785615622997,0.013854896649718285,0.05737109109759331,0.04727990925312042,169,0.013854896649718285
+94,0.011426065117120743,0.0134203489869833,0.03795648738741875,0.030324328690767288,169,0.0134203489869833
+95,0.0093757975846529,0.012697160243988037,0.045601826161146164,0.033690761774778366,169,0.012697160243988037
+96,0.011923435144126415,0.016629505902528763,0.056065432727336884,0.047692399471998215,169,0.016629505902528763
+97,0.011058664880692959,0.01329448539763689,0.032006919384002686,0.016896458342671394,169,0.01329448539763689
+98,0.01138075441122055,0.015385940670967102,0.05146330967545509,0.04310939460992813,169,0.015385940670967102
+99,0.013045013882219791,0.013653993606567383,0.03851466253399849,0.026293626055121422,169,0.013653993606567383
+100,0.012302204966545105,0.014130542986094952,0.035762690007686615,0.022860191762447357,169,0.014130542986094952
+101,0.011945460923016071,0.014006802812218666,0.037765637040138245,0.02936636470258236,169,0.014006802812218666
+102,0.011421266943216324,0.013197961263358593,0.03592289984226227,0.024019083008170128,169,0.013197961263358593
+103,0.010169126093387604,0.012749174609780312,0.03907473757863045,0.025423632934689522,169,0.012749174609780312
+104,0.013960035517811775,0.01524632703512907,0.0428694486618042,0.03195544332265854,169,0.01524632703512907
+105,0.011309154331684113,0.013272625394165516,0.04414605349302292,0.03513475880026817,169,0.013272625394165516
+106,0.012887483462691307,0.014263750985264778,0.038466136902570724,0.02893574722111225,169,0.014263750985264778
+107,0.010418248362839222,0.0132241640239954,0.03534715250134468,0.0219224002212286,169,0.0132241640239954
+108,0.0104298647493124,0.012546275742352009,0.040524281561374664,0.02821986749768257,169,0.012546275742352009
+109,0.012536444701254368,0.01516571082174778,0.04296722635626793,0.038234226405620575,169,0.01516571082174778
+110,0.011395121924579144,0.013311405666172504,0.051051679998636246,0.04025038704276085,169,0.013311405666172504
+111,0.011449193581938744,0.012998444959521294,0.04522872343659401,0.03472425043582916,169,0.012998444959521294
+112,0.01023749727755785,0.013374252244830132,0.04694291576743126,0.039650995284318924,169,0.013374252244830132
+113,0.010959742590785027,0.012853287160396576,0.03630496934056282,0.027408868074417114,169,0.012853287160396576
+114,0.009379418566823006,0.011747343465685844,0.03851553052663803,0.02557600662112236,169,0.011747343465685844
+115,0.011733079329133034,0.014642896130681038,0.043246395885944366,0.03311732038855553,169,0.014642896130681038
+116,0.009472950361669064,0.011532251723110676,0.05319765955209732,0.04259821027517319,169,0.011532251723110676
+117,0.010769214481115341,0.013947260566055775,0.039033327251672745,0.025379430502653122,169,0.013947260566055775
+118,0.010147248394787312,0.014274662360548973,0.03791704773902893,0.02775842696428299,169,0.014274662360548973
+119,0.0116241704672575,0.01297819148749113,0.03310256078839302,0.01670643500983715,169,0.01297819148749113
+120,0.012180589139461517,0.013908512890338898,0.05003681778907776,0.04666405916213989,169,0.013908512890338898
+121,0.009688323363661766,0.011229127645492554,0.0555601567029953,0.044923458248376846,169,0.011229127645492554
+122,0.011216823942959309,0.01337288785725832,0.05122964084148407,0.04427093639969826,169,0.01337288785725832
+123,0.010497553274035454,0.01319989562034607,0.04904099926352501,0.03820788487792015,169,0.01319989562034607
+124,0.011543292552232742,0.013738933019340038,0.035421013832092285,0.019722042605280876,169,0.013738933019340038
+125,0.012112629599869251,0.013487784191966057,0.033748809248209,0.017572980374097824,169,0.013487784191966057
+126,0.009981540963053703,0.013028053566813469,0.0378197506070137,0.025524163618683815,169,0.013028053566813469
+127,0.010737120173871517,0.01280774176120758,0.03183222934603691,0.01633126102387905,169,0.01280774176120758
+128,0.01081155240535736,0.012716777622699738,0.04658561199903488,0.039348237216472626,169,0.012716777622699738
+129,0.011384904384613037,0.013501519337296486,0.04408008232712746,0.03414246067404747,169,0.013501519337296486
+130,0.012546095997095108,0.014109174720942974,0.04149055480957031,0.03058960847556591,169,0.014109174720942974
+131,0.010902199894189835,0.013576202094554901,0.033703405410051346,0.018634911626577377,169,0.013576202094554901
+132,0.01020148303359747,0.014086559414863586,0.04268839582800865,0.03205270692706108,169,0.014086559414863586
+133,0.00999896414577961,0.01319984532892704,0.04648995399475098,0.035612281411886215,169,0.01319984532892704
+134,0.010713523253798485,0.012850928120315075,0.033944107592105865,0.020060069859027863,169,0.012850928120315075
+135,0.014568226411938667,0.016313280910253525,0.038877952843904495,0.027980564162135124,169,0.016313280910253525
+136,0.011958618648350239,0.01275539118796587,0.04051323980093002,0.033278461545705795,169,0.01275539118796587
+137,0.01021519210189581,0.012098973616957664,0.033313363790512085,0.016026752069592476,169,0.012098973616957664
+138,0.011580163612961769,0.012987379916012287,0.04385480657219887,0.034075044095516205,169,0.012987379916012287
+139,0.010937022045254707,0.012945261783897877,0.04624579846858978,0.037750598043203354,169,0.012945261783897877
+140,0.011381213553249836,0.01305419486016035,0.05109994858503342,0.047309860587120056,169,0.01305419486016035
+141,0.010733268223702908,0.014084281399846077,0.037378884851932526,0.022426651790738106,169,0.014084281399846077
+142,0.011734024621546268,0.013068070635199547,0.032459963113069534,0.015457697212696075,169,0.013068070635199547
+143,0.011101428419351578,0.013097638264298439,0.03390609100461006,0.021419335156679153,169,0.013097638264298439
+144,0.009687993675470352,0.012539966031908989,0.03301824629306793,0.016582172363996506,169,0.012539966031908989
+145,0.011292792856693268,0.01634063944220543,0.05055556818842888,0.04004611447453499,169,0.01634063944220543
+146,0.012077175080776215,0.014003257267177105,0.038964878767728806,0.030319588258862495,169,0.014003257267177105
+147,0.012714485637843609,0.013663055375218391,0.03288885951042175,0.015630880370736122,169,0.013663055375218391
+148,0.011552558280527592,0.012877839617431164,0.04266992583870888,0.03384780511260033,169,0.012877839617431164
+149,0.010763539001345634,0.01323053427040577,0.03194791451096535,0.017586681991815567,169,0.01323053427040577
+150,0.012582026422023773,0.013508206233382225,0.03606696426868439,0.021826153621077538,169,0.013508206233382225
+151,0.009468693286180496,0.012768697924911976,0.03970818594098091,0.025727666914463043,169,0.012768697924911976
+152,0.011852406896650791,0.014661249704658985,0.05484653264284134,0.04569438472390175,169,0.014661249704658985
+153,0.014044462703168392,0.017248934134840965,0.05234399065375328,0.04853992909193039,169,0.017248934134840965
+154,0.010357541032135487,0.014068993739783764,0.04956459254026413,0.04308084771037102,169,0.014068993739783764
+155,0.013366189785301685,0.013893465511500835,0.04672841727733612,0.042694538831710815,169,0.013893465511500835
+156,0.010643171146512032,0.014249448664486408,0.03719017282128334,0.026420580223202705,169,0.014249448664486408
+157,0.011591088958084583,0.013503086753189564,0.04370786249637604,0.03306364268064499,169,0.013503086753189564
+158,0.00989616010338068,0.013056893832981586,0.03297286853194237,0.015695571899414062,169,0.013056893832981586
+159,0.011586766690015793,0.013472901657223701,0.035449519753456116,0.024902580305933952,169,0.013472901657223701
+160,0.013098944909870625,0.013816324062645435,0.034029796719551086,0.01615544594824314,169,0.013816324062645435
+161,0.010500910691916943,0.013050666078925133,0.04874369129538536,0.03736964240670204,169,0.013050666078925133
+162,0.01118577178567648,0.014528946951031685,0.04310831055045128,0.03884045034646988,169,0.014528946951031685
+163,0.012011677958071232,0.01341983862221241,0.05134465545415878,0.042513590306043625,169,0.01341983862221241
+164,0.013405879959464073,0.014905637130141258,0.04514201730489731,0.036239515990018845,169,0.014905637130141258
+165,0.009494186379015446,0.012052798643708229,0.041412971913814545,0.028616128489375114,169,0.012052798643708229
+166,0.012127816677093506,0.014019397087395191,0.04333943873643875,0.034706249833106995,169,0.014019397087395191
+167,0.010303272865712643,0.014194652438163757,0.055221423506736755,0.04468696936964989,169,0.014194652438163757
+168,0.012286652810871601,0.013381886295974255,0.033620502799749374,0.02210846357047558,169,0.013381886295974255
+169,0.011941477656364441,0.013881554827094078,0.03089011088013649,0.0,169,0.013881554827094078
+170,0.010902199894189835,0.013418538495898247,0.03387140855193138,0.01865852251648903,169,0.013418538495898247
+171,0.011853178963065147,0.013395830988883972,0.042039621621370316,0.030933812260627747,169,0.013395830988883972
+172,0.011416839435696602,0.01397701632231474,0.03271806985139847,0.016180023550987244,169,0.01397701632231474
+173,0.009966660290956497,0.013320195488631725,0.05213264748454094,0.04265931248664856,169,0.013320195488631725
+174,0.012661087326705456,0.013532793149352074,0.036370061337947845,0.021564219146966934,169,0.013532793149352074
+175,0.010404961183667183,0.012791653163731098,0.041372835636138916,0.030284907668828964,169,0.012791653163731098
+176,0.010008202865719795,0.014492213726043701,0.041207801550626755,0.0305919349193573,169,0.014492213726043701
+177,0.011185373179614544,0.0126521410420537,0.03745856136083603,0.02843787893652916,169,0.0126521410420537
+178,0.010978900827467442,0.013392655178904533,0.04593479633331299,0.03497040644288063,169,0.013392655178904533
+179,0.010919999331235886,0.013185019604861736,0.03245563432574272,0.013809571042656898,169,0.013185019604861736
+180,0.013240884989500046,0.014226404950022697,0.04969528689980507,0.04548085853457451,169,0.014226404950022697
+181,0.010844903998076916,0.01364889182150364,0.035884786397218704,0.02572653815150261,169,0.01364889182150364
+182,0.013776145875453949,0.014983209781348705,0.04398051276803017,0.037584658712148666,169,0.014983209781348705
+183,0.013187915086746216,0.014454295858740807,0.037279848009347916,0.025696279481053352,169,0.014454295858740807
+184,0.011271454393863678,0.01350544672459364,0.041451554745435715,0.03617127612233162,169,0.01350544672459364
+185,0.010734915733337402,0.013494430109858513,0.04081443324685097,0.030338963493704796,169,0.013494430109858513
+186,0.011435337364673615,0.013222220353782177,0.035462502390146255,0.02654467150568962,169,0.013222220353782177
+187,0.011552011594176292,0.012920239940285683,0.03630758449435234,0.02515203133225441,169,0.012920239940285683
+188,0.010317728854715824,0.013369647786021233,0.037483301013708115,0.022090032696723938,169,0.013369647786021233
+189,0.011646714061498642,0.013204103335738182,0.03476595878601074,0.021336575970053673,169,0.013204103335738182
+190,0.011852406896650791,0.013308001682162285,0.04922044277191162,0.03965584188699722,169,0.013308001682162285
+191,0.010966521687805653,0.012762469239532948,0.04048958793282509,0.03342781960964203,169,0.012762469239532948
+192,0.010937022045254707,0.012231101281940937,0.04593294486403465,0.03725199028849602,169,0.012231101281940937
+193,0.012152441777288914,0.01398825366050005,0.042736880481243134,0.0297350212931633,169,0.01398825366050005
+194,0.011314183473587036,0.013666284270584583,0.05048230290412903,0.03960035368800163,169,0.013666284270584583
+195,0.011593335308134556,0.012917488813400269,0.03615304082632065,0.026454085484147072,169,0.012917488813400269
+196,0.010664497502148151,0.013268372043967247,0.034644268453121185,0.024839861318469048,169,0.013268372043967247
+197,0.011326564475893974,0.013753116130828857,0.038378432393074036,0.03084961511194706,169,0.013753116130828857
+198,0.011593335308134556,0.013276025652885437,0.03795047104358673,0.029317693784832954,169,0.013276025652885437
+199,0.011761326342821121,0.01400902308523655,0.04065585136413574,0.030628984794020653,169,0.01400902308523655
+200,0.013906985521316528,0.015549525618553162,0.05047433078289032,0.04199850931763649,169,0.015549525618553162
+201,0.011149405501782894,0.012744681909680367,0.04228198528289795,0.03451281040906906,169,0.012744681909680367
+202,0.009785549715161324,0.011674960143864155,0.038871996104717255,0.025036128237843513,169,0.011674960143864155
+203,0.011374140158295631,0.013030394911766052,0.0320621132850647,0.017362434417009354,169,0.013030394911766052
+204,0.01175325270742178,0.014249923638999462,0.03540676832199097,0.01935802958905697,169,0.014249923638999462
+205,0.010500910691916943,0.01384368073195219,0.04924718290567398,0.03789524734020233,169,0.01384368073195219
+206,0.011900225654244423,0.015052398666739464,0.0529504120349884,0.04871596023440361,169,0.015052398666739464
+207,0.010588170029222965,0.013083333149552345,0.04133707284927368,0.028324075043201447,169,0.013083333149552345
+208,0.010773645713925362,0.012463981285691261,0.04106004536151886,0.028805816546082497,169,0.012463981285691261
+209,0.011704918928444386,0.013949602842330933,0.033992912620306015,0.018555646762251854,169,0.013949602842330933
+210,0.010404402390122414,0.01244891993701458,0.04090415686368942,0.03225989267230034,169,0.01244891993701458
+211,0.012556365691125393,0.01468233298510313,0.05480525642633438,0.05208421126008034,169,0.01468233298510313
+212,0.011279853992164135,0.013348126783967018,0.03437398374080658,0.020899901166558266,169,0.013348126783967018
+213,0.011254072189331055,0.012846795842051506,0.03449279069900513,0.023174021393060684,169,0.012846795842051506
+214,0.012093866243958473,0.013572569005191326,0.03532348573207855,0.025463581085205078,169,0.013572569005191326
+215,0.011309154331684113,0.01417281199246645,0.0460660457611084,0.03799126297235489,169,0.01417281199246645
+216,0.013556423597037792,0.014698070473968983,0.049830250442028046,0.04471948742866516,169,0.014698070473968983
+217,0.01132768951356411,0.013261961750686169,0.05096665769815445,0.04057219251990318,169,0.013261961750686169
+218,0.0111992247402668,0.013286024332046509,0.03954698145389557,0.028011495247483253,169,0.013286024332046509
+219,0.01031485851854086,0.012021840550005436,0.032195452600717545,0.017869608476758003,169,0.012021840550005436
+220,0.011118012480437756,0.013005584478378296,0.04258975014090538,0.030280034989118576,169,0.013005584478378296
+221,0.01060844399034977,0.013753250241279602,0.04978196322917938,0.041274622082710266,169,0.013753250241279602
+222,0.012399422004818916,0.013856127858161926,0.04006905108690262,0.030034715309739113,169,0.013856127858161926
+223,0.012180589139461517,0.01464275736361742,0.050883948802948,0.04768113046884537,169,0.01464275736361742
+224,0.010737120173871517,0.013661992736160755,0.03230005130171776,0.016684923321008682,169,0.013661992736160755
+225,0.01118577178567648,0.014370235614478588,0.04185624420642853,0.03696420043706894,169,0.014370235614478588
+226,0.011775338090956211,0.012903176248073578,0.04528892785310745,0.040637001395225525,169,0.012903176248073578
+227,0.011723962612450123,0.014032225124537945,0.036280617117881775,0.028414545580744743,169,0.014032225124537945
+228,0.010898694396018982,0.012462596409022808,0.04677361622452736,0.035694871097803116,169,0.012462596409022808
+229,0.011871467344462872,0.014979556202888489,0.041836824268102646,0.035438526421785355,169,0.014979556202888489
+230,0.010629886761307716,0.01241560559719801,0.04484602063894272,0.032876092940568924,169,0.01241560559719801
+231,0.01084572821855545,0.013536319136619568,0.036565739661455154,0.02608935907483101,169,0.013536319136619568
+232,0.009468693286180496,0.01335203181952238,0.041276995092630386,0.02834169752895832,169,0.01335203181952238
+233,0.009688323363661766,0.01193708460777998,0.055913716554641724,0.045428838580846786,169,0.01193708460777998
+234,0.010404402390122414,0.01215923298150301,0.042591895908117294,0.03439779579639435,169,0.01215923298150301
+235,0.011250641196966171,0.013431185856461525,0.03315375745296478,0.017821624875068665,169,0.013431185856461525
+236,0.011402510106563568,0.013259418308734894,0.036382585763931274,0.020800350233912468,169,0.013259418308734894
+237,0.011215117760002613,0.014091539196670055,0.03316781297326088,0.01575203612446785,169,0.014091539196670055
+238,0.011338780634105206,0.013788002543151379,0.03968018293380737,0.026850352063775063,169,0.013788002543151379
+239,0.011914322152733803,0.014257891103625298,0.03895558416843414,0.030869802460074425,169,0.014257891103625298
+240,0.011900225654244423,0.01452569104731083,0.050763122737407684,0.04636944830417633,169,0.01452569104731083
+241,0.009687993675470352,0.01267771702259779,0.03335162624716759,0.01758125238120556,169,0.01267771702259779
+242,0.011502745561301708,0.014636045321822166,0.04077209532260895,0.031163720414042473,169,0.014636045321822166
+243,0.011712696403265,0.014891425147652626,0.043310243636369705,0.03803133964538574,169,0.014891425147652626
+244,0.013015382923185825,0.01707320287823677,0.04477338865399361,0.03923424333333969,169,0.01707320287823677
+245,0.011580163612961769,0.01340417005121708,0.04257560521364212,0.03189106285572052,169,0.01340417005121708
+246,0.010433965362608433,0.013321056962013245,0.03774134814739227,0.02268543653190136,169,0.013321056962013245
+247,0.011025742627680302,0.012925359420478344,0.03331631422042847,0.01699744164943695,169,0.012925359420478344
+248,0.012116420082747936,0.01328594982624054,0.03273608908057213,0.016844933852553368,169,0.01328594982624054
+249,0.014047614298760891,0.016055945307016373,0.05421910807490349,0.0452127680182457,169,0.016055945307016373
+250,0.011985601857304573,0.013645539991557598,0.04376092180609703,0.0379464365541935,169,0.013645539991557598
+251,0.010621599853038788,0.011806212365627289,0.04074833542108536,0.028166409581899643,169,0.011806212365627289
+252,0.011984668672084808,0.013859391212463379,0.044709473848342896,0.03504008427262306,169,0.013859391212463379
+253,0.014048502780497074,0.01578037068247795,0.05172326788306236,0.045805756002664566,169,0.01578037068247795
+254,0.010847714729607105,0.014215665869414806,0.03705144301056862,0.0243845172226429,169,0.014215665869414806
+255,0.010629886761307716,0.012572130188345909,0.04484247788786888,0.03311479836702347,169,0.012572130188345909
+256,0.010835663415491581,0.013756836764514446,0.03514743968844414,0.020892217755317688,169,0.013756836764514446
+257,0.010701730847358704,0.012353919446468353,0.04134836792945862,0.028551368042826653,169,0.012353919446468353
+258,0.0106181176379323,0.012380838394165039,0.045042283833026886,0.03322942182421684,169,0.012380838394165039
+259,0.010492865927517414,0.011924775317311287,0.051020462065935135,0.040196944028139114,169,0.011924775317311287
+260,0.011263162828981876,0.012754641473293304,0.034599419683218,0.0231328122317791,169,0.012754641473293304
+261,0.012209488078951836,0.01378460694104433,0.044566329568624496,0.03772500529885292,169,0.01378460694104433
+262,0.009379418566823006,0.011495020240545273,0.03889671340584755,0.026163702830672264,169,0.011495020240545273
+263,0.011518187820911407,0.013949600979685783,0.03681936487555504,0.028149288147687912,169,0.013949600979685783
+264,0.0106181176379323,0.012015251442790031,0.04292134568095207,0.03117644600570202,169,0.012015251442790031
+265,0.011027345433831215,0.012771273963153362,0.03366134688258171,0.01708228327333927,169,0.012771273963153362
+266,0.011999454349279404,0.015297437086701393,0.054215483367443085,0.050546418875455856,169,0.015297437086701393
+267,0.012770552188158035,0.013634378090500832,0.03186166286468506,0.012836679816246033,169,0.013634378090500832
+268,0.01184145174920559,0.01414093654602766,0.03743555396795273,0.029257316142320633,169,0.01414093654602766
+269,0.012451263144612312,0.014083700254559517,0.04816661775112152,0.037338968366384506,169,0.014083700254559517
+270,0.010716166347265244,0.0125215919688344,0.031957585364580154,0.017772184684872627,169,0.0125215919688344
+271,0.012766929343342781,0.014348057098686695,0.04647931829094887,0.040372334420681,169,0.014348057098686695
+272,0.011101428419351578,0.012598857283592224,0.03435538709163666,0.02300589345395565,169,0.012598857283592224
+273,0.011292792856693268,0.01587289571762085,0.05092231184244156,0.04084803909063339,169,0.01587289571762085
+274,0.010418248362839222,0.013520397245883942,0.03546997159719467,0.02225872501730919,169,0.013520397245883942
+275,0.011674921028316021,0.013833025470376015,0.04200941324234009,0.03708752617239952,169,0.013833025470376015
+276,0.0115562304854393,0.014109927229583263,0.0393529012799263,0.02967974916100502,169,0.014109927229583263
+277,0.010147248394787312,0.014034828171133995,0.03763913735747337,0.027820425108075142,169,0.014034828171133995
+278,0.01279937569051981,0.013704325072467327,0.03687933087348938,0.023071374744176865,169,0.013704325072467327
+279,0.009771635755896568,0.012723475694656372,0.043593693524599075,0.031952738761901855,169,0.012723475694656372
+280,0.010898694396018982,0.013816501013934612,0.047700121998786926,0.0368560329079628,169,0.013816501013934612
+281,0.011194635182619095,0.012918124906718731,0.037855420261621475,0.02674740180373192,169,0.012918124906718731
+282,0.010973755270242691,0.012561110779643059,0.04970328137278557,0.04216810688376427,169,0.012561110779643059
+283,0.011093860492110252,0.01270564179867506,0.038491856306791306,0.024157822132110596,169,0.01270564179867506
+284,0.011784722097218037,0.013356024399399757,0.03411378338932991,0.018084228038787842,169,0.013356024399399757
+285,0.011014118790626526,0.012760041281580925,0.03950611501932144,0.030720040202140808,169,0.012760041281580925
+286,0.010769214481115341,0.014049530029296875,0.0383586548268795,0.023915987461805344,169,0.014049530029296875
+287,0.01160685159265995,0.012895124964416027,0.03774568811058998,0.030350415036082268,169,0.012895124964416027
+288,0.013216843828558922,0.01465779822319746,0.05148077756166458,0.0472051277756691,169,0.01465779822319746
+289,0.01082831621170044,0.013497253879904747,0.03924482688307762,0.025891505181789398,169,0.013497253879904747
+290,0.010734915733337402,0.013525607995688915,0.040771592408418655,0.030581749975681305,169,0.013525607995688915
+291,0.01242693979293108,0.01481783390045166,0.045619718730449677,0.03363559767603874,169,0.01481783390045166
+292,0.01049918681383133,0.012741724029183388,0.035414036363363266,0.01822204701602459,169,0.012741724029183388
+293,0.012842979282140732,0.014050936326384544,0.04137042164802551,0.0322289913892746,169,0.014050936326384544
+294,0.010404961183667183,0.012477436102926731,0.04301716387271881,0.03303366154432297,169,0.012477436102926731
+295,0.011879093945026398,0.01442691683769226,0.038327816873788834,0.031135180965065956,169,0.01442691683769226
+296,0.012667828239500523,0.014976921491324902,0.055398453027009964,0.04485822096467018,169,0.014976921491324902
+297,0.01111126970499754,0.015619230456650257,0.05308528244495392,0.0446740984916687,169,0.015619230456650257
+298,0.010713523253798485,0.012582868337631226,0.03402352333068848,0.0204630009829998,169,0.012582868337631226
+299,0.011384904384613037,0.013018171302974224,0.04196666181087494,0.03222380205988884,169,0.013018171302974224
+300,0.011106611229479313,0.013038167729973793,0.04772979021072388,0.04318398982286453,169,0.013038167729973793
+301,0.011261088773608208,0.015199368819594383,0.05374119430780411,0.0443667471408844,169,0.015199368819594383
+302,0.01021083164960146,0.012359409593045712,0.05561232939362526,0.044815871864557266,169,0.012359409593045712
+303,0.009908991865813732,0.011942187324166298,0.043877702206373215,0.03525780513882637,169,0.011942187324166298
+304,0.011524110101163387,0.012652851641178131,0.0403379462659359,0.03254327550530434,169,0.012652851641178131
+305,0.012192179448902607,0.01390791218727827,0.04607390612363815,0.04234132170677185,169,0.01390791218727827
+306,0.012624957598745823,0.01399662159383297,0.04154222086071968,0.032618198543787,169,0.01399662159383297
+307,0.011299783363938332,0.012552117928862572,0.037130337208509445,0.02253635600209236,169,0.012552117928862572
+308,0.009472950361669064,0.011533142998814583,0.052787113934755325,0.0423688068985939,169,0.011533142998814583
+309,0.014249342493712902,0.015293577685952187,0.04698023200035095,0.04291485622525215,169,0.015293577685952187
+310,0.011149944737553596,0.013293052092194557,0.039458323270082474,0.029361853376030922,169,0.013293052092194557
+311,0.011713121086359024,0.013825195841491222,0.03903516009449959,0.028414130210876465,169,0.013825195841491222
+312,0.010572250001132488,0.013059193268418312,0.03662726283073425,0.027917927131056786,169,0.013059193268418312
+313,0.011585571803152561,0.01343546248972416,0.041047997772693634,0.028444934636354446,169,0.01343546248972416
+314,0.01111126970499754,0.014760300517082214,0.04933532699942589,0.04042944684624672,169,0.014760300517082214
+315,0.013681606389582157,0.014804547652602196,0.036404289305210114,0.02405526302754879,169,0.014804547652602196
+316,0.010353878140449524,0.01204890850931406,0.05171237140893936,0.04103763401508331,169,0.01204890850931406
+317,0.011279853992164135,0.013072160072624683,0.03387465327978134,0.02117427997291088,169,0.013072160072624683
+318,0.01253793016076088,0.015375259332358837,0.03979242220520973,0.030357925221323967,169,0.015375259332358837
+319,0.011914322152733803,0.01513690035790205,0.041032783687114716,0.033024031668901443,169,0.01513690035790205
+320,0.012958399020135403,0.014284281060099602,0.04854034259915352,0.04392723739147186,169,0.014284281060099602
+321,0.01256552990525961,0.014354756101965904,0.03495242819190025,0.0248173326253891,169,0.014354756101965904
+322,0.010172235779464245,0.013537396676838398,0.035071298480033875,0.01956842839717865,169,0.013537396676838398
+323,0.01217289175838232,0.015161019749939442,0.04367910325527191,0.03919535130262375,169,0.015161019749939442
+324,0.011131348088383675,0.013504673726856709,0.052435293793678284,0.04254194349050522,169,0.013504673726856709
+325,0.010640677064657211,0.013547057285904884,0.03401171416044235,0.020462622866034508,169,0.013547057285904884
+326,0.01100219041109085,0.012872489169239998,0.032631535083055496,0.017018219456076622,169,0.012872489169239998
+327,0.011241653002798557,0.012129148468375206,0.04455023631453514,0.03307415544986725,169,0.012129148468375206
+328,0.011647050268948078,0.013645490631461143,0.03970686346292496,0.032666027545928955,169,0.013645490631461143
+329,0.010721694678068161,0.013667549006640911,0.036653079092502594,0.025668881833553314,169,0.013667549006640911
+330,0.010294605977833271,0.013467902317643166,0.04521763324737549,0.03391348943114281,169,0.013467902317643166
+331,0.011215117760002613,0.013440619222819805,0.03278433531522751,0.016197016462683678,169,0.013440619222819805
+332,0.011264560744166374,0.012990276329219341,0.03428316488862038,0.021313704550266266,169,0.012990276329219341
+333,0.012929731979966164,0.01568654365837574,0.04359880089759827,0.038480885326862335,169,0.01568654365837574
+334,0.012304640375077724,0.014982655644416809,0.0558706633746624,0.05315463989973068,169,0.014982655644416809
+335,0.012872176244854927,0.01596795581281185,0.04035651311278343,0.029504546895623207,169,0.01596795581281185
+336,0.011101034469902515,0.013886317610740662,0.03754707798361778,0.028799667954444885,169,0.013886317610740662
+337,0.011724203824996948,0.013455711305141449,0.04276690632104874,0.03626361861824989,169,0.013455711305141449
+338,0.01068926602602005,0.013964165933430195,0.03484465926885605,0.022344250231981277,169,0.013964165933430195
+339,0.011256986297667027,0.015392757952213287,0.047873176634311676,0.03624037653207779,169,0.015392757952213287
+340,0.010969149880111217,0.012846079654991627,0.034604113548994064,0.016947701573371887,169,0.012846079654991627
+341,0.012304640375077724,0.01415590476244688,0.05288112536072731,0.049649760127067566,169,0.01415590476244688
+342,0.011502745561301708,0.014348392374813557,0.03836114704608917,0.02763376384973526,169,0.014348392374813557
+343,0.013473669067025185,0.014655254781246185,0.05079434812068939,0.044735733419656754,169,0.014655254781246185
+344,0.011208008043467999,0.013098162598907948,0.031128644943237305,0.013973859138786793,169,0.013098162598907948
+345,0.010763539001345634,0.013808119110763073,0.03266981616616249,0.018334975466132164,169,0.013808119110763073
+346,0.010716166347265244,0.012592824175953865,0.03193604573607445,0.017114853486418724,169,0.012592824175953865
+347,0.012302204966545105,0.013252618722617626,0.03356850892305374,0.01824871636927128,169,0.013252618722617626
+348,0.009596102871000767,0.013385375030338764,0.039268333464860916,0.025245755910873413,169,0.013385375030338764
+349,0.014256675727665424,0.015240371227264404,0.05306882783770561,0.04996904358267784,169,0.015240371227264404
+350,0.011674921028316021,0.01347727607935667,0.042439017444849014,0.036980386823415756,169,0.01347727607935667
+351,0.011474885046482086,0.013437802903354168,0.0592280775308609,0.048896778374910355,169,0.013437802903354168
+352,0.010744956322014332,0.01232806034386158,0.04719338193535805,0.039135608822107315,169,0.01232806034386158
+353,0.012013934552669525,0.01328761875629425,0.045754339545965195,0.0417117103934288,169,0.01328761875629425
+354,0.009722290560603142,0.011722131632268429,0.038862645626068115,0.02624615654349327,169,0.011722131632268429
+355,0.010303272865712643,0.014427637681365013,0.058610949665308,0.048269122838974,169,0.014427637681365013
+356,0.010294605977833271,0.012516127899289131,0.04643803834915161,0.03512959182262421,169,0.012516127899289131
+357,0.01239160168915987,0.014501561410725117,0.039593953639268875,0.02644091099500656,169,0.014501561410725117
+358,0.013519312255084515,0.014223639853298664,0.04677971452474594,0.040047094225883484,169,0.014223639853298664
+359,0.010585015639662743,0.012606780044734478,0.03524170070886612,0.01953369937837124,169,0.012606780044734478
+360,0.010919999331235886,0.01306373905390501,0.03361029550433159,0.017061803489923477,169,0.01306373905390501
+361,0.012484819628298283,0.014820491895079613,0.05054266005754471,0.040529705584049225,169,0.014820491895079613
+362,0.011999454349279404,0.014920206740498543,0.054000504314899445,0.050823621451854706,169,0.014920206740498543
+363,0.01290583610534668,0.013919848017394543,0.0331302210688591,0.018562424927949905,169,0.013919848017394543
+364,0.012582404538989067,0.01501145027577877,0.04325275123119354,0.03590875864028931,169,0.01501145027577877
+365,0.01021519210189581,0.012530295178294182,0.033557113260030746,0.01685057021677494,169,0.012530295178294182
+366,0.01189735159277916,0.01296002883464098,0.04239292070269585,0.0379229411482811,169,0.01296002883464098
+367,0.012009236961603165,0.013409510254859924,0.03311298042535782,0.016662130132317543,169,0.013409510254859924
+368,0.010852149687707424,0.012518242001533508,0.03331515192985535,0.01931639201939106,169,0.012518242001533508
+369,0.010645338334143162,0.013527944684028625,0.03738512471318245,0.022130673751235008,169,0.013527944684028625
+370,0.012168092653155327,0.013566744513809681,0.038858089596033096,0.031534746289253235,169,0.013566744513809681
+371,0.011254072189331055,0.013185888528823853,0.03479038178920746,0.022767961025238037,169,0.013185888528823853
+372,0.010835663415491581,0.013771791942417622,0.03414636477828026,0.01903950236737728,169,0.013771791942417622
+373,0.011749468743801117,0.012823881581425667,0.03183409944176674,0.014446389861404896,169,0.012823881581425667
+374,0.012556365691125393,0.014065970666706562,0.05184389278292656,0.0488707609474659,169,0.014065970666706562
+375,0.012902868911623955,0.014970242977142334,0.04165992513298988,0.030973339453339577,169,0.014970242977142334
+376,0.011830451898276806,0.013627223670482635,0.03686891868710518,0.028846431523561478,169,0.013627223670482635
+377,0.010664497502148151,0.013226961717009544,0.0355469174683094,0.026578662917017937,169,0.013226961717009544
+378,0.012152480892837048,0.013518902473151684,0.03525332361459732,0.019831296056509018,169,0.013518902473151684
+379,0.011435337364673615,0.013270871713757515,0.0362459197640419,0.028474286198616028,169,0.013270871713757515
+380,0.01284860260784626,0.01427440159022808,0.04558807611465454,0.0346531942486763,169,0.01427440159022808
+381,0.00997646152973175,0.012371323071420193,0.032385360449552536,0.015307781286537647,169,0.012371323071420193
+382,0.013737078756093979,0.01617712900042534,0.041486166417598724,0.030546698719263077,169,0.01617712900042534
+383,0.010876658372581005,0.01353788934648037,0.035648148506879807,0.025036590173840523,169,0.01353788934648037
+384,0.011958618648350239,0.012710563838481903,0.04036254063248634,0.03271256014704704,169,0.012710563838481903
+385,0.0128812026232481,0.0141056003049016,0.053287796676158905,0.042226776480674744,169,0.0141056003049016
+386,0.01023749727755785,0.013071788474917412,0.05088447034358978,0.044489964842796326,169,0.013071788474917412
+387,0.012253457680344582,0.013351500034332275,0.0343947596848011,0.019790131598711014,169,0.013351500034332275
+388,0.01060737669467926,0.012527463026344776,0.04877273365855217,0.04097197949886322,169,0.012527463026344776
+389,0.010643171146512032,0.013898898847401142,0.03631775826215744,0.024825120344758034,169,0.013898898847401142
+390,0.009771635755896568,0.012238021939992905,0.04323168843984604,0.03173412010073662,169,0.012238021939992905
+391,0.01193608995527029,0.014044588431715965,0.03772968053817749,0.028422044590115547,169,0.014044588431715965
+392,0.011793768033385277,0.01313529908657074,0.03806609287858009,0.030496101826429367,169,0.01313529908657074
+393,0.01148789469152689,0.01308235339820385,0.033183395862579346,0.015516383573412895,169,0.01308235339820385
+394,0.012744762934744358,0.015058371238410473,0.044011324644088745,0.03576405346393585,169,0.015058371238410473
+395,0.010293045081198215,0.012216633185744286,0.039328910410404205,0.02862493321299553,169,0.012216633185744286
+396,0.010172235779464245,0.012786383740603924,0.03360748291015625,0.016711728647351265,169,0.012786383740603924
+397,0.011129582300782204,0.012871727347373962,0.042603909969329834,0.03256916254758835,169,0.012871727347373962
+398,0.009785549715161324,0.011759649962186813,0.03985002636909485,0.0268724225461483,169,0.011759649962186813
+399,0.009596102871000767,0.014139826409518719,0.04005825147032738,0.026860535144805908,169,0.014139826409518719
+400,0.011700201779603958,0.013475763611495495,0.03371121734380722,0.02359761670231819,169,0.013475763611495495
+401,0.010261986404657364,0.012031346559524536,0.04011647775769234,0.026644228026270866,169,0.012031346559524536
+402,0.011449193581938744,0.012839250266551971,0.04728670418262482,0.03784340247511864,169,0.012839250266551971
+403,0.011941477656364441,0.013127408921718597,0.03200003132224083,0.011941477656364441,169,0.013127408921718597
+404,0.014568226411938667,0.016187746077775955,0.03973543271422386,0.03033081255853176,169,0.016187746077775955
+405,0.010662689805030823,0.012342896312475204,0.04192105680704117,0.031320057809352875,169,0.012342896312475204
+406,0.011062847450375557,0.012749915011227131,0.03801323473453522,0.029659781605005264,169,0.012749915011227131
+407,0.011062847450375557,0.012949889525771141,0.037674058228731155,0.028814004734158516,169,0.012949889525771141
+408,0.010414063930511475,0.013115162029862404,0.03219809755682945,0.01804395206272602,169,0.013115162029862404
+409,0.01344795897603035,0.016309278085827827,0.04495489224791527,0.03290826082229614,169,0.016309278085827827
+410,0.013592495582997799,0.016851071268320084,0.04120953753590584,0.03554894030094147,169,0.016851071268320084
+411,0.01172932144254446,0.015955111011862755,0.05069011077284813,0.04260530322790146,169,0.015955111011862755
+412,0.010533158667385578,0.012732614763081074,0.04298386350274086,0.03557179123163223,169,0.012732614763081074
+413,0.012536444701254368,0.01508051436394453,0.044687606394290924,0.040021952241659164,169,0.01508051436394453
+414,0.009722290560603142,0.012455372139811516,0.03903847560286522,0.026695434004068375,169,0.012455372139811516
+415,0.011871467344462872,0.014389729127287865,0.03998595103621483,0.03251646086573601,169,0.014389729127287865
+416,0.01060844399034977,0.013991735875606537,0.047929998487234116,0.039323173463344574,169,0.013991735875606537
+417,0.01116905640810728,0.012517529539763927,0.03815656900405884,0.028111614286899567,169,0.012517529539763927
+418,0.010537550784647465,0.012231585569679737,0.03855698183178902,0.025055645033717155,169,0.012231585569679737
+419,0.01202941033989191,0.01332334615290165,0.049178726971149445,0.04547365754842758,169,0.01332334615290165
+420,0.011376433074474335,0.013042221777141094,0.03176899999380112,0.013143966905772686,169,0.013042221777141094
+421,0.01386694610118866,0.014952793717384338,0.03946322575211525,0.029929067939519882,169,0.014952793717384338
+422,0.009981540963053703,0.01385992020368576,0.038756947964429855,0.02637789398431778,169,0.01385992020368576
+423,0.011326564475893974,0.014314940199255943,0.038698531687259674,0.03102252632379532,169,0.014314940199255943
+424,0.01049918681383133,0.012104692868888378,0.03529524430632591,0.019101645797491074,169,0.012104692868888378
+425,0.012272350490093231,0.014401910826563835,0.03885762393474579,0.0272480770945549,169,0.014401910826563835
+426,0.01031485851854086,0.012696944177150726,0.03318814933300018,0.020270904526114464,169,0.012696944177150726
+427,0.00999896414577961,0.013950143940746784,0.04800792783498764,0.03748738020658493,169,0.013950143940746784
+428,0.013130404986441135,0.01482921838760376,0.05300518497824669,0.049231573939323425,169,0.01482921838760376
+429,0.011161500588059425,0.01347436010837555,0.05090436711907387,0.04675958305597305,169,0.01347436010837555
+430,0.010334083810448647,0.014153271913528442,0.03484718129038811,0.018939299508929253,169,0.014153271913528442
+431,0.012065279297530651,0.016462082043290138,0.04039688780903816,0.032666921615600586,169,0.016462082043290138
+432,0.012308320961892605,0.014306699857115746,0.04700513556599617,0.03915759548544884,169,0.014306699857115746
+433,0.012123226188123226,0.014597946777939796,0.04226088151335716,0.029868291690945625,169,0.014597946777939796
+434,0.011614962480962276,0.013494836166501045,0.04259425401687622,0.03403805196285248,169,0.013494836166501045
+435,0.010411379858851433,0.012479161843657494,0.04708386957645416,0.039323750883340836,169,0.012479161843657494
+436,0.010127785615622997,0.013992873020470142,0.057245511561632156,0.04705150052905083,169,0.013992873020470142
+437,0.010533158667385578,0.01322057843208313,0.04287748783826828,0.03527224436402321,169,0.01322057843208313
+438,0.011326202191412449,0.013846119865775108,0.04674619808793068,0.03973197564482689,169,0.013846119865775108
+439,0.010840436443686485,0.013305353000760078,0.03428425267338753,0.02106061950325966,169,0.013305353000760078
+440,0.010411379858851433,0.012962639331817627,0.049549706280231476,0.04245487228035927,169,0.012962639331817627
+441,0.011700201779603958,0.013162133283913136,0.034989647567272186,0.024864083155989647,169,0.013162133283913136
+442,0.012020773254334927,0.013480487279593945,0.05027369037270546,0.04323965311050415,169,0.013480487279593945
+443,0.010713888332247734,0.01235243771225214,0.03822498396039009,0.024227097630500793,169,0.01235243771225214
+444,0.012695809826254845,0.015164680778980255,0.035802166908979416,0.023154892027378082,169,0.015164680778980255
+445,0.010287892073392868,0.012952195480465889,0.04350863769650459,0.031178707256913185,169,0.012952195480465889
+446,0.011664236895740032,0.01460784487426281,0.03722401335835457,0.022541629150509834,169,0.01460784487426281
+447,0.011985601857304573,0.013121304102241993,0.04323851689696312,0.03795093670487404,169,0.013121304102241993
+448,0.01226156298071146,0.013901589438319206,0.03695835545659065,0.023168260231614113,169,0.013901589438319206
+449,0.011433697305619717,0.013366259634494781,0.03804056718945503,0.02975204773247242,169,0.013366259634494781
+450,0.012838225811719894,0.015198421664536,0.056335002183914185,0.053541671484708786,169,0.015198421664536
+451,0.011518187820911407,0.01434458326548338,0.03711710497736931,0.027630897238850594,169,0.01434458326548338
+452,0.011666446924209595,0.013746371492743492,0.05292442813515663,0.042540282011032104,169,0.013746371492743492
+453,0.012167887762188911,0.014110488817095757,0.03446318209171295,0.018688729032874107,169,0.014110488817095757
+454,0.011261088773608208,0.015848923474550247,0.05262729525566101,0.04289459437131882,169,0.015848923474550247
+455,0.013035432435572147,0.014874687418341637,0.04698437079787254,0.03863150626420975,169,0.014874687418341637
+456,0.01138075441122055,0.014898248016834259,0.050584178417921066,0.04214472323656082,169,0.014898248016834259
+457,0.01100219041109085,0.01265255082398653,0.03163490816950798,0.014218493364751339,169,0.01265255082398653
+458,0.01390835177153349,0.015758724883198738,0.048307765275239944,0.04122592508792877,169,0.015758724883198738
+459,0.009900936856865883,0.013090578839182854,0.0485294908285141,0.03718580678105354,169,0.013090578839182854
+460,0.010578181594610214,0.012716004624962807,0.037349846214056015,0.022684309631586075,169,0.012716004624962807
+461,0.01244143396615982,0.013857945799827576,0.03277789428830147,0.014670368283987045,169,0.013857945799827576
+462,0.01217289175838232,0.01535832416266203,0.04394792020320892,0.03992081433534622,169,0.01535832416266203
+463,0.01149698719382286,0.01251645851880312,0.04349060729146004,0.031623780727386475,169,0.01251645851880312
+464,0.011250641196966171,0.013202948495745659,0.032039791345596313,0.01667441800236702,169,0.013202948495745659
+465,0.012440490536391735,0.013187365606427193,0.03749479353427887,0.029777249321341515,169,0.013187365606427193
+466,0.011251366697251797,0.013508327305316925,0.03421378880739212,0.019331984221935272,169,0.013508327305316925
+467,0.01170630194246769,0.013742240145802498,0.0402403362095356,0.030547933652997017,169,0.013742240145802498
+468,0.012561491690576077,0.015599250793457031,0.04072604700922966,0.030703727155923843,169,0.015599250793457031
+469,0.01219884678721428,0.01357879675924778,0.043891943991184235,0.03385287523269653,169,0.01357879675924778
+470,0.009494186379015446,0.012200693599879742,0.041178394109010696,0.02875075489282608,169,0.012200693599879742
+471,0.012499282136559486,0.013820203952491283,0.04084892198443413,0.03379608690738678,169,0.013820203952491283
+472,0.011118012480437756,0.014307585544884205,0.040674395859241486,0.02801940031349659,169,0.014307585544884205
+473,0.01052606850862503,0.013284148648381233,0.03655902296304703,0.022960618138313293,169,0.013284148648381233
+474,0.01334916427731514,0.015826841816306114,0.041302766650915146,0.03303410857915878,169,0.015826841816306114
+475,0.011185373179614544,0.012519290670752525,0.03643612563610077,0.026945525780320168,169,0.012519290670752525
+476,0.011194635182619095,0.013152946718037128,0.03994535654783249,0.03009725920855999,169,0.013152946718037128
+477,0.012065279297530651,0.01622794196009636,0.04177310690283775,0.03511044755578041,169,0.01622794196009636
+478,0.010497553274035454,0.012781670317053795,0.04595334082841873,0.03482566028833389,169,0.012781670317053795
+479,0.01020148303359747,0.013986272737383842,0.04394415393471718,0.03378850966691971,169,0.013986272737383842
+480,0.012561491690576077,0.017036424949765205,0.0440656840801239,0.03537356108427048,169,0.017036424949765205
+481,0.010847714729607105,0.013209044933319092,0.03479795902967453,0.020611511543393135,169,0.013209044933319092
+482,0.014083867892622948,0.015957290306687355,0.0486031100153923,0.0447525680065155,169,0.015957290306687355
+483,0.011003854684531689,0.013646930456161499,0.042470622807741165,0.03002808429300785,169,0.013646930456161499
+484,0.011465747840702534,0.01291580032557249,0.03448352590203285,0.0247759148478508,169,0.01291580032557249
+485,0.012123226188123226,0.014213001355528831,0.04130512848496437,0.02816959284245968,169,0.014213001355528831
+486,0.01060036476701498,0.01324393879622221,0.04024256020784378,0.030539169907569885,169,0.01324393879622221
+487,0.01009734719991684,0.014226220548152924,0.03857142850756645,0.02588411048054695,169,0.014226220548152924
+488,0.010966521687805653,0.013614113442599773,0.04241935536265373,0.03536076471209526,169,0.013614113442599773
+489,0.012984336353838444,0.015226838178932667,0.04822791367769241,0.04345835745334625,169,0.015226838178932667
+490,0.011264560744166374,0.013342037796974182,0.0346875935792923,0.020903106778860092,169,0.013342037796974182
+491,0.013232775963842869,0.015678653493523598,0.043904103338718414,0.03891860693693161,169,0.015678653493523598
+492,0.011682662181556225,0.013284395448863506,0.03513486683368683,0.0248078852891922,169,0.013284395448863506
+493,0.011712696403265,0.014100313186645508,0.045555680990219116,0.04119318723678589,169,0.014100313186645508
+494,0.011545045301318169,0.013091130182147026,0.03629612550139427,0.021169286221265793,169,0.013091130182147026
+495,0.00986604206264019,0.011616474017500877,0.04121987521648407,0.02933608368039131,169,0.011616474017500877
+496,0.01114719919860363,0.013925781473517418,0.03816527500748634,0.027845658361911774,169,0.013925781473517418
+497,0.011110068298876286,0.014427614398300648,0.04978848248720169,0.04332207143306732,169,0.014427614398300648
+498,0.01191819179803133,0.013523724861443043,0.040370114147663116,0.03006620891392231,169,0.013523724861443043
+499,0.01132768951356411,0.012958891689777374,0.04938659816980362,0.038739945739507675,169,0.012958891689777374
+500,0.010293045081198215,0.012191682122647762,0.039953917264938354,0.029962992295622826,169,0.012191682122647762

splits/compute_chamfer_splits.py

@@ -0,0 +1,735 @@
+#!/usr/bin/env python3
+"""Compute STL-based Chamfer geometry splits for AhmedML.
+
+This is intentionally standalone so it can be copied to the machine that has
+the STL files. It expects a AhmedML-style directory layout:
+
+    DATA_ROOT/
+      run_1/ahmed_1.stl
+      run_2/ahmed_2.stl
+      ...
+
+Outputs:
+  - sampled point clouds cached as NPZ files
+  - chamfer_metrics.csv with nearest-neighbor and outlier scores
+  - chamfer_manifest.json with geometry_{train,val,test}
+  - optional chamfer_distance_matrix.npy, a symmetric NxN float32 matrix
+  - optional sparse train subsets when a base manifest with full_train exists
+
+Install dependencies on the data machine:
+
+    python -m pip install numpy scipy trimesh
+
+Example:
+
+    python compute_chamfer_splits.py \
+      --data-root /data/ahmedml \
+      --output-dir /data/ahmedml_chamfer \
+      --samples 4096 \
+      --workers 16 \
+      --base-manifest /path/to/ahmedml/splits/manifest.json
+"""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import hashlib
+import json
+import math
+import random
+import sys
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Iterable
+
+import numpy as np
+
+
+N_CASES = 500
+PUBLIC_RUN_IDS = list(range(1, N_CASES + 1))
+DEFAULT_TEST_FRACTION = 0.2
+DEFAULT_VAL_FRACTION = 0.1
+DEFAULT_SEED = 42
+cKDTree = None
+trimesh = None
+
+
+@dataclass(frozen=True)
+class RunFile:
+    run_id: int
+    stl_path: Path
+
+
+def case_id(run_id: int) -> str:
+    return f"run_{run_id}"
+
+
+def run_id(case: str) -> int:
+    if not case.startswith("run_"):
+        raise ValueError(f"bad case id: {case!r}")
+    return int(case.split("_", 1)[1])
+
+
+def require_dependencies() -> None:
+    global cKDTree, trimesh
+    try:
+        from scipy.spatial import cKDTree as scipy_ckdtree
+    except Exception as exc:  # pragma: no cover - dependency guard
+        raise SystemExit(
+            "Missing dependency scipy. Install with: python -m pip install numpy scipy trimesh"
+        ) from exc
+    try:
+        import trimesh as trimesh_module
+    except Exception as exc:  # pragma: no cover - dependency guard
+        raise SystemExit(
+            "Missing dependency trimesh. Install with: python -m pip install numpy scipy trimesh"
+        ) from exc
+    cKDTree = scipy_ckdtree
+    trimesh = trimesh_module
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Compute STL-surface Chamfer distances and AhmedML geometry splits.",
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
+    )
+    parser.add_argument(
+        "--data-root",
+        type=Path,
+        required=True,
+        help="Directory containing run_N/ahmed_N.stl files.",
+    )
+    parser.add_argument(
+        "--output-dir",
+        type=Path,
+        required=True,
+        help="Directory where matrices, metrics, and manifests will be written.",
+    )
+    parser.add_argument(
+        "--samples",
+        type=int,
+        default=4096,
+        help="Surface sample count per STL. 4096 is a practical first pass; 10000+ is better for final splits.",
+    )
+    parser.add_argument(
+        "--workers",
+        type=int,
+        default=8,
+        help="Thread workers used for pairwise nearest-neighbor queries.",
+    )
+    parser.add_argument(
+        "--sample-workers",
+        type=int,
+        default=1,
+        help="Thread workers used while loading and sampling STLs. Keep this low for large AhmedML files.",
+    )
+    parser.add_argument(
+        "--seed",
+        type=int,
+        default=DEFAULT_SEED,
+        help="Base random seed for deterministic surface sampling and split selection.",
+    )
+    parser.add_argument(
+        "--k-neighbors",
+        type=int,
+        default=10,
+        help="K used for the local-isolation geometry score.",
+    )
+    parser.add_argument(
+        "--test-fraction",
+        type=float,
+        default=DEFAULT_TEST_FRACTION,
+        help="Fraction held out as OOD test for geometry.",
+    )
+    parser.add_argument(
+        "--val-fraction",
+        type=float,
+        default=DEFAULT_VAL_FRACTION,
+        help="Overall validation fraction. Validation is sampled from the train-side pool.",
+    )
+    parser.add_argument(
+        "--score",
+        choices=["knn", "medoid", "mean"],
+        default="knn",
+        help="Score used to rank OOD geometry cases.",
+    )
+    parser.add_argument(
+        "--center",
+        choices=["none", "bbox", "centroid"],
+        default="none",
+        help="How to remove translation before Chamfer. Use none when STLs share a common coordinate frame.",
+    )
+    parser.add_argument(
+        "--scale-mode",
+        choices=["global_median_bbox", "per_mesh_bbox", "none"],
+        default="global_median_bbox",
+        help="How to scale coordinates before Chamfer. global_median_bbox keeps real relative vehicle size.",
+    )
+    parser.add_argument(
+        "--runs",
+        type=str,
+        default="public",
+        help=(
+            "Run IDs to process: public, all, or a comma/range expression like "
+            "1,2,10-20. For AhmedML, public and all both mean 1..500."
+        ),
+    )
+    parser.add_argument(
+        "--base-manifest",
+        type=Path,
+        default=None,
+        help=(
+            "Optional existing split manifest. If it contains full_train/full_val/full_test, "
+            "the script also writes geometry_medium/scarce/super_scarce splits."
+        ),
+    )
+    parser.add_argument(
+        "--force-resample",
+        action="store_true",
+        help="Ignore cached point clouds and resample all STLs.",
+    )
+    parser.add_argument(
+        "--force-matrix",
+        action="store_true",
+        help="Recompute the Chamfer matrix even if a compatible matrix already exists.",
+    )
+    parser.add_argument(
+        "--write-matrix",
+        action="store_true",
+        help="Write chamfer_distance_matrix.npy and its metadata JSON. Omitted by default to keep the split package lean.",
+    )
+    parser.add_argument(
+        "--allow-missing",
+        action="store_true",
+        help="Process the subset of requested runs whose STLs exist. Without this, missing STLs are an error.",
+    )
+    parser.add_argument(
+        "--write-csv-matrix",
+        action="store_true",
+        help="Also write chamfer_distance_matrix.csv from the in-memory matrix.",
+    )
+    return parser.parse_args()
+
+
+def parse_run_expression(expr: str) -> list[int]:
+    expr = expr.strip().lower()
+    if expr == "public":
+        return PUBLIC_RUN_IDS.copy()
+    if expr == "all":
+        return list(range(1, N_CASES + 1))
+
+    result: set[int] = set()
+    for token in expr.split(","):
+        token = token.strip()
+        if not token:
+            continue
+        if "-" in token:
+            start_s, end_s = token.split("-", 1)
+            start, end = int(start_s), int(end_s)
+            if start > end:
+                start, end = end, start
+            result.update(range(start, end + 1))
+        else:
+            result.add(int(token))
+    runs = sorted(result)
+    bad = [rid for rid in runs if rid < 1 or rid > N_CASES]
+    if bad:
+        raise SystemExit(f"Run IDs must be in 1..{N_CASES}; bad values: {bad}")
+    return runs
+
+
+def discover_files(data_root: Path, requested_runs: Iterable[int], allow_missing: bool) -> list[RunFile]:
+    files: list[RunFile] = []
+    missing: list[int] = []
+    for rid in requested_runs:
+        path = data_root / f"run_{rid}" / f"ahmed_{rid}.stl"
+        if path.exists() and path.stat().st_size > 0:
+            files.append(RunFile(rid, path))
+        else:
+            missing.append(rid)
+
+    if missing and not allow_missing:
+        preview = ", ".join(str(x) for x in missing[:20])
+        suffix = " ..." if len(missing) > 20 else ""
+        raise SystemExit(
+            f"Missing {len(missing)} requested STL files under {data_root}: {preview}{suffix}\n"
+            "Use --allow-missing to compute with the available subset."
+        )
+    if not files:
+        raise SystemExit(f"No STL files found under {data_root}")
+    return files
+
+
+def load_mesh(path: Path) -> "trimesh.Trimesh":
+    mesh = trimesh.load_mesh(path, process=False)
+    if isinstance(mesh, trimesh.Scene):
+        geometries = [g for g in mesh.geometry.values() if len(g.faces) > 0]
+        if not geometries:
+            raise ValueError(f"{path} did not contain any mesh geometry")
+        mesh = trimesh.util.concatenate(geometries)
+    if not isinstance(mesh, trimesh.Trimesh):
+        raise ValueError(f"{path} loaded as unsupported object: {type(mesh)!r}")
+    if len(mesh.faces) == 0:
+        raise ValueError(f"{path} has no faces")
+    return mesh
+
+
+def sample_mesh_surface(mesh: "trimesh.Trimesh", count: int, seed: int) -> np.ndarray:
+    """Area-sample points from a triangular mesh using a local RNG."""
+    rng = np.random.default_rng(seed)
+    areas = np.asarray(mesh.area_faces, dtype=np.float64)
+    total_area = float(np.sum(areas))
+    if not math.isfinite(total_area) or total_area <= 0.0:
+        raise ValueError("mesh surface area is zero or invalid")
+
+    face_indices = rng.choice(len(mesh.faces), size=count, replace=True, p=areas / total_area)
+    triangles = np.asarray(mesh.vertices[mesh.faces[face_indices]], dtype=np.float64)
+
+    u = rng.random(count)
+    v = rng.random(count)
+    outside = (u + v) > 1.0
+    u[outside] = 1.0 - u[outside]
+    v[outside] = 1.0 - v[outside]
+    points = triangles[:, 0] + u[:, None] * (triangles[:, 1] - triangles[:, 0]) + v[:, None] * (
+        triangles[:, 2] - triangles[:, 0]
+    )
+    return np.asarray(points, dtype=np.float32)
+
+
+def cache_path(cache_dir: Path, run: RunFile, samples: int, seed: int) -> Path:
+    source = f"{run.stl_path.resolve()}:{run.stl_path.stat().st_size}:{samples}:{seed}:{run.run_id}"
+    digest = hashlib.sha256(source.encode("utf-8")).hexdigest()[:16]
+    return cache_dir / f"run_{run.run_id:03d}_samples_{samples}_{digest}.npz"
+
+
+def sample_one(run: RunFile, cache_dir: Path, samples: int, seed: int, force: bool) -> tuple[int, np.ndarray, np.ndarray, np.ndarray]:
+    cache = cache_path(cache_dir, run, samples, seed)
+    if cache.exists() and not force:
+        data = np.load(cache)
+        points = np.asarray(data["points"], dtype=np.float32)
+        bbox_min = np.asarray(data["bbox_min"], dtype=np.float32)
+        bbox_max = np.asarray(data["bbox_max"], dtype=np.float32)
+        if points.shape == (samples, 3):
+            return run.run_id, points, bbox_min, bbox_max
+
+    mesh = load_mesh(run.stl_path)
+    points = sample_mesh_surface(mesh, samples, seed + run.run_id)
+    bbox_min = np.asarray(mesh.bounds[0], dtype=np.float32)
+    bbox_max = np.asarray(mesh.bounds[1], dtype=np.float32)
+    np.savez_compressed(
+        cache,
+        run_id=np.asarray(run.run_id, dtype=np.int32),
+        points=points,
+        bbox_min=bbox_min,
+        bbox_max=bbox_max,
+        source=str(run.stl_path),
+        samples=np.asarray(samples, dtype=np.int32),
+        seed=np.asarray(seed, dtype=np.int32),
+    )
+    return run.run_id, points, bbox_min, bbox_max
+
+
+def sample_point_clouds(
+    runs: list[RunFile],
+    cache_dir: Path,
+    samples: int,
+    seed: int,
+    workers: int,
+    force: bool,
+) -> tuple[list[int], list[np.ndarray], np.ndarray, np.ndarray]:
+    cache_dir.mkdir(parents=True, exist_ok=True)
+    started = time.time()
+    print(f"Sampling/caching {len(runs)} STL point clouds with {samples} points each...")
+
+    outputs: list[tuple[int, np.ndarray, np.ndarray, np.ndarray]] = []
+    with ThreadPoolExecutor(max_workers=max(1, workers)) as pool:
+        futures = [pool.submit(sample_one, run, cache_dir, samples, seed, force) for run in runs]
+        for idx, future in enumerate(as_completed(futures), start=1):
+            outputs.append(future.result())
+            if idx == len(futures) or idx % 25 == 0:
+                print(f"  sampled {idx}/{len(futures)}")
+
+    outputs.sort(key=lambda x: x[0])
+    run_ids = [x[0] for x in outputs]
+    clouds = [x[1] for x in outputs]
+    bbox_min = np.stack([x[2] for x in outputs])
+    bbox_max = np.stack([x[3] for x in outputs])
+    print(f"Sampling complete in {time.time() - started:.1f}s")
+    return run_ids, clouds, bbox_min, bbox_max
+
+
+def normalize_clouds(
+    clouds: list[np.ndarray],
+    bbox_min: np.ndarray,
+    bbox_max: np.ndarray,
+    center: str,
+    scale_mode: str,
+) -> tuple[list[np.ndarray], dict[str, float | str]]:
+    result: list[np.ndarray] = []
+    bbox_diag = np.linalg.norm(bbox_max - bbox_min, axis=1)
+    global_scale = float(np.median(bbox_diag))
+    if not math.isfinite(global_scale) or global_scale <= 0:
+        global_scale = 1.0
+
+    for idx, points in enumerate(clouds):
+        pts = points.astype(np.float32, copy=True)
+        if center == "bbox":
+            pts -= ((bbox_min[idx] + bbox_max[idx]) * 0.5).astype(np.float32)
+        elif center == "centroid":
+            pts -= pts.mean(axis=0, keepdims=True)
+
+        if scale_mode == "global_median_bbox":
+            scale = global_scale
+        elif scale_mode == "per_mesh_bbox":
+            scale = float(bbox_diag[idx]) if bbox_diag[idx] > 0 else 1.0
+        else:
+            scale = 1.0
+        pts /= np.float32(scale)
+        result.append(pts)
+
+    metadata: dict[str, float | str] = {
+        "center": center,
+        "scale_mode": scale_mode,
+        "global_median_bbox_diag": global_scale,
+    }
+    return result, metadata
+
+
+def pair_chamfer_rms(i: int, j: int, clouds: list[np.ndarray], trees: list[cKDTree]) -> tuple[int, int, float]:
+    a_to_b, _ = trees[j].query(clouds[i], k=1)
+    b_to_a, _ = trees[i].query(clouds[j], k=1)
+    chamfer = float(np.sqrt(0.5 * (np.mean(a_to_b * a_to_b) + np.mean(b_to_a * b_to_a))))
+    return i, j, chamfer
+
+
+def matrix_metadata_path(output_dir: Path) -> Path:
+    return output_dir / "chamfer_distance_matrix.meta.json"
+
+
+def matrix_is_compatible(output_dir: Path, run_ids: list[int], args: argparse.Namespace) -> bool:
+    matrix_path = output_dir / "chamfer_distance_matrix.npy"
+    meta_path = matrix_metadata_path(output_dir)
+    if not matrix_path.exists() or not meta_path.exists():
+        return False
+    try:
+        meta = json.loads(meta_path.read_text(encoding="utf-8"))
+    except Exception:
+        return False
+    return (
+        meta.get("run_ids") == run_ids
+        and meta.get("samples") == args.samples
+        and meta.get("seed") == args.seed
+        and meta.get("center") == args.center
+        and meta.get("scale_mode") == args.scale_mode
+        and meta.get("metric") == "symmetric_chamfer_rms"
+    )
+
+
+def compute_chamfer_matrix(
+    run_ids: list[int],
+    clouds: list[np.ndarray],
+    output_dir: Path,
+    args: argparse.Namespace,
+    normalization_metadata: dict[str, float | str],
+) -> np.ndarray:
+    matrix_path = output_dir / "chamfer_distance_matrix.npy"
+    if matrix_is_compatible(output_dir, run_ids, args) and not args.force_matrix:
+        print(f"Loading existing compatible matrix: {matrix_path}")
+        return np.load(matrix_path)
+
+    n = len(clouds)
+    print(f"Building {n} KD trees...")
+    trees = [cKDTree(points) for points in clouds]
+    matrix = np.zeros((n, n), dtype=np.float32)
+    pairs = [(i, j) for i in range(n) for j in range(i + 1, n)]
+    started = time.time()
+    print(f"Computing {len(pairs)} pairwise symmetric Chamfer RMS distances...")
+
+    with ThreadPoolExecutor(max_workers=max(1, args.workers)) as pool:
+        futures = [pool.submit(pair_chamfer_rms, i, j, clouds, trees) for i, j in pairs]
+        for done, future in enumerate(as_completed(futures), start=1):
+            i, j, value = future.result()
+            matrix[i, j] = matrix[j, i] = np.float32(value)
+            if done == len(futures) or done % 1000 == 0:
+                elapsed = time.time() - started
+                rate = done / elapsed if elapsed > 0 else 0.0
+                remaining = (len(futures) - done) / rate if rate > 0 else float("nan")
+                print(
+                    f"  pairs {done}/{len(futures)} "
+                    f"({100 * done / len(futures):5.1f}%), ETA {remaining / 60:5.1f} min"
+                )
+
+    if args.write_matrix:
+        np.save(matrix_path, matrix)
+        metadata = {
+            "run_ids": run_ids,
+            "samples": args.samples,
+            "seed": args.seed,
+            "center": args.center,
+            "scale_mode": args.scale_mode,
+            "metric": "symmetric_chamfer_rms",
+            "created_unix_time": time.time(),
+            **normalization_metadata,
+        }
+        matrix_metadata_path(output_dir).write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8")
+        print(f"Matrix written: {matrix_path}")
+    return matrix
+
+
+def write_csv_matrix(path: Path, run_ids: list[int], matrix: np.ndarray) -> None:
+    with path.open("w", encoding="utf-8", newline="") as f:
+        writer = csv.writer(f)
+        writer.writerow(["run", *[case_id(rid) for rid in run_ids]])
+        for rid, row in zip(run_ids, matrix):
+            writer.writerow([case_id(rid), *[f"{float(x):.8g}" for x in row]])
+
+
+def metric_values(run_ids: list[int], matrix: np.ndarray, k_neighbors: int) -> tuple[list[dict[str, float | int]], dict[int, float]]:
+    n = len(run_ids)
+    if n < 2:
+        raise SystemExit("At least two STL files are required to compute Chamfer metrics")
+    k = min(max(1, k_neighbors), n - 1)
+    means = matrix.sum(axis=1) / (n - 1)
+    medoid_index = int(np.argmin(means))
+    medoid_run = run_ids[medoid_index]
+    rows: list[dict[str, float | int]] = []
+    knn_scores: dict[int, float] = {}
+
+    for idx, rid in enumerate(run_ids):
+        nonself = np.delete(matrix[idx], idx)
+        sorted_dist = np.sort(nonself)
+        nearest = float(sorted_dist[0])
+        knn_mean = float(np.mean(sorted_dist[:k]))
+        mean_all = float(means[idx])
+        medoid_distance = float(matrix[idx, medoid_index])
+        knn_scores[rid] = knn_mean
+        rows.append(
+            {
+                "run": rid,
+                "nearest_neighbor_chamfer": nearest,
+                f"mean_{k}_nn_chamfer": knn_mean,
+                "mean_all_chamfer": mean_all,
+                "medoid_chamfer": medoid_distance,
+                "medoid_run": medoid_run,
+            }
+        )
+    return rows, knn_scores
+
+
+def score_map(
+    run_ids: list[int],
+    matrix: np.ndarray,
+    metrics: list[dict[str, float | int]],
+    score_name: str,
+    k_neighbors: int,
+) -> dict[int, float]:
+    if score_name == "knn":
+        key = f"mean_{min(max(1, k_neighbors), len(run_ids) - 1)}_nn_chamfer"
+    elif score_name == "medoid":
+        key = "medoid_chamfer"
+    else:
+        key = "mean_all_chamfer"
+    return {int(row["run"]): float(row[key]) for row in metrics}
+
+
+def split_pool(pool: list[int], val_fraction_of_pool: float, seed: int, salt: str) -> tuple[list[int], list[int]]:
+    rng_seed = hashlib.sha256(f"{seed}:{salt}".encode("utf-8")).digest()[:8]
+    rng = random.Random(int.from_bytes(rng_seed, "big"))
+    shuffled = pool.copy()
+    rng.shuffle(shuffled)
+    n_val = round(len(pool) * val_fraction_of_pool)
+    val = sorted(shuffled[:n_val])
+    train = sorted(shuffled[n_val:])
+    return train, val
+
+
+def ranked_ood_split(
+    scores: dict[int, float],
+    test_fraction: float,
+    val_fraction: float,
+    seed: int,
+    salt: str,
+) -> tuple[list[int], list[int], list[int]]:
+    ranked = sorted(scores, key=lambda rid: (scores[rid], rid))
+    n_test = round(len(ranked) * test_fraction)
+    test = sorted(ranked[-n_test:])
+    pool = sorted(ranked[:-n_test])
+    val_fraction_of_pool = val_fraction / (1.0 - test_fraction)
+    train, val = split_pool(pool, val_fraction_of_pool, seed, salt)
+    return train, val, test
+
+
+def make_case_ids(values: Iterable[int]) -> list[str]:
+    return [case_id(rid) for rid in sorted(values)]
+
+
+def farthest_order(pool: list[int], run_to_index: dict[int, int], matrix: np.ndarray, seed: int) -> list[int]:
+    if not pool:
+        return []
+
+    mean_dist = {
+        rid: float(np.mean([matrix[run_to_index[rid], run_to_index[other]] for other in pool if other != rid]))
+        for rid in pool
+    }
+    first = max(pool, key=lambda rid: (mean_dist[rid], -rid))
+    selected = [first]
+    remaining = [rid for rid in pool if rid != first]
+
+    rng_seed = hashlib.sha256(f"{seed}:geometry_sparse_order".encode("utf-8")).digest()[:8]
+    rng = random.Random(int.from_bytes(rng_seed, "big"))
+    tie_break = {rid: rng.random() for rid in pool}
+
+    while remaining:
+        next_rid = max(
+            remaining,
+            key=lambda rid: (
+                min(matrix[run_to_index[rid], run_to_index[chosen]] for chosen in selected),
+                tie_break[rid],
+            ),
+        )
+        selected.append(next_rid)
+        remaining.remove(next_rid)
+    return selected
+
+
+def load_base_manifest(path: Path | None) -> dict[str, list[str]]:
+    if path is None:
+        candidate = Path(__file__).resolve().parents[1] / "splits" / "manifest.json"
+        if not candidate.exists():
+            return {}
+        path = candidate
+    if not path.exists():
+        raise SystemExit(f"Base manifest does not exist: {path}")
+    return json.loads(path.read_text(encoding="utf-8"))
+
+
+def write_metrics_csv(path: Path, metrics: list[dict[str, float | int]], scores: dict[int, float]) -> None:
+    fieldnames = list(metrics[0].keys()) + ["ood_score"]
+    with path.open("w", encoding="utf-8", newline="") as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        for row in metrics:
+            out = dict(row)
+            out["ood_score"] = scores[int(row["run"])]
+            writer.writerow(out)
+
+
+def build_manifest(
+    run_ids: list[int],
+    matrix: np.ndarray,
+    scores: dict[int, float],
+    args: argparse.Namespace,
+) -> dict[str, list[str]]:
+    train, val, test = ranked_ood_split(
+        scores,
+        test_fraction=args.test_fraction,
+        val_fraction=args.val_fraction,
+        seed=args.seed,
+        salt="geometry_val_selection",
+    )
+    manifest: dict[str, list[str]] = {
+        "geometry_train": make_case_ids(train),
+        "geometry_val": make_case_ids(val),
+        "geometry_test": make_case_ids(test),
+    }
+
+    base = load_base_manifest(args.base_manifest)
+    required = {"full_train", "full_val", "full_test"}
+    if not required <= set(base):
+        return manifest
+
+    available = set(run_ids)
+    full_train = [run_id(cid) for cid in base["full_train"] if run_id(cid) in available]
+    if len(full_train) < 20:
+        return manifest
+
+    run_to_index = {rid: idx for idx, rid in enumerate(run_ids)}
+    order = farthest_order(full_train, run_to_index, matrix, args.seed)
+    medium = round(len(order) / 3)
+    scarce = round(len(order) / 6)
+    super_scarce = max(1, round(len(order) / 36))
+    sparse_sets = {
+        "geometry_medium": sorted(order[:medium]),
+        "geometry_scarce": sorted(order[:scarce]),
+        "geometry_super_scarce": sorted(order[:super_scarce]),
+    }
+    for name, ids in sparse_sets.items():
+        manifest[f"{name}_train"] = make_case_ids(ids)
+        manifest[f"{name}_val"] = [cid for cid in base["full_val"] if run_id(cid) in available]
+        manifest[f"{name}_test"] = [cid for cid in base["full_test"] if run_id(cid) in available]
+    manifest["geometry_sparse_order"] = make_case_ids(order)
+    return manifest
+
+
+def summarize_split(name: str, manifest: dict[str, list[str]]) -> str:
+    return (
+        f"{name}: "
+        f"train={len(manifest.get(name + '_train', []))}, "
+        f"val={len(manifest.get(name + '_val', []))}, "
+        f"test={len(manifest.get(name + '_test', []))}"
+    )
+
+
+def main() -> None:
+    args = parse_args()
+    require_dependencies()
+    args.data_root = args.data_root.expanduser().resolve()
+    args.output_dir = args.output_dir.expanduser().resolve()
+    args.output_dir.mkdir(parents=True, exist_ok=True)
+
+    requested_runs = parse_run_expression(args.runs)
+    run_files = discover_files(args.data_root, requested_runs, args.allow_missing)
+    print(f"Found {len(run_files)} STL files under {args.data_root}")
+
+    run_ids, raw_clouds, bbox_min, bbox_max = sample_point_clouds(
+        run_files,
+        cache_dir=args.output_dir / "point_cloud_cache",
+        samples=args.samples,
+        seed=args.seed,
+        workers=args.sample_workers,
+        force=args.force_resample,
+    )
+    clouds, normalization_metadata = normalize_clouds(
+        raw_clouds,
+        bbox_min,
+        bbox_max,
+        center=args.center,
+        scale_mode=args.scale_mode,
+    )
+    matrix = compute_chamfer_matrix(run_ids, clouds, args.output_dir, args, normalization_metadata)
+    if args.write_csv_matrix:
+        write_csv_matrix(args.output_dir / "chamfer_distance_matrix.csv", run_ids, matrix)
+
+    metrics, _knn_scores = metric_values(run_ids, matrix, args.k_neighbors)
+    scores = score_map(run_ids, matrix, metrics, args.score, args.k_neighbors)
+    write_metrics_csv(args.output_dir / "chamfer_metrics.csv", metrics, scores)
+
+    manifest = build_manifest(run_ids, matrix, scores, args)
+    manifest_path = args.output_dir / "chamfer_manifest.json"
+    manifest_path.write_text(json.dumps(manifest, indent=2) + "\n", encoding="utf-8")
+
+    print()
+    print("Chamfer split summary")
+    print("=" * 60)
+    print(f"Runs: {len(run_ids)}")
+    print(f"Metric: symmetric Chamfer RMS; score={args.score}")
+    print(f"Metrics: {args.output_dir / 'chamfer_metrics.csv'}")
+    if args.write_matrix:
+        print(f"Matrix:  {args.output_dir / 'chamfer_distance_matrix.npy'}")
+    else:
+        print("Matrix:  not written; pass --write-matrix to save the full NPY")
+    print(f"Manifest: {manifest_path}")
+    print("  " + summarize_split("geometry", manifest))
+    for prefix in ["geometry_medium", "geometry_scarce", "geometry_super_scarce"]:
+        if f"{prefix}_train" in manifest:
+            print("  " + summarize_split(prefix, manifest))
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except KeyboardInterrupt:
+        sys.exit("Interrupted")

splits/compute_image_metrics.py

@@ -0,0 +1,122 @@
+"""Compute AhmedML image-derived wake scores from downloaded UxMean PNGs.
+
+The metric is intentionally simple and reproducible: for each run, it reads the
+Y-4 centreline/near-centreline slice and three near-base X slices, crops out the
+upper background-dominated band, and averages a color-intensity score over the
+lower flow region. The resulting score is used for the `image_wake` OOD split.
+"""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import os
+from pathlib import Path
+
+import numpy as np
+from PIL import Image
+
+
+N_CASES = 500
+RUN_IDS = list(range(1, N_CASES + 1))
+SCRIPT_DIR = Path(__file__).resolve().parent
+REPOSITORY_ROOT = SCRIPT_DIR.parent
+DATA_DIR = SCRIPT_DIR
+DEFAULT_ASSET_ROOT = Path(os.environ.get("AHMEDML_ASSET_ROOT", REPOSITORY_ROOT.parent / "ahmedml_hf_assets"))
+CENTERLINE_Y_INDEX = 4
+NEAR_WAKE_X_INDICES = (14, 15, 16)
+
+
+def default_data_root() -> Path:
+    return DEFAULT_ASSET_ROOT if DEFAULT_ASSET_ROOT.exists() else DATA_DIR
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Compute AhmedML UxMean image wake metrics.")
+    parser.add_argument("--data-root", type=Path, default=default_data_root(), help="Directory containing run_*/images/UxMean PNGs.")
+    parser.add_argument("--output", type=Path, default=DATA_DIR / "image_metrics.csv", help="CSV output path.")
+    parser.add_argument("--min-images", type=int, default=4, help="Minimum targeted UxMean PNGs required for an observed score.")
+    return parser.parse_args()
+
+
+def _valid_png(path: Path) -> bool:
+    if path.name.startswith("._"):
+        return False
+    try:
+        with path.open("rb") as f:
+            return f.read(8) == b"\x89PNG\r\n\x1a\n"
+    except OSError:
+        return False
+
+
+def _image_score(path: Path) -> float | None:
+    if not _valid_png(path):
+        return None
+    try:
+        with Image.open(path) as img:
+            rgb = np.asarray(img.convert("RGB"), dtype=np.float32) / 255.0
+    except Exception:
+        return None
+
+    height, width, _ = rgb.shape
+    # The upper band in AhmedML UxMean slices is mostly uniform far-field color.
+    # The lower central region contains the visible body/wake/ground structure.
+    crop = rgb[int(0.30 * height):int(0.96 * height), int(0.08 * width):int(0.92 * width)]
+    brightness = crop.mean(axis=2)
+    chroma = crop.max(axis=2) - crop.min(axis=2)
+    warm = np.clip(crop[:, :, 0] - crop[:, :, 2], 0.0, 1.0)
+    return float(0.45 * brightness.mean() + 0.35 * chroma.mean() + 0.20 * warm.mean())
+
+
+def uxmean_path(data_root: Path, run: int, axis: str, index: int) -> Path:
+    return data_root / f"run_{run}" / "images" / "UxMean" / f"run_{run}-slice-UMean-0-{axis}-{index}.png"
+
+
+def target_uxmean_images(data_root: Path, run: int) -> list[tuple[str, Path]]:
+    return [
+        (f"Y-{CENTERLINE_Y_INDEX}", uxmean_path(data_root, run, "Y", CENTERLINE_Y_INDEX)),
+        *[(f"X-{index}", uxmean_path(data_root, run, "X", index)) for index in NEAR_WAKE_X_INDICES],
+    ]
+
+
+def run_image_score(data_root: Path, run: int, min_images: int) -> tuple[float, int]:
+    scores: dict[str, float] = {}
+    for label, path in target_uxmean_images(data_root, run):
+        score = _image_score(path)
+        if score is not None:
+            scores[label] = score
+    if len(scores) < min_images:
+        return float("nan"), len(scores)
+    y_score = scores.get(f"Y-{CENTERLINE_Y_INDEX}")
+    x_scores = [scores[f"X-{index}"] for index in NEAR_WAKE_X_INDICES if f"X-{index}" in scores]
+    if y_score is None or len(x_scores) != len(NEAR_WAKE_X_INDICES):
+        return float("nan"), len(scores)
+    return float(0.5 * y_score + 0.5 * np.mean(x_scores)), len(scores)
+
+
+def main() -> None:
+    args = parse_args()
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+    with args.output.open("w", encoding="utf-8", newline="") as f:
+        writer = csv.DictWriter(
+            f,
+            fieldnames=["run", "image_wake_score", "image_wake_observed", "uxmean_images", "uxmean_slices"],
+        )
+        writer.writeheader()
+        for run in RUN_IDS:
+            score, count = run_image_score(args.data_root, run, args.min_images)
+            observed = not np.isnan(score)
+            writer.writerow(
+                {
+                    "run": run,
+                    "image_wake_score": "" if not observed else score,
+                    "image_wake_observed": str(observed).lower(),
+                    "uxmean_images": count,
+                    "uxmean_slices": "Y-4;X-14;X-15;X-16",
+                }
+            )
+    print(f"Wrote {args.output}")
+
+
+if __name__ == "__main__":
+    main()

splits/create_example_figures.py

@@ -0,0 +1,251 @@
+"""Create low/high example figures for AhmedML split report."""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import os
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import numpy as np
+from PIL import Image
+
+
+SCRIPT_DIR = Path(__file__).resolve().parent
+REPOSITORY_ROOT = SCRIPT_DIR.parent
+DATA_DIR = SCRIPT_DIR
+DEFAULT_ASSET_ROOT = Path(os.environ.get("AHMEDML_ASSET_ROOT", REPOSITORY_ROOT.parent / "ahmedml_hf_assets"))
+DOCS_DIR = SCRIPT_DIR
+
+
+def default_asset_root() -> Path:
+    return DEFAULT_ASSET_ROOT if DEFAULT_ASSET_ROOT.exists() else DATA_DIR
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Create AhmedML report example figures.")
+    parser.add_argument("--data-root", type=Path, default=DATA_DIR, help="Directory containing force and metric CSV files.")
+    parser.add_argument("--force-root", type=Path, default=REPOSITORY_ROOT, help="Directory containing force_mom_all.csv.")
+    parser.add_argument("--asset-root", type=Path, default=default_asset_root(), help="Directory containing run_*/ STL and PNG assets.")
+    parser.add_argument("--output-dir", type=Path, default=DOCS_DIR, help="Directory for PNG figures.")
+    parser.add_argument("--max-stl-points", type=int, default=140000, help="Maximum STL vertices plotted per run.")
+    return parser.parse_args()
+
+
+def clean_row(row: dict[str, str]) -> dict[str, str]:
+    return {key.strip(): value.strip() for key, value in row.items() if key is not None}
+
+
+def load_force(force_root: Path) -> dict[int, dict[str, float]]:
+    candidates = [
+        force_root / "force_mom_all.csv",
+        force_root / "data" / "force_mom_all.csv",
+        REPOSITORY_ROOT / "force_mom_all.csv",
+        REPOSITORY_ROOT / "data" / "force_mom_all.csv",
+    ]
+    force_path = next((path for path in candidates if path.exists()), candidates[0])
+    with force_path.open(encoding="utf-8-sig", newline="") as f:
+        return {
+            int(clean_row(row)["run"]): {
+                "cd": float(clean_row(row)["cd"]),
+                "cl": float(clean_row(row)["cl"]),
+            }
+            for row in csv.DictReader(f)
+        }
+
+
+def load_metric(path: Path, column: str) -> dict[int, float]:
+    with path.open(encoding="utf-8-sig", newline="") as f:
+        return {int(row["run"]): float(row[column]) for row in csv.DictReader(f)}
+
+
+def low_high(scores: dict[int, float]) -> tuple[int, int]:
+    ordered = sorted(scores, key=lambda run: (scores[run], run))
+    return ordered[0], ordered[-1]
+
+
+def valid_png(path: Path) -> bool:
+    try:
+        with path.open("rb") as f:
+            return f.read(8) == b"\x89PNG\r\n\x1a\n"
+    except OSError:
+        return False
+
+
+def read_rgb(path: Path) -> np.ndarray:
+    if not valid_png(path):
+        raise ValueError(f"Not a valid PNG: {path}")
+    with Image.open(path) as img:
+        return np.asarray(img.convert("RGB"), dtype=np.float32) / 255.0
+
+
+def scored_crop(path: Path) -> np.ndarray:
+    rgb = read_rgb(path)
+    height, width, _ = rgb.shape
+    return rgb[int(0.30 * height):int(0.96 * height), int(0.08 * width):int(0.92 * width)]
+
+
+def image_score(path: Path) -> float:
+    crop = scored_crop(path)
+    brightness = crop.mean(axis=2)
+    chroma = crop.max(axis=2) - crop.min(axis=2)
+    warm = np.clip(crop[:, :, 0] - crop[:, :, 2], 0.0, 1.0)
+    return float(0.45 * brightness.mean() + 0.35 * chroma.mean() + 0.20 * warm.mean())
+
+
+def uxmean_path(data_root: Path, run: int, axis: str, index: int) -> Path:
+    return data_root / f"run_{run}" / "images" / "UxMean" / f"run_{run}-slice-UMean-0-{axis}-{index}.png"
+
+
+def display_crop(path: Path, axis: str) -> np.ndarray:
+    rgb = read_rgb(path)
+    height, width, _ = rgb.shape
+    if axis == "Y":
+        return rgb[int(0.34 * height):int(0.98 * height), :]
+    return rgb[int(0.20 * height):int(0.98 * height), int(0.04 * width):int(0.96 * width)]
+
+
+def make_wake_examples(data_root: Path, asset_root: Path, output_dir: Path, force: dict[int, dict[str, float]]) -> None:
+    scores = load_metric(data_root / "image_metrics.csv", "image_wake_score")
+    low_run, high_run = low_high(scores)
+    slices = [("Y", 4, "Y-4 centreline wake"), ("X", 14, "X-14 near-base wake")]
+
+    fig, axes = plt.subplots(len(slices), 2, figsize=(10.8, 5.8), constrained_layout=True)
+    if len(slices) == 1:
+        axes = np.asarray([axes])
+    for row, (axis, index, slice_label) in enumerate(slices):
+        for col, (label, run) in enumerate([("Low image-wake score", low_run), ("High image-wake score", high_run)]):
+            ax = axes[row, col]
+            ax.imshow(display_crop(uxmean_path(asset_root, run, axis, index), axis))
+            ax.set_axis_off()
+            ax.set_title(
+                f"{label}\n"
+                f"run_{run}, {slice_label}, score={scores[run]:.4f}, "
+                f"Cd={force[run]['cd']:.4f}, Cl={force[run]['cl']:.4f}",
+                fontsize=9,
+            )
+    fig.suptitle("Image-wake split examples from UxMean Y-4 and near-base X-slice PNGs", fontsize=12)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    fig.savefig(output_dir / "wake_score_examples.png", dpi=180)
+    plt.close(fig)
+
+
+def load_binary_stl_vertices(path: Path) -> np.ndarray:
+    size = path.stat().st_size
+    with path.open("rb") as f:
+        f.read(80)
+        count_data = f.read(4)
+        if len(count_data) != 4:
+            raise ValueError(f"Malformed STL: {path}")
+        face_count = int(np.frombuffer(count_data, dtype="<u4")[0])
+        expected_size = 84 + face_count * 50
+        if expected_size != size:
+            raise ValueError(f"Expected binary STL size {expected_size}, got {size}: {path}")
+        dtype = np.dtype(
+            [
+                ("normal", "<f4", (3,)),
+                ("vertices", "<f4", (3, 3)),
+                ("attribute", "<u2"),
+            ]
+        )
+        data = np.fromfile(f, dtype=dtype, count=face_count)
+    return np.asarray(data["vertices"].reshape(-1, 3), dtype=np.float32)
+
+
+def sample_vertices(vertices: np.ndarray, max_points: int, seed: int) -> np.ndarray:
+    if len(vertices) <= max_points:
+        return vertices
+    rng = np.random.default_rng(seed)
+    idx = rng.choice(len(vertices), size=max_points, replace=False)
+    return vertices[idx]
+
+
+def setup_projection_axis(ax, all_vertices: np.ndarray, dims: tuple[int, int], title: str) -> None:
+    x = all_vertices[:, dims[0]]
+    y = all_vertices[:, dims[1]]
+    pad_x = 0.04 * max(1e-9, float(x.max() - x.min()))
+    pad_y = 0.08 * max(1e-9, float(y.max() - y.min()))
+    ax.set_xlim(float(x.min() - pad_x), float(x.max() + pad_x))
+    ax.set_ylim(float(y.min() - pad_y), float(y.max() + pad_y))
+    ax.set_aspect("equal", adjustable="box")
+    ax.set_title(title, fontsize=10)
+    ax.grid(True, color="#e1e6eb", lw=0.6)
+    ax.spines["top"].set_visible(False)
+    ax.spines["right"].set_visible(False)
+
+
+def scatter_projection(ax, vertices: np.ndarray, dims: tuple[int, int], color: str, alpha: float, label: str | None = None) -> None:
+    ax.scatter(vertices[:, dims[0]], vertices[:, dims[1]], s=0.18, color=color, alpha=alpha, linewidth=0, label=label)
+
+
+def make_geometry_examples(data_root: Path, asset_root: Path, output_dir: Path, force: dict[int, dict[str, float]], max_points: int) -> None:
+    scores = load_metric(data_root / "chamfer_metrics.csv", "ood_score")
+    low_run, high_run = low_high(scores)
+    low_vertices = sample_vertices(load_binary_stl_vertices(asset_root / f"run_{low_run}" / f"ahmed_{low_run}.stl"), max_points, low_run)
+    high_vertices = sample_vertices(load_binary_stl_vertices(asset_root / f"run_{high_run}" / f"ahmed_{high_run}.stl"), max_points, high_run)
+    all_vertices = np.vstack([low_vertices, high_vertices])
+
+    low_color = "#6b7280"
+    high_color = "#087f8c"
+    fig, axes = plt.subplots(2, 2, figsize=(10.4, 6.2), constrained_layout=True)
+
+    setup_projection_axis(axes[0, 0], all_vertices, (0, 2), "Low geometry score: side view")
+    scatter_projection(axes[0, 0], low_vertices, (0, 2), low_color, 0.08)
+    axes[0, 0].set_xlabel("x")
+    axes[0, 0].set_ylabel("z")
+    axes[0, 0].text(
+        0.02,
+        0.96,
+        f"run_{low_run}\nscore={scores[low_run]:.5f}\nCd={force[low_run]['cd']:.4f}, Cl={force[low_run]['cl']:.4f}",
+        transform=axes[0, 0].transAxes,
+        va="top",
+        fontsize=8,
+        bbox={"facecolor": "white", "edgecolor": "#d8dee6", "alpha": 0.85, "pad": 3},
+    )
+
+    setup_projection_axis(axes[0, 1], all_vertices, (0, 2), "High geometry score: side view")
+    scatter_projection(axes[0, 1], high_vertices, (0, 2), high_color, 0.08)
+    axes[0, 1].set_xlabel("x")
+    axes[0, 1].set_ylabel("z")
+    axes[0, 1].text(
+        0.02,
+        0.96,
+        f"run_{high_run}\nscore={scores[high_run]:.5f}\nCd={force[high_run]['cd']:.4f}, Cl={force[high_run]['cl']:.4f}",
+        transform=axes[0, 1].transAxes,
+        va="top",
+        fontsize=8,
+        bbox={"facecolor": "white", "edgecolor": "#d8dee6", "alpha": 0.85, "pad": 3},
+    )
+
+    setup_projection_axis(axes[1, 0], all_vertices, (0, 1), "Overlay: top view")
+    scatter_projection(axes[1, 0], low_vertices, (0, 1), low_color, 0.06, "low")
+    scatter_projection(axes[1, 0], high_vertices, (0, 1), high_color, 0.06, "high")
+    axes[1, 0].set_xlabel("x")
+    axes[1, 0].set_ylabel("y")
+    axes[1, 0].legend(frameon=False, markerscale=6, loc="upper left")
+
+    setup_projection_axis(axes[1, 1], all_vertices, (0, 2), "Overlay: side view")
+    scatter_projection(axes[1, 1], low_vertices, (0, 2), low_color, 0.06, "low")
+    scatter_projection(axes[1, 1], high_vertices, (0, 2), high_color, 0.06, "high")
+    axes[1, 1].set_xlabel("x")
+    axes[1, 1].set_ylabel("z")
+    axes[1, 1].legend(frameon=False, markerscale=6, loc="upper left")
+
+    fig.suptitle("STL-Chamfer geometry split examples", fontsize=12)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    fig.savefig(output_dir / "geometry_score_examples.png", dpi=180)
+    plt.close(fig)
+
+
+def main() -> None:
+    args = parse_args()
+    force = load_force(args.force_root)
+    make_wake_examples(args.data_root, args.asset_root, args.output_dir, force)
+    make_geometry_examples(args.data_root, args.asset_root, args.output_dir, force, args.max_stl_points)
+    print(f"Wrote {args.output_dir / 'wake_score_examples.png'}")
+    print(f"Wrote {args.output_dir / 'geometry_score_examples.png'}")
+
+
+if __name__ == "__main__":
+    main()

splits/download_hf_inputs.py

@@ -0,0 +1,193 @@
+"""Download AhmedML split-regeneration inputs from Hugging Face.
+
+Default behavior downloads only the aggregate CSV files needed by
+splits/generate_splits.py:
+
+    python3 splits/download_hf_inputs.py --output-dir data
+
+For STL/PNG assets, prefer a directory outside this split package:
+
+    python3 splits/download_hf_inputs.py --output-dir ../ahmedml_hf_assets \
+      --include-stls --include-wake-images
+"""
+
+from __future__ import annotations
+
+import argparse
+from concurrent.futures import ThreadPoolExecutor, as_completed
+import os
+from pathlib import Path
+import random
+import time
+from urllib.error import HTTPError, URLError
+from urllib.parse import quote
+from urllib.request import Request, urlopen
+
+
+REPO_ID = "neashton/ahmedml"
+REVISION = "main"
+AGGREGATE_FILES = [
+    "force_mom_all.csv",
+    "geo_parameters_all.csv",
+]
+OPTIONAL_AGGREGATE_FILES = [
+    "force_mom_varref_all.csv",
+]
+IMAGE_SLICES = {
+    "X": range(22),
+    "Y": range(9),
+    "Z": range(6),
+}
+WAKE_IMAGE_SLICES = {
+    "X": (14, 15, 16),
+    "Y": (4,),
+}
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Download source inputs for AhmedML split regeneration.")
+    parser.add_argument("--repo-id", default=REPO_ID, help=f"Hugging Face dataset repo ID. Default: {REPO_ID}")
+    parser.add_argument("--revision", default=REVISION, help=f"Hub revision, branch, or tag. Default: {REVISION}")
+    parser.add_argument("--output-dir", type=Path, default=Path("data"), help="Directory where files are written.")
+    parser.add_argument("--include-varref", action="store_true", help="Also download force_mom_varref_all.csv.")
+    parser.add_argument("--include-images", action="store_true", help="Download all CpT and UxMean PNG images. This is large.")
+    parser.add_argument("--include-wake-images", action="store_true", help="Download only UxMean PNG slices needed for the image_wake split.")
+    parser.add_argument("--include-stls", action="store_true", help="Download run_*/ahmed_*.stl files. This is large.")
+    parser.add_argument("--runs", default="all", help="Run IDs for optional STL/image downloads: all or a comma/range expression.")
+    parser.add_argument("--workers", type=int, default=4, help="Parallel downloads. Default: 4.")
+    parser.add_argument("--retries", type=int, default=6, help="Retries for HTTP 429 and transient network errors.")
+    parser.add_argument("--retry-sleep", type=float, default=4.0, help="Initial retry sleep in seconds.")
+    parser.add_argument("--overwrite", action="store_true", help="Redownload files that already exist.")
+    parser.add_argument("--dry-run", action="store_true", help="Print the file list without downloading.")
+    return parser.parse_args()
+
+
+def parse_run_expression(expr: str) -> list[int]:
+    expr = expr.strip().lower()
+    if expr == "all":
+        return list(range(1, 501))
+    runs: set[int] = set()
+    for part in expr.split(","):
+        part = part.strip()
+        if not part:
+            continue
+        if "-" in part:
+            start, end = [int(value) for value in part.split("-", 1)]
+            if start > end:
+                start, end = end, start
+            runs.update(range(start, end + 1))
+        else:
+            runs.add(int(part))
+    invalid = sorted(run for run in runs if run < 1 or run > 500)
+    if invalid:
+        raise SystemExit(f"Invalid run IDs: {invalid[:10]}")
+    return sorted(runs)
+
+
+def hub_url(repo_id: str, revision: str, path: str) -> str:
+    quoted_path = "/".join(quote(part) for part in path.split("/"))
+    return f"https://huggingface.co/datasets/{repo_id}/resolve/{quote(revision, safe='')}/{quoted_path}"
+
+
+def retry_delay(base_sleep: float, attempt: int, retry_after: str | None = None) -> float:
+    if retry_after:
+        try:
+            return max(0.0, float(retry_after))
+        except ValueError:
+            pass
+    jitter = random.uniform(0.0, base_sleep)
+    return min(90.0, base_sleep * (2 ** max(0, attempt - 1)) + jitter)
+
+
+def download_one(
+    repo_id: str,
+    revision: str,
+    output_dir: Path,
+    rel_path: str,
+    overwrite: bool,
+    retries: int,
+    retry_sleep: float,
+) -> tuple[str, str]:
+    destination = output_dir / rel_path
+    if destination.exists() and not overwrite:
+        return rel_path, "skip"
+    destination.parent.mkdir(parents=True, exist_ok=True)
+    headers = {}
+    if os.environ.get("HF_TOKEN"):
+        headers["Authorization"] = f"Bearer {os.environ['HF_TOKEN']}"
+    for attempt in range(retries + 1):
+        request = Request(hub_url(repo_id, revision, rel_path), headers=headers)
+        try:
+            with urlopen(request, timeout=120) as response:
+                destination.write_bytes(response.read())
+            return rel_path, "ok"
+        except HTTPError as exc:
+            retryable = exc.code == 429 or 500 <= exc.code <= 599
+            if retryable and attempt < retries:
+                time.sleep(retry_delay(retry_sleep, attempt, exc.headers.get("Retry-After")))
+                continue
+            return rel_path, f"http_{exc.code}"
+        except URLError as exc:
+            if attempt < retries:
+                time.sleep(retry_delay(retry_sleep, attempt))
+                continue
+            return rel_path, f"url_error:{exc.reason}"
+    return rel_path, "failed"
+
+
+def main() -> None:
+    args = parse_args()
+    files = AGGREGATE_FILES.copy()
+    if args.include_varref:
+        files.extend(OPTIONAL_AGGREGATE_FILES)
+    runs = parse_run_expression(args.runs)
+    if args.include_stls:
+        files.extend(f"run_{run}/ahmed_{run}.stl" for run in runs)
+    if args.include_images:
+        for run in runs:
+            for axis, indices in IMAGE_SLICES.items():
+                files.extend(
+                    f"run_{run}/images/CpT/run_{run}-slice-total(p)_coeffMean-{axis}-{idx}.png"
+                    for idx in indices
+                )
+                files.extend(
+                    f"run_{run}/images/UxMean/run_{run}-slice-UMean-0-{axis}-{idx}.png"
+                    for idx in indices
+                )
+    elif args.include_wake_images:
+        for run in runs:
+            for axis, indices in WAKE_IMAGE_SLICES.items():
+                files.extend(
+                    f"run_{run}/images/UxMean/run_{run}-slice-UMean-0-{axis}-{idx}.png"
+                    for idx in indices
+                )
+
+    print(f"Repository: {args.repo_id}@{args.revision}")
+    print(f"Output dir: {args.output_dir}")
+    print(f"Files: {len(files)}")
+    if args.dry_run:
+        for path in files:
+            print(path)
+        return
+
+    with ThreadPoolExecutor(max_workers=max(1, args.workers)) as pool:
+        futures = [
+            pool.submit(
+                download_one,
+                args.repo_id,
+                args.revision,
+                args.output_dir,
+                path,
+                args.overwrite,
+                args.retries,
+                args.retry_sleep,
+            )
+            for path in files
+        ]
+        for future in as_completed(futures):
+            path, status = future.result()
+            print(f"{status:<8s} {path}")
+
+
+if __name__ == "__main__":
+    main()

splits/generate_splits.py

@@ -0,0 +1,512 @@
+"""Generate deterministic train/val/test splits for the AhmedML dataset.
+
+Outputs:
+  - splits/manifest.json
+  - splits/parameter_geometry_metrics.csv
+
+Split families:
+  1. full         - Noether-compatible seed-42 random split, 400/50/50
+  2. medium       - same val/test as full, train is 1/3 subsample
+  3. scarce       - same val/test as full, train is 1/6 subsample
+  4. super_scarce - same val/test as full, train is 1/36 subsample
+  5. geometry     - OOD STL-Chamfer local-isolation split
+  6. high_drag    - OOD high-drag split from force_mom_all.csv
+  7. low_drag     - OOD low-drag split from force_mom_all.csv
+  8. image_wake   - OOD image-derived wake split from UxMean PNGs
+
+For every OOD split, the validation set is drawn from the training-side
+population so hyperparameter tuning does not see the held-out extreme regime.
+"""
+
+from __future__ import annotations
+
+import csv
+import hashlib
+import json
+import math
+import os
+import random
+from pathlib import Path
+
+import numpy as np
+
+
+SCRIPT_DIR = Path(__file__).resolve().parent
+REPOSITORY_ROOT = SCRIPT_DIR.parent
+PACKAGE_ROOT = SCRIPT_DIR
+DATA_DIR = SCRIPT_DIR
+SPLITS_DIR = SCRIPT_DIR
+DATA_ROOT = Path(os.environ.get("AHMEDML_DATA_ROOT", REPOSITORY_ROOT))
+CHAMFER_METRICS = "chamfer_metrics.csv"
+IMAGE_METRICS = "image_metrics.csv"
+PARAMETER_GEOMETRY_METRICS = "parameter_geometry_metrics.csv"
+
+N_CASES = 500
+RUN_IDS = list(range(1, N_CASES + 1))
+SEED = 42
+MEDIUM_FRACTION = 1 / 3
+SCARCE_FRACTION = 1 / 6
+SUPER_SCARCE_FRACTION = 1 / 36
+OOD_TEST_FRACTION = 0.2
+VAL_FRACTION = 0.1
+TEST_FRACTION = 0.2
+VAL_FRACTION_OF_POOL = VAL_FRACTION / (1 - TEST_FRACTION)
+
+
+# Noether AhmedMLDefaultSplitIDs, generated from torch.randperm(500) with seed
+# 42. AhmedML has no hidden-test exclusion in that default split.
+FULL_TRAIN_IDS = [
+    1, 2, 3, 5, 6, 7, 8, 9, 10, 13, 14, 15, 16, 17, 18, 21, 23, 25, 27, 28, 30, 31, 32, 33, 34, 35, 36, 37, 39, 40,
+    42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 57, 58, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72,
+    73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100,
+    101, 102, 103, 105, 106, 107, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 125, 126,
+    128, 129, 130, 131, 132, 134, 135, 136, 137, 138, 139, 140, 141, 143, 144, 145, 146, 147, 148, 149, 151, 152,
+    153, 154, 155, 156, 157, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 174, 175, 176, 178,
+    179, 181, 182, 183, 184, 185, 186, 189, 190, 192, 193, 194, 195, 198, 200, 201, 202, 204, 206, 209, 211, 212,
+    213, 214, 216, 217, 218, 219, 220, 221, 223, 224, 225, 227, 229, 231, 232, 233, 235, 236, 237, 238, 239, 240,
+    242, 243, 244, 245, 246, 248, 249, 250, 251, 254, 255, 256, 257, 259, 261, 262, 264, 265, 266, 267, 268, 269,
+    270, 272, 273, 274, 276, 277, 278, 279, 282, 283, 285, 286, 287, 288, 289, 292, 293, 294, 296, 297, 299, 300,
+    301, 302, 305, 306, 307, 308, 309, 310, 313, 314, 315, 316, 317, 318, 319, 320, 323, 325, 326, 327, 330, 331,
+    332, 333, 334, 335, 336, 338, 339, 340, 342, 343, 344, 345, 346, 347, 348, 349, 351, 353, 355, 356, 357, 358,
+    359, 360, 361, 362, 365, 367, 368, 369, 370, 371, 373, 374, 375, 377, 378, 379, 381, 383, 384, 385, 386, 388,
+    389, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 402, 404, 406, 407, 408, 409, 411, 412, 413, 414, 415,
+    416, 417, 418, 419, 420, 421, 422, 425, 426, 427, 430, 431, 433, 434, 435, 437, 438, 439, 440, 442, 443, 444,
+    445, 446, 448, 449, 450, 451, 452, 453, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467, 468, 469,
+    470, 471, 473, 474, 475, 476, 477, 478, 479, 480, 481, 482, 483, 484, 485, 486, 488, 489, 490, 491, 492, 493,
+    495, 496, 497, 498, 499, 500,
+]
+
+FULL_VAL_IDS = [
+    24, 26, 29, 38, 41, 55, 59, 104, 108, 124, 133, 142, 158, 173, 180, 188, 196, 197, 199, 205, 207, 210, 222, 226,
+    230, 258, 263, 280, 281, 284, 290, 291, 295, 304, 312, 337, 350, 354, 363, 372, 387, 403, 405, 424, 428, 429,
+    432, 455, 472, 494,
+]
+
+FULL_TEST_IDS = [
+    4, 11, 12, 19, 20, 22, 56, 109, 127, 150, 165, 177, 187, 191, 203, 208, 215, 228, 234, 241, 247, 252, 253, 260,
+    271, 275, 298, 303, 311, 321, 322, 324, 328, 329, 341, 352, 364, 366, 376, 380, 382, 390, 401, 410, 423, 436,
+    441, 447, 454, 487,
+]
+
+
+def case_id(run_id: int) -> str:
+    return f"run_{run_id}"
+
+
+def run_id(case_id_value: str) -> int:
+    if not case_id_value.startswith("run_"):
+        raise ValueError(f"Malformed case ID: {case_id_value!r}")
+    return int(case_id_value.split("_", 1)[1])
+
+
+def make_case_ids(values: list[int]) -> list[str]:
+    return [case_id(value) for value in sorted(values)]
+
+
+def _rng(salt: str) -> random.Random:
+    seed_bytes = hashlib.sha256(f"{SEED}:{salt}".encode("utf-8")).digest()[:8]
+    return random.Random(int.from_bytes(seed_bytes, "big"))
+
+
+def _unit_hash(run: int, salt: str) -> float:
+    seed = hashlib.sha256(f"{SEED}:{salt}:{run}".encode("utf-8")).digest()[:8]
+    return int.from_bytes(seed, "big") / 2**64
+
+
+def _split_pool(pool: list[int], *, salt: str) -> tuple[list[int], list[int]]:
+    shuffled = pool.copy()
+    _rng(salt).shuffle(shuffled)
+    n_val = round(len(pool) * VAL_FRACTION_OF_POOL)
+    val = sorted(shuffled[:n_val])
+    train = sorted(shuffled[n_val:])
+    return train, val
+
+
+def _candidate_paths(filename: str) -> list[Path]:
+    roots = [
+        DATA_ROOT,
+        DATA_ROOT / "data",
+        DATA_ROOT / "dataset",
+        DATA_DIR,
+        DATA_DIR / "dataset",
+        PACKAGE_ROOT,
+        PACKAGE_ROOT / "dataset",
+        REPOSITORY_ROOT,
+        REPOSITORY_ROOT / "data",
+        Path.cwd(),
+        Path.cwd() / "data",
+    ]
+    seen: set[Path] = set()
+    paths: list[Path] = []
+    for root in roots:
+        path = root / filename
+        key = path.resolve() if path.exists() else path.absolute()
+        if key not in seen:
+            paths.append(path)
+            seen.add(key)
+    return paths
+
+
+def _clean_row(row: dict[str, str]) -> dict[str, str]:
+    return {key.strip(): value.strip() for key, value in row.items() if key is not None}
+
+
+def _float(value: str) -> float:
+    return float(value.replace(" ", ""))
+
+
+def load_force_mom() -> tuple[dict[int, dict[str, float]], str]:
+    for path in _candidate_paths("force_mom_all.csv"):
+        if not path.exists():
+            continue
+        records: dict[int, dict[str, float]] = {}
+        with path.open(encoding="utf-8-sig", newline="") as f:
+            for row in csv.DictReader(f):
+                clean = _clean_row(row)
+                rid = int(clean["run"])
+                records[rid] = {"cd": _float(clean["cd"]), "cl": _float(clean["cl"])}
+        missing = sorted(set(RUN_IDS) - set(records))
+        if missing:
+            raise ValueError(f"{path} is missing run IDs: {missing[:10]}")
+        return records, str(path)
+    raise FileNotFoundError("force_mom_all.csv not found; run splits/download_hf_inputs.py")
+
+
+def load_geo_parameters() -> tuple[dict[int, dict[str, float]], dict[int, bool], str]:
+    for path in _candidate_paths("geo_parameters_all.csv"):
+        if not path.exists():
+            continue
+        observed_records: dict[int, dict[str, float]] = {}
+        with path.open(encoding="utf-8-sig", newline="") as f:
+            for row in csv.DictReader(f):
+                clean = _clean_row(row)
+                rid = int(clean["run"])
+                observed_records[rid] = {
+                    key: _float(value)
+                    for key, value in clean.items()
+                    if key != "run"
+                }
+        if not observed_records:
+            raise ValueError(f"{path} has no geometry rows")
+
+        keys = sorted(next(iter(observed_records.values())).keys())
+        means = {
+            key: float(np.mean([record[key] for record in observed_records.values()]))
+            for key in keys
+        }
+        records: dict[int, dict[str, float]] = {}
+        observed: dict[int, bool] = {}
+        for rid in RUN_IDS:
+            if rid in observed_records:
+                records[rid] = observed_records[rid]
+                observed[rid] = True
+            else:
+                records[rid] = means.copy()
+                observed[rid] = False
+        return records, observed, str(path)
+    raise FileNotFoundError("geo_parameters_all.csv not found; run splits/download_hf_inputs.py")
+
+
+def _standardized_matrix(records: dict[int, dict[str, float]], fields: list[str], runs: list[int]) -> np.ndarray:
+    matrix = np.asarray([[records[rid][field] for field in fields] for rid in runs], dtype=float)
+    means = matrix.mean(axis=0)
+    stds = matrix.std(axis=0)
+    stds[stds == 0.0] = 1.0
+    return (matrix - means) / stds
+
+
+def _pairwise_distances(matrix: np.ndarray) -> np.ndarray:
+    diff = matrix[:, None, :] - matrix[None, :, :]
+    return np.sqrt(np.sum(diff * diff, axis=2))
+
+
+def geometry_isolation_scores(
+    geo_records: dict[int, dict[str, float]],
+) -> tuple[dict[int, float], dict[int, float]]:
+    runs = RUN_IDS.copy()
+    fields = sorted(next(iter(geo_records.values())).keys())
+    matrix = _standardized_matrix(geo_records, fields, runs)
+    distances = _pairwise_distances(matrix)
+    np.fill_diagonal(distances, np.inf)
+    nearest_10 = np.sort(distances, axis=1)[:, :10]
+    finite = np.where(np.isfinite(distances), distances, np.nan)
+    mean_10 = nearest_10.mean(axis=1)
+    mean_all = np.nanmean(finite, axis=1)
+    return (
+        {rid: float(mean_10[idx]) for idx, rid in enumerate(runs)},
+        {rid: float(mean_all[idx]) for idx, rid in enumerate(runs)},
+    )
+
+
+def write_parameter_geometry_metrics(
+    scores: dict[int, float],
+    mean_all: dict[int, float],
+    observed: dict[int, bool],
+) -> None:
+    DATA_DIR.mkdir(parents=True, exist_ok=True)
+    path = DATA_DIR / PARAMETER_GEOMETRY_METRICS
+    with path.open("w", encoding="utf-8", newline="") as f:
+        writer = csv.DictWriter(
+            f,
+            fieldnames=[
+                "run",
+                "geometry_observed",
+                "ood_score",
+                "mean_10_nn_parameter_distance",
+                "mean_all_parameter_distance",
+            ],
+        )
+        writer.writeheader()
+        for rid in RUN_IDS:
+            writer.writerow(
+                {
+                    "run": rid,
+                    "geometry_observed": str(observed[rid]).lower(),
+                    "ood_score": scores[rid],
+                    "mean_10_nn_parameter_distance": scores[rid],
+                    "mean_all_parameter_distance": mean_all[rid],
+                }
+            )
+
+
+def load_metric_scores(
+    filename: str,
+    value_column: str,
+    *,
+    observed_column: str | None = None,
+) -> tuple[dict[int, float], str]:
+    for path in _candidate_paths(filename):
+        if not path.exists():
+            continue
+        scores: dict[int, float] = {}
+        unobserved: list[int] = []
+        with path.open(encoding="utf-8-sig", newline="") as f:
+            for row in csv.DictReader(f):
+                clean = _clean_row(row)
+                rid = int(clean["run"])
+                if observed_column and clean.get(observed_column, "true").lower() != "true":
+                    unobserved.append(rid)
+                    continue
+                value = clean.get(value_column, "")
+                if value == "":
+                    unobserved.append(rid)
+                    continue
+                scores[rid] = _float(value)
+        missing = sorted(set(RUN_IDS) - set(scores))
+        if missing or unobserved:
+            missing_preview = ", ".join(str(x) for x in (missing + unobserved)[:12])
+            raise ValueError(
+                f"{path} does not contain complete observed {value_column} scores; "
+                f"missing/unobserved runs include: {missing_preview}"
+            )
+        return scores, str(path)
+    raise FileNotFoundError(f"{filename} not found; compute it before running this generator")
+
+
+def load_chamfer_scores() -> tuple[dict[int, float], str]:
+    return load_metric_scores(CHAMFER_METRICS, "ood_score")
+
+
+def load_image_wake_scores() -> tuple[dict[int, float], str]:
+    return load_metric_scores(IMAGE_METRICS, "image_wake_score", observed_column="image_wake_observed")
+
+
+def force_scores(records: dict[int, dict[str, float]]) -> dict[str, dict[int, float]]:
+    cd = {rid: row["cd"] for rid, row in records.items()}
+    return {
+        "high_drag": cd,
+        "low_drag": {rid: -value for rid, value in cd.items()},
+    }
+
+
+def ranked_ood_split(scores: dict[int, float], *, salt: str) -> tuple[list[int], list[int], list[int]]:
+    ranked = sorted(scores, key=lambda rid: (scores[rid], rid))
+    n_test = round(len(ranked) * OOD_TEST_FRACTION)
+    test = sorted(ranked[-n_test:])
+    pool = sorted(ranked[:-n_test])
+    train, val = _split_pool(pool, salt=salt)
+    return train, val, test
+
+
+def diverse_training_order(
+    force_records: dict[int, dict[str, float]],
+    geo_records: dict[int, dict[str, float]],
+) -> list[int]:
+    train_pool = FULL_TRAIN_IDS.copy()
+    feature_rows: dict[int, list[float]] = {rid: [] for rid in train_pool}
+
+    for field in ["cd", "cl"]:
+        values = np.asarray([force_records[rid][field] for rid in train_pool], dtype=float)
+        mean = float(values.mean())
+        std = float(values.std()) or 1.0
+        for rid in train_pool:
+            feature_rows[rid].append((force_records[rid][field] - mean) / std)
+
+    for field in sorted(next(iter(geo_records.values())).keys()):
+        values = np.asarray([geo_records[rid][field] for rid in train_pool], dtype=float)
+        mean = float(values.mean())
+        std = float(values.std()) or 1.0
+        for rid in train_pool:
+            feature_rows[rid].append((geo_records[rid][field] - mean) / std)
+
+    def distance(a: int, b: int) -> float:
+        return math.sqrt(sum((x - y) ** 2 for x, y in zip(feature_rows[a], feature_rows[b])))
+
+    first = max(
+        train_pool,
+        key=lambda rid: (
+            math.sqrt(sum(value * value for value in feature_rows[rid])),
+            _unit_hash(rid, "scarce_first_tie_break"),
+        ),
+    )
+    selected = [first]
+    remaining = [rid for rid in train_pool if rid != first]
+    while remaining:
+        next_rid = max(
+            remaining,
+            key=lambda rid: (
+                min(distance(rid, chosen) for chosen in selected),
+                _unit_hash(rid, "scarce_tie_break"),
+            ),
+        )
+        selected.append(next_rid)
+        remaining.remove(next_rid)
+    return selected
+
+
+def _validate_noether_full_split() -> None:
+    groups = {
+        "full_train": FULL_TRAIN_IDS,
+        "full_val": FULL_VAL_IDS,
+        "full_test": FULL_TEST_IDS,
+    }
+    seen: dict[int, str] = {}
+    for name, ids in groups.items():
+        if len(ids) != len(set(ids)):
+            raise AssertionError(f"{name} contains duplicate run IDs")
+        for rid in ids:
+            if rid < 1 or rid > N_CASES:
+                raise AssertionError(f"{name} has invalid run ID {rid}")
+            if rid in seen:
+                raise AssertionError(f"run {rid} appears in {seen[rid]} and {name}")
+            seen[rid] = name
+    if set(seen) != set(RUN_IDS):
+        raise AssertionError("full split does not cover all AhmedML runs")
+    if (len(FULL_TRAIN_IDS), len(FULL_VAL_IDS), len(FULL_TEST_IDS)) != (400, 50, 50):
+        raise AssertionError("unexpected full split sizes")
+
+
+def generate_splits() -> tuple[dict[str, list[str]], str, str, str, str]:
+    _validate_noether_full_split()
+    force_records, force_source = load_force_mom()
+    geo_records, geo_observed, geo_source = load_geo_parameters()
+    parameter_geometry_scores, parameter_geometry_mean_all = geometry_isolation_scores(geo_records)
+    write_parameter_geometry_metrics(parameter_geometry_scores, parameter_geometry_mean_all, geo_observed)
+    geometry_scores, geometry_source = load_chamfer_scores()
+    image_wake_scores, image_source = load_image_wake_scores()
+
+    splits: dict[str, list[str]] = {}
+    splits["full_train"] = make_case_ids(FULL_TRAIN_IDS)
+    splits["full_val"] = make_case_ids(FULL_VAL_IDS)
+    splits["full_test"] = make_case_ids(FULL_TEST_IDS)
+
+    order = diverse_training_order(force_records, geo_records)
+    n_medium = round(len(FULL_TRAIN_IDS) * MEDIUM_FRACTION)
+    n_scarce = round(len(FULL_TRAIN_IDS) * SCARCE_FRACTION)
+    n_super_scarce = round(len(FULL_TRAIN_IDS) * SUPER_SCARCE_FRACTION)
+    splits["medium_train"] = make_case_ids(sorted(order[:n_medium]))
+    splits["medium_val"] = splits["full_val"]
+    splits["medium_test"] = splits["full_test"]
+    splits["scarce_train"] = make_case_ids(sorted(order[:n_scarce]))
+    splits["scarce_val"] = splits["full_val"]
+    splits["scarce_test"] = splits["full_test"]
+    splits["super_scarce_train"] = make_case_ids(sorted(order[:n_super_scarce]))
+    splits["super_scarce_val"] = splits["full_val"]
+    splits["super_scarce_test"] = splits["full_test"]
+
+    for name, scores in force_scores(force_records).items():
+        train, val, test = ranked_ood_split(scores, salt=f"{name}_val_selection")
+        splits[f"{name}_train"] = make_case_ids(train)
+        splits[f"{name}_val"] = make_case_ids(val)
+        splits[f"{name}_test"] = make_case_ids(test)
+
+    train, val, test = ranked_ood_split(geometry_scores, salt="geometry_val_selection")
+    splits["geometry_train"] = make_case_ids(train)
+    splits["geometry_val"] = make_case_ids(val)
+    splits["geometry_test"] = make_case_ids(test)
+
+    train, val, test = ranked_ood_split(image_wake_scores, salt="image_wake_val_selection")
+    splits["image_wake_train"] = make_case_ids(train)
+    splits["image_wake_val"] = make_case_ids(val)
+    splits["image_wake_test"] = make_case_ids(test)
+    return splits, force_source, geo_source, geometry_source, image_source
+
+
+def validate_splits(splits: dict[str, list[str]]) -> None:
+    all_cases = {case_id(rid) for rid in RUN_IDS}
+    split_names = sorted({key.rsplit("_", 1)[0] for key in splits})
+    for name in split_names:
+        train = set(splits[f"{name}_train"])
+        val = set(splits[f"{name}_val"])
+        test = set(splits[f"{name}_test"])
+        assert not (train & val), f"{name}: train/val overlap"
+        assert not (train & test), f"{name}: train/test overlap"
+        assert not (val & test), f"{name}: val/test overlap"
+        assert train | val | test <= all_cases, f"{name}: non-AhmedML run included"
+
+    assert (len(splits["full_train"]), len(splits["full_val"]), len(splits["full_test"])) == (400, 50, 50)
+    for prefix in ["medium", "scarce", "super_scarce"]:
+        assert splits[f"{prefix}_val"] == splits["full_val"], f"{prefix}_val must equal full_val"
+        assert splits[f"{prefix}_test"] == splits["full_test"], f"{prefix}_test must equal full_test"
+
+    assert set(splits["super_scarce_train"]) < set(splits["scarce_train"])
+    assert set(splits["scarce_train"]) < set(splits["medium_train"])
+    assert set(splits["medium_train"]) < set(splits["full_train"])
+
+    for prefix in ["full", "geometry", "high_drag", "low_drag", "image_wake"]:
+        total = len(splits[f"{prefix}_train"]) + len(splits[f"{prefix}_val"]) + len(splits[f"{prefix}_test"])
+        assert total == N_CASES, f"{prefix}: expected {N_CASES} cases, got {total}"
+
+    for prefix in ["geometry", "high_drag", "low_drag", "image_wake"]:
+        assert (
+            len(splits[f"{prefix}_train"]),
+            len(splits[f"{prefix}_val"]),
+            len(splits[f"{prefix}_test"]),
+        ) == (350, 50, 100), f"{prefix}: unexpected OOD split sizes"
+
+
+def main() -> None:
+    splits, force_source, geo_source, geometry_source, image_source = generate_splits()
+    validate_splits(splits)
+
+    SPLITS_DIR.mkdir(parents=True, exist_ok=True)
+    output = SPLITS_DIR / "manifest.json"
+    output.write_text(json.dumps(splits, indent=4) + "\n", encoding="utf-8")
+
+    print("AhmedML Splits")
+    print("=" * 60)
+    print(f"  Runs: {N_CASES}")
+    print(f"  Seed: {SEED}")
+    print(f"  Force/moment source: {force_source}")
+    print(f"  Geometry-parameter source: {geo_source}")
+    print(f"  STL-Chamfer source: {geometry_source}")
+    print(f"  Image-wake source: {image_source}")
+    print()
+    print(f"  {'Split':<18s} {'Train':>6s} {'Val':>6s} {'Test':>6s} {'Total':>6s}")
+    print(f"  {'-' * 46}")
+    for name in sorted({key.rsplit('_', 1)[0] for key in splits}):
+        n_train = len(splits[f"{name}_train"])
+        n_val = len(splits[f"{name}_val"])
+        n_test = len(splits[f"{name}_test"])
+        print(f"  {name:<18s} {n_train:>6d} {n_val:>6d} {n_test:>6d} {n_train + n_val + n_test:>6d}")
+    print()
+    print(f"  Manifest: {output}")
+    print(f"  STL-Chamfer metrics: {DATA_DIR / CHAMFER_METRICS}")
+    print(f"  Image metrics: {DATA_DIR / IMAGE_METRICS}")
+    print(f"  Parameter geometry metrics: {DATA_DIR / PARAMETER_GEOMETRY_METRICS}")
+    print(f"  Keys: {len(splits)}")
+    print("All validations passed.")
+
+
+if __name__ == "__main__":
+    main()

splits/image_metrics.csv

@@ -0,0 +1,501 @@
+run,image_wake_score,image_wake_observed,uxmean_images,uxmean_slices
+1,0.3461875865856806,true,4,Y-4;X-14;X-15;X-16
+2,0.351776992281278,true,4,Y-4;X-14;X-15;X-16
+3,0.3631688902775446,true,4,Y-4;X-14;X-15;X-16
+4,0.357019712527593,true,4,Y-4;X-14;X-15;X-16
+5,0.351190984249115,true,4,Y-4;X-14;X-15;X-16
+6,0.38376043240229285,true,4,Y-4;X-14;X-15;X-16
+7,0.3426342258850733,true,4,Y-4;X-14;X-15;X-16
+8,0.3663400808970133,true,4,Y-4;X-14;X-15;X-16
+9,0.3424033423264822,true,4,Y-4;X-14;X-15;X-16
+10,0.3609602749347687,true,4,Y-4;X-14;X-15;X-16
+11,0.33907972276210785,true,4,Y-4;X-14;X-15;X-16
+12,0.381332149108251,true,4,Y-4;X-14;X-15;X-16
+13,0.3817415783802668,true,4,Y-4;X-14;X-15;X-16
+14,0.38172480960687005,true,4,Y-4;X-14;X-15;X-16
+15,0.3451130837202072,true,4,Y-4;X-14;X-15;X-16
+16,0.3515610297520956,true,4,Y-4;X-14;X-15;X-16
+17,0.3418483932813009,true,4,Y-4;X-14;X-15;X-16
+18,0.3787537068128586,true,4,Y-4;X-14;X-15;X-16
+19,0.34178511798381805,true,4,Y-4;X-14;X-15;X-16
+20,0.3646722733974457,true,4,Y-4;X-14;X-15;X-16
+21,0.37189146379629767,true,4,Y-4;X-14;X-15;X-16
+22,0.35373611748218536,true,4,Y-4;X-14;X-15;X-16
+23,0.37545597056547797,true,4,Y-4;X-14;X-15;X-16
+24,0.3514778365691503,true,4,Y-4;X-14;X-15;X-16
+25,0.3433034668366114,true,4,Y-4;X-14;X-15;X-16
+26,0.3771053453286489,true,4,Y-4;X-14;X-15;X-16
+27,0.35247685511906945,true,4,Y-4;X-14;X-15;X-16
+28,0.34733870128790534,true,4,Y-4;X-14;X-15;X-16
+29,0.3406224399805069,true,4,Y-4;X-14;X-15;X-16
+30,0.34595009684562683,true,4,Y-4;X-14;X-15;X-16
+31,0.3439881602923075,true,4,Y-4;X-14;X-15;X-16
+32,0.3570915758609772,true,4,Y-4;X-14;X-15;X-16
+33,0.3531102190415064,true,4,Y-4;X-14;X-15;X-16
+34,0.3848358591397604,true,4,Y-4;X-14;X-15;X-16
+35,0.34848642349243164,true,4,Y-4;X-14;X-15;X-16
+36,0.36217309534549713,true,4,Y-4;X-14;X-15;X-16
+37,0.3483212242523829,true,4,Y-4;X-14;X-15;X-16
+38,0.33929045498371124,true,4,Y-4;X-14;X-15;X-16
+39,0.3441611478726069,true,4,Y-4;X-14;X-15;X-16
+40,0.3530972748994827,true,4,Y-4;X-14;X-15;X-16
+41,0.3833983143170675,true,4,Y-4;X-14;X-15;X-16
+42,0.3438113828500112,true,4,Y-4;X-14;X-15;X-16
+43,0.34940263628959656,true,4,Y-4;X-14;X-15;X-16
+44,0.3445309648911158,true,4,Y-4;X-14;X-15;X-16
+45,0.3501907934745153,true,4,Y-4;X-14;X-15;X-16
+46,0.35458605984846747,true,4,Y-4;X-14;X-15;X-16
+47,0.35241089264551795,true,4,Y-4;X-14;X-15;X-16
+48,0.3517579436302185,true,4,Y-4;X-14;X-15;X-16
+49,0.3495356192191442,true,4,Y-4;X-14;X-15;X-16
+50,0.34348961214224494,true,4,Y-4;X-14;X-15;X-16
+51,0.34816386799017585,true,4,Y-4;X-14;X-15;X-16
+52,0.35219246645768487,true,4,Y-4;X-14;X-15;X-16
+53,0.3491321156422297,true,4,Y-4;X-14;X-15;X-16
+54,0.33699025213718414,true,4,Y-4;X-14;X-15;X-16
+55,0.3794567584991455,true,4,Y-4;X-14;X-15;X-16
+56,0.34859101474285126,true,4,Y-4;X-14;X-15;X-16
+57,0.36692495147387183,true,4,Y-4;X-14;X-15;X-16
+58,0.3380003720521927,true,4,Y-4;X-14;X-15;X-16
+59,0.3460168093442917,true,4,Y-4;X-14;X-15;X-16
+60,0.3728251407543818,true,4,Y-4;X-14;X-15;X-16
+61,0.35211281975110376,true,4,Y-4;X-14;X-15;X-16
+62,0.35988104840119683,true,4,Y-4;X-14;X-15;X-16
+63,0.3542174200216929,true,4,Y-4;X-14;X-15;X-16
+64,0.34016656378904975,true,4,Y-4;X-14;X-15;X-16
+65,0.35790068904558814,true,4,Y-4;X-14;X-15;X-16
+66,0.3379997809727987,true,4,Y-4;X-14;X-15;X-16
+67,0.35174825290838874,true,4,Y-4;X-14;X-15;X-16
+68,0.347043921550115,true,4,Y-4;X-14;X-15;X-16
+69,0.3739911367495855,true,4,Y-4;X-14;X-15;X-16
+70,0.3401368806759516,true,4,Y-4;X-14;X-15;X-16
+71,0.37611752251784003,true,4,Y-4;X-14;X-15;X-16
+72,0.34206880132357276,true,4,Y-4;X-14;X-15;X-16
+73,0.3790056655804316,true,4,Y-4;X-14;X-15;X-16
+74,0.3843736698230108,true,4,Y-4;X-14;X-15;X-16
+75,0.3820884426434835,true,4,Y-4;X-14;X-15;X-16
+76,0.37921497225761414,true,4,Y-4;X-14;X-15;X-16
+77,0.35149887204170227,true,4,Y-4;X-14;X-15;X-16
+78,0.3554788132508596,true,4,Y-4;X-14;X-15;X-16
+79,0.3666918079058329,true,4,Y-4;X-14;X-15;X-16
+80,0.3449161251386007,true,4,Y-4;X-14;X-15;X-16
+81,0.38276966412862146,true,4,Y-4;X-14;X-15;X-16
+82,0.3812187761068344,true,4,Y-4;X-14;X-15;X-16
+83,0.3520535876353582,true,4,Y-4;X-14;X-15;X-16
+84,0.3463789075613022,true,4,Y-4;X-14;X-15;X-16
+85,0.34566865861415863,true,4,Y-4;X-14;X-15;X-16
+86,0.3460667977730433,true,4,Y-4;X-14;X-15;X-16
+87,0.3419599632422129,true,4,Y-4;X-14;X-15;X-16
+88,0.3476617236932119,true,4,Y-4;X-14;X-15;X-16
+89,0.3706303735574087,true,4,Y-4;X-14;X-15;X-16
+90,0.34644113977750146,true,4,Y-4;X-14;X-15;X-16
+91,0.3523273169994354,true,4,Y-4;X-14;X-15;X-16
+92,0.3582994739214579,true,4,Y-4;X-14;X-15;X-16
+93,0.3538975119590759,true,4,Y-4;X-14;X-15;X-16
+94,0.3790803998708725,true,4,Y-4;X-14;X-15;X-16
+95,0.3475813915332159,true,4,Y-4;X-14;X-15;X-16
+96,0.3395392199357351,true,4,Y-4;X-14;X-15;X-16
+97,0.3470164636770884,true,4,Y-4;X-14;X-15;X-16
+98,0.3476693630218506,true,4,Y-4;X-14;X-15;X-16
+99,0.3503692001104355,true,4,Y-4;X-14;X-15;X-16
+100,0.3483559240897497,true,4,Y-4;X-14;X-15;X-16
+101,0.3559330254793167,true,4,Y-4;X-14;X-15;X-16
+102,0.3472145597139994,true,4,Y-4;X-14;X-15;X-16
+103,0.3489331056674322,true,4,Y-4;X-14;X-15;X-16
+104,0.3521694938341776,true,4,Y-4;X-14;X-15;X-16
+105,0.3445242245992025,true,4,Y-4;X-14;X-15;X-16
+106,0.34721974531809485,true,4,Y-4;X-14;X-15;X-16
+107,0.37994764248530066,true,4,Y-4;X-14;X-15;X-16
+108,0.3588190277417501,true,4,Y-4;X-14;X-15;X-16
+109,0.34389907121658325,true,4,Y-4;X-14;X-15;X-16
+110,0.36150750021139777,true,4,Y-4;X-14;X-15;X-16
+111,0.3548774917920431,true,4,Y-4;X-14;X-15;X-16
+112,0.3599904775619507,true,4,Y-4;X-14;X-15;X-16
+113,0.3504287203152975,true,4,Y-4;X-14;X-15;X-16
+114,0.35895157853762305,true,4,Y-4;X-14;X-15;X-16
+115,0.3705332577228546,true,4,Y-4;X-14;X-15;X-16
+116,0.3560210565725962,true,4,Y-4;X-14;X-15;X-16
+117,0.3468300700187683,true,4,Y-4;X-14;X-15;X-16
+118,0.3560007860263189,true,4,Y-4;X-14;X-15;X-16
+119,0.3539868493874868,true,4,Y-4;X-14;X-15;X-16
+120,0.33614909152189887,true,4,Y-4;X-14;X-15;X-16
+121,0.34865080316861474,true,4,Y-4;X-14;X-15;X-16
+122,0.3515466203292211,true,4,Y-4;X-14;X-15;X-16
+123,0.35888379315535224,true,4,Y-4;X-14;X-15;X-16
+124,0.34799129267533624,true,4,Y-4;X-14;X-15;X-16
+125,0.3448231518268585,true,4,Y-4;X-14;X-15;X-16
+126,0.3577128052711487,true,4,Y-4;X-14;X-15;X-16
+127,0.350676084558169,true,4,Y-4;X-14;X-15;X-16
+128,0.3510897010564804,true,4,Y-4;X-14;X-15;X-16
+129,0.3526856452226639,true,4,Y-4;X-14;X-15;X-16
+130,0.36019415160020196,true,4,Y-4;X-14;X-15;X-16
+131,0.351927121480306,true,4,Y-4;X-14;X-15;X-16
+132,0.33989305794239044,true,4,Y-4;X-14;X-15;X-16
+133,0.34871854881445563,true,4,Y-4;X-14;X-15;X-16
+134,0.34301573038101196,true,4,Y-4;X-14;X-15;X-16
+135,0.35838388899962104,true,4,Y-4;X-14;X-15;X-16
+136,0.34030034144719445,true,4,Y-4;X-14;X-15;X-16
+137,0.3529029885927836,true,4,Y-4;X-14;X-15;X-16
+138,0.3525668432315191,true,4,Y-4;X-14;X-15;X-16
+139,0.34885677198568976,true,4,Y-4;X-14;X-15;X-16
+140,0.35446813205877936,true,4,Y-4;X-14;X-15;X-16
+141,0.3821118424336115,true,4,Y-4;X-14;X-15;X-16
+142,0.34470916291077935,true,4,Y-4;X-14;X-15;X-16
+143,0.34355669716993964,true,4,Y-4;X-14;X-15;X-16
+144,0.3483401437600454,true,4,Y-4;X-14;X-15;X-16
+145,0.35135207076867425,true,4,Y-4;X-14;X-15;X-16
+146,0.3505520621935526,true,4,Y-4;X-14;X-15;X-16
+147,0.3455455402533213,true,4,Y-4;X-14;X-15;X-16
+148,0.37942679723103845,true,4,Y-4;X-14;X-15;X-16
+149,0.3494625886281332,true,4,Y-4;X-14;X-15;X-16
+150,0.35253724455833435,true,4,Y-4;X-14;X-15;X-16
+151,0.35309371848901117,true,4,Y-4;X-14;X-15;X-16
+152,0.38439940909544623,true,4,Y-4;X-14;X-15;X-16
+153,0.34806210796038306,true,4,Y-4;X-14;X-15;X-16
+154,0.3449005136887232,true,4,Y-4;X-14;X-15;X-16
+155,0.34405648708343506,true,4,Y-4;X-14;X-15;X-16
+156,0.34394633273283637,true,4,Y-4;X-14;X-15;X-16
+157,0.3540697246789932,true,4,Y-4;X-14;X-15;X-16
+158,0.34131025771299994,true,4,Y-4;X-14;X-15;X-16
+159,0.3388786961634954,true,4,Y-4;X-14;X-15;X-16
+160,0.34834580620129907,true,4,Y-4;X-14;X-15;X-16
+161,0.35110872983932495,true,4,Y-4;X-14;X-15;X-16
+162,0.3438411752382914,true,4,Y-4;X-14;X-15;X-16
+163,0.3473967661460241,true,4,Y-4;X-14;X-15;X-16
+164,0.34460139771302545,true,4,Y-4;X-14;X-15;X-16
+165,0.3480007996161779,true,4,Y-4;X-14;X-15;X-16
+166,0.35239056746164954,true,4,Y-4;X-14;X-15;X-16
+167,0.35786113142967224,true,4,Y-4;X-14;X-15;X-16
+168,0.3494000732898712,true,4,Y-4;X-14;X-15;X-16
+169,0.3460137049357096,true,4,Y-4;X-14;X-15;X-16
+170,0.3431391716003418,true,4,Y-4;X-14;X-15;X-16
+171,0.3851578583319982,true,4,Y-4;X-14;X-15;X-16
+172,0.35515041152636206,true,4,Y-4;X-14;X-15;X-16
+173,0.3826426019271215,true,4,Y-4;X-14;X-15;X-16
+174,0.38391278187433875,true,4,Y-4;X-14;X-15;X-16
+175,0.3438434898853302,true,4,Y-4;X-14;X-15;X-16
+176,0.3598177284002304,true,4,Y-4;X-14;X-15;X-16
+177,0.3673089941342672,true,4,Y-4;X-14;X-15;X-16
+178,0.3768031944831213,true,4,Y-4;X-14;X-15;X-16
+179,0.3490620503822962,true,4,Y-4;X-14;X-15;X-16
+180,0.33987779915332794,true,4,Y-4;X-14;X-15;X-16
+181,0.3394384731849035,true,4,Y-4;X-14;X-15;X-16
+182,0.3503947754700979,true,4,Y-4;X-14;X-15;X-16
+183,0.3419712036848068,true,4,Y-4;X-14;X-15;X-16
+184,0.3630563020706177,true,4,Y-4;X-14;X-15;X-16
+185,0.3388984054327011,true,4,Y-4;X-14;X-15;X-16
+186,0.34217621882756555,true,4,Y-4;X-14;X-15;X-16
+187,0.37315139174461365,true,4,Y-4;X-14;X-15;X-16
+188,0.35754188895225525,true,4,Y-4;X-14;X-15;X-16
+189,0.36037887136141455,true,4,Y-4;X-14;X-15;X-16
+190,0.3776668608188629,true,4,Y-4;X-14;X-15;X-16
+191,0.3724663952986399,true,4,Y-4;X-14;X-15;X-16
+192,0.35704356928666436,true,4,Y-4;X-14;X-15;X-16
+193,0.362161785364151,true,4,Y-4;X-14;X-15;X-16
+194,0.3536730309327444,true,4,Y-4;X-14;X-15;X-16
+195,0.3517526239156723,true,4,Y-4;X-14;X-15;X-16
+196,0.371451790134112,true,4,Y-4;X-14;X-15;X-16
+197,0.37124224503835046,true,4,Y-4;X-14;X-15;X-16
+198,0.34649654726187384,true,4,Y-4;X-14;X-15;X-16
+199,0.353202223777771,true,4,Y-4;X-14;X-15;X-16
+200,0.3502361923456192,true,4,Y-4;X-14;X-15;X-16
+201,0.35929346084594727,true,4,Y-4;X-14;X-15;X-16
+202,0.35148817797501886,true,4,Y-4;X-14;X-15;X-16
+203,0.3406952073176702,true,4,Y-4;X-14;X-15;X-16
+204,0.381009096900622,true,4,Y-4;X-14;X-15;X-16
+205,0.3872829874356588,true,4,Y-4;X-14;X-15;X-16
+206,0.33981030682722724,true,4,Y-4;X-14;X-15;X-16
+207,0.3466399163007736,true,4,Y-4;X-14;X-15;X-16
+208,0.36445675790309906,true,4,Y-4;X-14;X-15;X-16
+209,0.3407026926676432,true,4,Y-4;X-14;X-15;X-16
+210,0.3551466266314188,true,4,Y-4;X-14;X-15;X-16
+211,0.34485720098018646,true,4,Y-4;X-14;X-15;X-16
+212,0.3440009454886118,true,4,Y-4;X-14;X-15;X-16
+213,0.3424153576294581,true,4,Y-4;X-14;X-15;X-16
+214,0.3421995540459951,true,4,Y-4;X-14;X-15;X-16
+215,0.34202566246191657,true,4,Y-4;X-14;X-15;X-16
+216,0.37500021855036414,true,4,Y-4;X-14;X-15;X-16
+217,0.37941510478655494,true,4,Y-4;X-14;X-15;X-16
+218,0.3497963696718216,true,4,Y-4;X-14;X-15;X-16
+219,0.3684433698654175,true,4,Y-4;X-14;X-15;X-16
+220,0.3769674847523371,true,4,Y-4;X-14;X-15;X-16
+221,0.34530789156754815,true,4,Y-4;X-14;X-15;X-16
+222,0.3581785559654236,true,4,Y-4;X-14;X-15;X-16
+223,0.3498951494693756,true,4,Y-4;X-14;X-15;X-16
+224,0.34697679678599036,true,4,Y-4;X-14;X-15;X-16
+225,0.34473489721616113,true,4,Y-4;X-14;X-15;X-16
+226,0.340437059601148,true,4,Y-4;X-14;X-15;X-16
+227,0.3490713139375051,true,4,Y-4;X-14;X-15;X-16
+228,0.38375222186247504,true,4,Y-4;X-14;X-15;X-16
+229,0.3442392200231552,true,4,Y-4;X-14;X-15;X-16
+230,0.3454762349526087,true,4,Y-4;X-14;X-15;X-16
+231,0.34291665752728784,true,4,Y-4;X-14;X-15;X-16
+232,0.37564513087272644,true,4,Y-4;X-14;X-15;X-16
+233,0.3497479359308878,true,4,Y-4;X-14;X-15;X-16
+234,0.3589025388161341,true,4,Y-4;X-14;X-15;X-16
+235,0.3473763714234034,true,4,Y-4;X-14;X-15;X-16
+236,0.3824052860339483,true,4,Y-4;X-14;X-15;X-16
+237,0.3558925986289978,true,4,Y-4;X-14;X-15;X-16
+238,0.35916968186696374,true,4,Y-4;X-14;X-15;X-16
+239,0.34153202176094055,true,4,Y-4;X-14;X-15;X-16
+240,0.34265541533629096,true,4,Y-4;X-14;X-15;X-16
+241,0.35240791738033295,true,4,Y-4;X-14;X-15;X-16
+242,0.3386556704839071,true,4,Y-4;X-14;X-15;X-16
+243,0.34625618159770966,true,4,Y-4;X-14;X-15;X-16
+244,0.351157416899999,true,4,Y-4;X-14;X-15;X-16
+245,0.35502021511395776,true,4,Y-4;X-14;X-15;X-16
+246,0.3564945856730143,true,4,Y-4;X-14;X-15;X-16
+247,0.3474683612585068,true,4,Y-4;X-14;X-15;X-16
+248,0.34446648756663,true,4,Y-4;X-14;X-15;X-16
+249,0.3455555637677511,true,4,Y-4;X-14;X-15;X-16
+250,0.3782188842693964,true,4,Y-4;X-14;X-15;X-16
+251,0.35077094535032904,true,4,Y-4;X-14;X-15;X-16
+252,0.3635778079430262,true,4,Y-4;X-14;X-15;X-16
+253,0.354649489124616,true,4,Y-4;X-14;X-15;X-16
+254,0.3548106998205185,true,4,Y-4;X-14;X-15;X-16
+255,0.34576425453027093,true,4,Y-4;X-14;X-15;X-16
+256,0.35002402464548743,true,4,Y-4;X-14;X-15;X-16
+257,0.3495090752840042,true,4,Y-4;X-14;X-15;X-16
+258,0.3466632068157196,true,4,Y-4;X-14;X-15;X-16
+259,0.3690320551395416,true,4,Y-4;X-14;X-15;X-16
+260,0.34037482738494873,true,4,Y-4;X-14;X-15;X-16
+261,0.34345343708992004,true,4,Y-4;X-14;X-15;X-16
+262,0.3650604536135992,true,4,Y-4;X-14;X-15;X-16
+263,0.34499774376551307,true,4,Y-4;X-14;X-15;X-16
+264,0.37977453569571173,true,4,Y-4;X-14;X-15;X-16
+265,0.35875863333543145,true,4,Y-4;X-14;X-15;X-16
+266,0.3475164522727331,true,4,Y-4;X-14;X-15;X-16
+267,0.3482773005962372,true,4,Y-4;X-14;X-15;X-16
+268,0.3405672460794449,true,4,Y-4;X-14;X-15;X-16
+269,0.3572241266568502,true,4,Y-4;X-14;X-15;X-16
+270,0.3448906342188517,true,4,Y-4;X-14;X-15;X-16
+271,0.3483964304129282,true,4,Y-4;X-14;X-15;X-16
+272,0.3476826250553131,true,4,Y-4;X-14;X-15;X-16
+273,0.34731615583101905,true,4,Y-4;X-14;X-15;X-16
+274,0.38033088545004523,true,4,Y-4;X-14;X-15;X-16
+275,0.33805207908153534,true,4,Y-4;X-14;X-15;X-16
+276,0.3640700429677963,true,4,Y-4;X-14;X-15;X-16
+277,0.3545143554608027,true,4,Y-4;X-14;X-15;X-16
+278,0.36435363690058387,true,4,Y-4;X-14;X-15;X-16
+279,0.36810829738775885,true,4,Y-4;X-14;X-15;X-16
+280,0.3888605982065201,true,4,Y-4;X-14;X-15;X-16
+281,0.34714781741301215,true,4,Y-4;X-14;X-15;X-16
+282,0.359450618426005,true,4,Y-4;X-14;X-15;X-16
+283,0.3845570832490921,true,4,Y-4;X-14;X-15;X-16
+284,0.3869556188583374,true,4,Y-4;X-14;X-15;X-16
+285,0.34323231875896454,true,4,Y-4;X-14;X-15;X-16
+286,0.348299965262413,true,4,Y-4;X-14;X-15;X-16
+287,0.34269658724466956,true,4,Y-4;X-14;X-15;X-16
+288,0.3378242154916128,true,4,Y-4;X-14;X-15;X-16
+289,0.3740089684724808,true,4,Y-4;X-14;X-15;X-16
+290,0.3476622998714447,true,4,Y-4;X-14;X-15;X-16
+291,0.3476293583710989,true,4,Y-4;X-14;X-15;X-16
+292,0.34939021368821466,true,4,Y-4;X-14;X-15;X-16
+293,0.33575011789798737,true,4,Y-4;X-14;X-15;X-16
+294,0.3457184433937073,true,4,Y-4;X-14;X-15;X-16
+295,0.38511334359645844,true,4,Y-4;X-14;X-15;X-16
+296,0.35279574493567145,true,4,Y-4;X-14;X-15;X-16
+297,0.34358036021391547,true,4,Y-4;X-14;X-15;X-16
+298,0.3502629200617472,true,4,Y-4;X-14;X-15;X-16
+299,0.3467387855052948,true,4,Y-4;X-14;X-15;X-16
+300,0.34392705063025153,true,4,Y-4;X-14;X-15;X-16
+301,0.34053896864255273,true,4,Y-4;X-14;X-15;X-16
+302,0.35219883918762207,true,4,Y-4;X-14;X-15;X-16
+303,0.3475775818030039,true,4,Y-4;X-14;X-15;X-16
+304,0.3614586343367895,true,4,Y-4;X-14;X-15;X-16
+305,0.3770411560932795,true,4,Y-4;X-14;X-15;X-16
+306,0.3752357612053553,true,4,Y-4;X-14;X-15;X-16
+307,0.3482293287913004,true,4,Y-4;X-14;X-15;X-16
+308,0.34796094397703803,true,4,Y-4;X-14;X-15;X-16
+309,0.38664020101229346,true,4,Y-4;X-14;X-15;X-16
+310,0.34728945295015967,true,4,Y-4;X-14;X-15;X-16
+311,0.3736526668071747,true,4,Y-4;X-14;X-15;X-16
+312,0.3479018211364746,true,4,Y-4;X-14;X-15;X-16
+313,0.35944311817487085,true,4,Y-4;X-14;X-15;X-16
+314,0.3531447649002075,true,4,Y-4;X-14;X-15;X-16
+315,0.38570666313171387,true,4,Y-4;X-14;X-15;X-16
+316,0.37387348214785254,true,4,Y-4;X-14;X-15;X-16
+317,0.3439298371473948,true,4,Y-4;X-14;X-15;X-16
+318,0.3353104988733927,true,4,Y-4;X-14;X-15;X-16
+319,0.35397067169348395,true,4,Y-4;X-14;X-15;X-16
+320,0.34497537215550744,true,4,Y-4;X-14;X-15;X-16
+321,0.3730275183916092,true,4,Y-4;X-14;X-15;X-16
+322,0.35320294400056207,true,4,Y-4;X-14;X-15;X-16
+323,0.34038640558719635,true,4,Y-4;X-14;X-15;X-16
+324,0.35159268478552497,true,4,Y-4;X-14;X-15;X-16
+325,0.3487641463677088,true,4,Y-4;X-14;X-15;X-16
+326,0.36885527769724524,true,4,Y-4;X-14;X-15;X-16
+327,0.3770240445931753,true,4,Y-4;X-14;X-15;X-16
+328,0.3373940040667852,true,4,Y-4;X-14;X-15;X-16
+329,0.3417827934026718,true,4,Y-4;X-14;X-15;X-16
+330,0.36818612118562066,true,4,Y-4;X-14;X-15;X-16
+331,0.38465508818626404,true,4,Y-4;X-14;X-15;X-16
+332,0.35987590750058496,true,4,Y-4;X-14;X-15;X-16
+333,0.3430445690949758,true,4,Y-4;X-14;X-15;X-16
+334,0.35079263647397363,true,4,Y-4;X-14;X-15;X-16
+335,0.353685071070989,true,4,Y-4;X-14;X-15;X-16
+336,0.37935538589954376,true,4,Y-4;X-14;X-15;X-16
+337,0.36845419804255164,true,4,Y-4;X-14;X-15;X-16
+338,0.35356859862804413,true,4,Y-4;X-14;X-15;X-16
+339,0.38415307799975074,true,4,Y-4;X-14;X-15;X-16
+340,0.3448419074217478,true,4,Y-4;X-14;X-15;X-16
+341,0.33807396392027533,true,4,Y-4;X-14;X-15;X-16
+342,0.3400692890087763,true,4,Y-4;X-14;X-15;X-16
+343,0.3474159588416418,true,4,Y-4;X-14;X-15;X-16
+344,0.34031561513741815,true,4,Y-4;X-14;X-15;X-16
+345,0.3524336864550909,true,4,Y-4;X-14;X-15;X-16
+346,0.3512922525405884,true,4,Y-4;X-14;X-15;X-16
+347,0.34686236580212915,true,4,Y-4;X-14;X-15;X-16
+348,0.35651628176371253,true,4,Y-4;X-14;X-15;X-16
+349,0.3492033779621124,true,4,Y-4;X-14;X-15;X-16
+350,0.33871320386727655,true,4,Y-4;X-14;X-15;X-16
+351,0.37716287871201837,true,4,Y-4;X-14;X-15;X-16
+352,0.3464714835087458,true,4,Y-4;X-14;X-15;X-16
+353,0.3731672714153925,true,4,Y-4;X-14;X-15;X-16
+354,0.36943192283312476,true,4,Y-4;X-14;X-15;X-16
+355,0.3616795291503271,true,4,Y-4;X-14;X-15;X-16
+356,0.37275543312231696,true,4,Y-4;X-14;X-15;X-16
+357,0.34717809657255805,true,4,Y-4;X-14;X-15;X-16
+358,0.34475871423880255,true,4,Y-4;X-14;X-15;X-16
+359,0.3459272483984629,true,4,Y-4;X-14;X-15;X-16
+360,0.3452642112970352,true,4,Y-4;X-14;X-15;X-16
+361,0.34677750865618384,true,4,Y-4;X-14;X-15;X-16
+362,0.3854369670152664,true,4,Y-4;X-14;X-15;X-16
+363,0.3514308830102285,true,4,Y-4;X-14;X-15;X-16
+364,0.35168311993281043,true,4,Y-4;X-14;X-15;X-16
+365,0.3759202708800634,true,4,Y-4;X-14;X-15;X-16
+366,0.36055166025956475,true,4,Y-4;X-14;X-15;X-16
+367,0.3768100539843241,true,4,Y-4;X-14;X-15;X-16
+368,0.3437922795613607,true,4,Y-4;X-14;X-15;X-16
+369,0.35572500030199683,true,4,Y-4;X-14;X-15;X-16
+370,0.3481576244036356,true,4,Y-4;X-14;X-15;X-16
+371,0.34677781661351526,true,4,Y-4;X-14;X-15;X-16
+372,0.3462307006120682,true,4,Y-4;X-14;X-15;X-16
+373,0.34408063193162286,true,4,Y-4;X-14;X-15;X-16
+374,0.3673677444458008,true,4,Y-4;X-14;X-15;X-16
+375,0.3577894320090612,true,4,Y-4;X-14;X-15;X-16
+376,0.34725103775660193,true,4,Y-4;X-14;X-15;X-16
+377,0.36311831076939904,true,4,Y-4;X-14;X-15;X-16
+378,0.37317823370297754,true,4,Y-4;X-14;X-15;X-16
+379,0.34453919529914856,true,4,Y-4;X-14;X-15;X-16
+380,0.34403005739053094,true,4,Y-4;X-14;X-15;X-16
+381,0.362069254120191,true,4,Y-4;X-14;X-15;X-16
+382,0.3786483357350031,true,4,Y-4;X-14;X-15;X-16
+383,0.37158315877119696,true,4,Y-4;X-14;X-15;X-16
+384,0.3579365611076355,true,4,Y-4;X-14;X-15;X-16
+385,0.3491609344879786,true,4,Y-4;X-14;X-15;X-16
+386,0.3504734287659327,true,4,Y-4;X-14;X-15;X-16
+387,0.352478489279747,true,4,Y-4;X-14;X-15;X-16
+388,0.37410006423791253,true,4,Y-4;X-14;X-15;X-16
+389,0.3443691978851954,true,4,Y-4;X-14;X-15;X-16
+390,0.35530663530031836,true,4,Y-4;X-14;X-15;X-16
+391,0.3454461544752121,true,4,Y-4;X-14;X-15;X-16
+392,0.3381874511639277,true,4,Y-4;X-14;X-15;X-16
+393,0.3424451450506846,true,4,Y-4;X-14;X-15;X-16
+394,0.3355673452218374,true,4,Y-4;X-14;X-15;X-16
+395,0.3489955017964045,true,4,Y-4;X-14;X-15;X-16
+396,0.34954857329527533,true,4,Y-4;X-14;X-15;X-16
+397,0.3521981487671534,true,4,Y-4;X-14;X-15;X-16
+398,0.3658656279246012,true,4,Y-4;X-14;X-15;X-16
+399,0.3798428177833557,true,4,Y-4;X-14;X-15;X-16
+400,0.369214211901029,true,4,Y-4;X-14;X-15;X-16
+401,0.3476307690143585,true,4,Y-4;X-14;X-15;X-16
+402,0.3536028365294138,true,4,Y-4;X-14;X-15;X-16
+403,0.35013577838738763,true,4,Y-4;X-14;X-15;X-16
+404,0.34336300690968835,true,4,Y-4;X-14;X-15;X-16
+405,0.34875816106796265,true,4,Y-4;X-14;X-15;X-16
+406,0.34492671986420953,true,4,Y-4;X-14;X-15;X-16
+407,0.36977139115333557,true,4,Y-4;X-14;X-15;X-16
+408,0.3459320912758509,true,4,Y-4;X-14;X-15;X-16
+409,0.3594076087077459,true,4,Y-4;X-14;X-15;X-16
+410,0.34356271723906195,true,4,Y-4;X-14;X-15;X-16
+411,0.3477467944224676,true,4,Y-4;X-14;X-15;X-16
+412,0.346897155046463,true,4,Y-4;X-14;X-15;X-16
+413,0.34138357639312744,true,4,Y-4;X-14;X-15;X-16
+414,0.3776817421118418,true,4,Y-4;X-14;X-15;X-16
+415,0.3462575376033783,true,4,Y-4;X-14;X-15;X-16
+416,0.3464354624350866,true,4,Y-4;X-14;X-15;X-16
+417,0.3485209047794342,true,4,Y-4;X-14;X-15;X-16
+418,0.356380358338356,true,4,Y-4;X-14;X-15;X-16
+419,0.3622211366891861,true,4,Y-4;X-14;X-15;X-16
+420,0.3802957087755203,true,4,Y-4;X-14;X-15;X-16
+421,0.3781160612901052,true,4,Y-4;X-14;X-15;X-16
+422,0.3568030347426733,true,4,Y-4;X-14;X-15;X-16
+423,0.35437195996443427,true,4,Y-4;X-14;X-15;X-16
+424,0.37449347476164496,true,4,Y-4;X-14;X-15;X-16
+425,0.3599476863940557,true,4,Y-4;X-14;X-15;X-16
+426,0.36587279538313544,true,4,Y-4;X-14;X-15;X-16
+427,0.34473858277002967,true,4,Y-4;X-14;X-15;X-16
+428,0.3718041181564331,true,4,Y-4;X-14;X-15;X-16
+429,0.34474337100982666,true,4,Y-4;X-14;X-15;X-16
+430,0.34659707049528754,true,4,Y-4;X-14;X-15;X-16
+431,0.3479718863964081,true,4,Y-4;X-14;X-15;X-16
+432,0.35053028662999475,true,4,Y-4;X-14;X-15;X-16
+433,0.3540815462668737,true,4,Y-4;X-14;X-15;X-16
+434,0.34361936151981354,true,4,Y-4;X-14;X-15;X-16
+435,0.37571406861146295,true,4,Y-4;X-14;X-15;X-16
+436,0.3649645745754242,true,4,Y-4;X-14;X-15;X-16
+437,0.34506329894065857,true,4,Y-4;X-14;X-15;X-16
+438,0.34225400288899743,true,4,Y-4;X-14;X-15;X-16
+439,0.3474563558896383,true,4,Y-4;X-14;X-15;X-16
+440,0.36787719527880347,true,4,Y-4;X-14;X-15;X-16
+441,0.3471689820289612,true,4,Y-4;X-14;X-15;X-16
+442,0.3479464302460352,true,4,Y-4;X-14;X-15;X-16
+443,0.38068240880966187,true,4,Y-4;X-14;X-15;X-16
+444,0.35062445203463233,true,4,Y-4;X-14;X-15;X-16
+445,0.354264035820961,true,4,Y-4;X-14;X-15;X-16
+446,0.3536220441261927,true,4,Y-4;X-14;X-15;X-16
+447,0.36916952331860864,true,4,Y-4;X-14;X-15;X-16
+448,0.3783888866504034,true,4,Y-4;X-14;X-15;X-16
+449,0.34361507991949714,true,4,Y-4;X-14;X-15;X-16
+450,0.3384132583936056,true,4,Y-4;X-14;X-15;X-16
+451,0.34414616227149963,true,4,Y-4;X-14;X-15;X-16
+452,0.3513202567895254,true,4,Y-4;X-14;X-15;X-16
+453,0.3832464963197708,true,4,Y-4;X-14;X-15;X-16
+454,0.3387235005696615,true,4,Y-4;X-14;X-15;X-16
+455,0.3844067851702372,true,4,Y-4;X-14;X-15;X-16
+456,0.3385346929232279,true,4,Y-4;X-14;X-15;X-16
+457,0.380063513914744,true,4,Y-4;X-14;X-15;X-16
+458,0.35745571057001746,true,4,Y-4;X-14;X-15;X-16
+459,0.38174636165301,true,4,Y-4;X-14;X-15;X-16
+460,0.37748314440250397,true,4,Y-4;X-14;X-15;X-16
+461,0.35280184944470727,true,4,Y-4;X-14;X-15;X-16
+462,0.3719876011212667,true,4,Y-4;X-14;X-15;X-16
+463,0.3541833211978277,true,4,Y-4;X-14;X-15;X-16
+464,0.35079393784205115,true,4,Y-4;X-14;X-15;X-16
+465,0.3714203735192617,true,4,Y-4;X-14;X-15;X-16
+466,0.3437741746505102,true,4,Y-4;X-14;X-15;X-16
+467,0.34273725251356757,true,4,Y-4;X-14;X-15;X-16
+468,0.34443433582782745,true,4,Y-4;X-14;X-15;X-16
+469,0.3401193171739578,true,4,Y-4;X-14;X-15;X-16
+470,0.3783543755610784,true,4,Y-4;X-14;X-15;X-16
+471,0.35678810377915704,true,4,Y-4;X-14;X-15;X-16
+472,0.34605470299720764,true,4,Y-4;X-14;X-15;X-16
+473,0.3520689904689789,true,4,Y-4;X-14;X-15;X-16
+474,0.35699044167995453,true,4,Y-4;X-14;X-15;X-16
+475,0.34458105762799585,true,4,Y-4;X-14;X-15;X-16
+476,0.3486201812823614,true,4,Y-4;X-14;X-15;X-16
+477,0.3526142040888468,true,4,Y-4;X-14;X-15;X-16
+478,0.36300720274448395,true,4,Y-4;X-14;X-15;X-16
+479,0.33719394604365027,true,4,Y-4;X-14;X-15;X-16
+480,0.34897851943969727,true,4,Y-4;X-14;X-15;X-16
+481,0.35256916781266534,true,4,Y-4;X-14;X-15;X-16
+482,0.34800294538338983,true,4,Y-4;X-14;X-15;X-16
+483,0.37704024712244666,true,4,Y-4;X-14;X-15;X-16
+484,0.35326315462589264,true,4,Y-4;X-14;X-15;X-16
+485,0.35626330971717834,true,4,Y-4;X-14;X-15;X-16
+486,0.3522376914819082,true,4,Y-4;X-14;X-15;X-16
+487,0.3575465728839239,true,4,Y-4;X-14;X-15;X-16
+488,0.36018089950084686,true,4,Y-4;X-14;X-15;X-16
+489,0.34959494074185693,true,4,Y-4;X-14;X-15;X-16
+490,0.343854159116745,true,4,Y-4;X-14;X-15;X-16
+491,0.3493368774652481,true,4,Y-4;X-14;X-15;X-16
+492,0.3809196650981903,true,4,Y-4;X-14;X-15;X-16
+493,0.34368080397446954,true,4,Y-4;X-14;X-15;X-16
+494,0.37078663210074103,true,4,Y-4;X-14;X-15;X-16
+495,0.3694220334291458,true,4,Y-4;X-14;X-15;X-16
+496,0.34718597431977594,true,4,Y-4;X-14;X-15;X-16
+497,0.37630823254585266,true,4,Y-4;X-14;X-15;X-16
+498,0.34428847332795465,true,4,Y-4;X-14;X-15;X-16
+499,0.3428070147832235,true,4,Y-4;X-14;X-15;X-16
+500,0.37927844623724616,true,4,Y-4;X-14;X-15;X-16

splits/manifest.json

@@ -0,0 +1,3061 @@
+{
+    "full_train": [
+        "run_1",
+        "run_2",
+        "run_3",
+        "run_5",
+        "run_6",
+        "run_7",
+        "run_8",
+        "run_9",
+        "run_10",
+        "run_13",
+        "run_14",
+        "run_15",
+        "run_16",
+        "run_17",
+        "run_18",
+        "run_21",
+        "run_23",
+        "run_25",
+        "run_27",
+        "run_28",
+        "run_30",
+        "run_31",
+        "run_32",
+        "run_33",
+        "run_34",
+        "run_35",
+        "run_36",
+        "run_37",
+        "run_39",
+        "run_40",
+        "run_42",
+        "run_43",
+        "run_44",
+        "run_45",
+        "run_46",
+        "run_47",
+        "run_48",
+        "run_49",
+        "run_50",
+        "run_51",
+        "run_52",
+        "run_53",
+        "run_54",
+        "run_57",
+        "run_58",
+        "run_60",
+        "run_61",
+        "run_62",
+        "run_63",
+        "run_64",
+        "run_65",
+        "run_66",
+        "run_67",
+        "run_68",
+        "run_69",
+        "run_70",
+        "run_71",
+        "run_72",
+        "run_73",
+        "run_74",
+        "run_75",
+        "run_76",
+        "run_77",
+        "run_78",
+        "run_79",
+        "run_80",
+        "run_81",
+        "run_82",
+        "run_83",
+        "run_84",
+        "run_85",
+        "run_86",
+        "run_87",
+        "run_88",
+        "run_89",
+        "run_90",
+        "run_91",
+        "run_92",
+        "run_93",
+        "run_94",
+        "run_95",
+        "run_96",
+        "run_97",
+        "run_98",
+        "run_99",
+        "run_100",
+        "run_101",
+        "run_102",
+        "run_103",
+        "run_105",
+        "run_106",
+        "run_107",
+        "run_110",
+        "run_111",
+        "run_112",
+        "run_113",
+        "run_114",
+        "run_115",
+        "run_116",
+        "run_117",
+        "run_118",
+        "run_119",
+        "run_120",
+        "run_121",
+        "run_122",
+        "run_123",
+        "run_125",
+        "run_126",
+        "run_128",
+        "run_129",
+        "run_130",
+        "run_131",
+        "run_132",
+        "run_134",
+        "run_135",
+        "run_136",
+        "run_137",
+        "run_138",
+        "run_139",
+        "run_140",
+        "run_141",
+        "run_143",
+        "run_144",
+        "run_145",
+        "run_146",
+        "run_147",
+        "run_148",
+        "run_149",
+        "run_151",
+        "run_152",
+        "run_153",
+        "run_154",
+        "run_155",
+        "run_156",
+        "run_157",
+        "run_159",
+        "run_160",
+        "run_161",
+        "run_162",
+        "run_163",
+        "run_164",
+        "run_166",
+        "run_167",
+        "run_168",
+        "run_169",
+        "run_170",
+        "run_171",
+        "run_172",
+        "run_174",
+        "run_175",
+        "run_176",
+        "run_178",
+        "run_179",
+        "run_181",
+        "run_182",
+        "run_183",
+        "run_184",
+        "run_185",
+        "run_186",
+        "run_189",
+        "run_190",
+        "run_192",
+        "run_193",
+        "run_194",
+        "run_195",
+        "run_198",
+        "run_200",
+        "run_201",
+        "run_202",
+        "run_204",
+        "run_206",
+        "run_209",
+        "run_211",
+        "run_212",
+        "run_213",
+        "run_214",
+        "run_216",
+        "run_217",
+        "run_218",
+        "run_219",
+        "run_220",
+        "run_221",
+        "run_223",
+        "run_224",
+        "run_225",
+        "run_227",
+        "run_229",
+        "run_231",
+        "run_232",
+        "run_233",
+        "run_235",
+        "run_236",
+        "run_237",
+        "run_238",
+        "run_239",
+        "run_240",
+        "run_242",
+        "run_243",
+        "run_244",
+        "run_245",
+        "run_246",
+        "run_248",
+        "run_249",
+        "run_250",
+        "run_251",
+        "run_254",
+        "run_255",
+        "run_256",
+        "run_257",
+        "run_259",
+        "run_261",
+        "run_262",
+        "run_264",
+        "run_265",
+        "run_266",
+        "run_267",
+        "run_268",
+        "run_269",
+        "run_270",
+        "run_272",
+        "run_273",
+        "run_274",
+        "run_276",
+        "run_277",
+        "run_278",
+        "run_279",
+        "run_282",
+        "run_283",
+        "run_285",
+        "run_286",
+        "run_287",
+        "run_288",
+        "run_289",
+        "run_292",
+        "run_293",
+        "run_294",
+        "run_296",
+        "run_297",
+        "run_299",
+        "run_300",
+        "run_301",
+        "run_302",
+        "run_305",
+        "run_306",
+        "run_307",
+        "run_308",
+        "run_309",
+        "run_310",
+        "run_313",
+        "run_314",
+        "run_315",
+        "run_316",
+        "run_317",
+        "run_318",
+        "run_319",
+        "run_320",
+        "run_323",
+        "run_325",
+        "run_326",
+        "run_327",
+        "run_330",
+        "run_331",
+        "run_332",
+        "run_333",
+        "run_334",
+        "run_335",
+        "run_336",
+        "run_338",
+        "run_339",
+        "run_340",
+        "run_342",
+        "run_343",
+        "run_344",
+        "run_345",
+        "run_346",
+        "run_347",
+        "run_348",
+        "run_349",
+        "run_351",
+        "run_353",
+        "run_355",
+        "run_356",
+        "run_357",
+        "run_358",
+        "run_359",
+        "run_360",
+        "run_361",
+        "run_362",
+        "run_365",
+        "run_367",
+        "run_368",
+        "run_369",
+        "run_370",
+        "run_371",
+        "run_373",
+        "run_374",
+        "run_375",
+        "run_377",
+        "run_378",
+        "run_379",
+        "run_381",
+        "run_383",
+        "run_384",
+        "run_385",
+        "run_386",
+        "run_388",
+        "run_389",
+        "run_391",
+        "run_392",
+        "run_393",
+        "run_394",
+        "run_395",
+        "run_396",
+        "run_397",
+        "run_398",
+        "run_399",
+        "run_400",
+        "run_402",
+        "run_404",
+        "run_406",
+        "run_407",
+        "run_408",
+        "run_409",
+        "run_411",
+        "run_412",
+        "run_413",
+        "run_414",
+        "run_415",
+        "run_416",
+        "run_417",
+        "run_418",
+        "run_419",
+        "run_420",
+        "run_421",
+        "run_422",
+        "run_425",
+        "run_426",
+        "run_427",
+        "run_430",
+        "run_431",
+        "run_433",
+        "run_434",
+        "run_435",
+        "run_437",
+        "run_438",
+        "run_439",
+        "run_440",
+        "run_442",
+        "run_443",
+        "run_444",
+        "run_445",
+        "run_446",
+        "run_448",
+        "run_449",
+        "run_450",
+        "run_451",
+        "run_452",
+        "run_453",
+        "run_456",
+        "run_457",
+        "run_458",
+        "run_459",
+        "run_460",
+        "run_461",
+        "run_462",
+        "run_463",
+        "run_464",
+        "run_465",
+        "run_466",
+        "run_467",
+        "run_468",
+        "run_469",
+        "run_470",
+        "run_471",
+        "run_473",
+        "run_474",
+        "run_475",
+        "run_476",
+        "run_477",
+        "run_478",
+        "run_479",
+        "run_480",
+        "run_481",
+        "run_482",
+        "run_483",
+        "run_484",
+        "run_485",
+        "run_486",
+        "run_488",
+        "run_489",
+        "run_490",
+        "run_491",
+        "run_492",
+        "run_493",
+        "run_495",
+        "run_496",
+        "run_497",
+        "run_498",
+        "run_499",
+        "run_500"
+    ],
+    "full_val": [
+        "run_24",
+        "run_26",
+        "run_29",
+        "run_38",
+        "run_41",
+        "run_55",
+        "run_59",
+        "run_104",
+        "run_108",
+        "run_124",
+        "run_133",
+        "run_142",
+        "run_158",
+        "run_173",
+        "run_180",
+        "run_188",
+        "run_196",
+        "run_197",
+        "run_199",
+        "run_205",
+        "run_207",
+        "run_210",
+        "run_222",
+        "run_226",
+        "run_230",
+        "run_258",
+        "run_263",
+        "run_280",
+        "run_281",
+        "run_284",
+        "run_290",
+        "run_291",
+        "run_295",
+        "run_304",
+        "run_312",
+        "run_337",
+        "run_350",
+        "run_354",
+        "run_363",
+        "run_372",
+        "run_387",
+        "run_403",
+        "run_405",
+        "run_424",
+        "run_428",
+        "run_429",
+        "run_432",
+        "run_455",
+        "run_472",
+        "run_494"
+    ],
+    "full_test": [
+        "run_4",
+        "run_11",
+        "run_12",
+        "run_19",
+        "run_20",
+        "run_22",
+        "run_56",
+        "run_109",
+        "run_127",
+        "run_150",
+        "run_165",
+        "run_177",
+        "run_187",
+        "run_191",
+        "run_203",
+        "run_208",
+        "run_215",
+        "run_228",
+        "run_234",
+        "run_241",
+        "run_247",
+        "run_252",
+        "run_253",
+        "run_260",
+        "run_271",
+        "run_275",
+        "run_298",
+        "run_303",
+        "run_311",
+        "run_321",
+        "run_322",
+        "run_324",
+        "run_328",
+        "run_329",
+        "run_341",
+        "run_352",
+        "run_364",
+        "run_366",
+        "run_376",
+        "run_380",
+        "run_382",
+        "run_390",
+        "run_401",
+        "run_410",
+        "run_423",
+        "run_436",
+        "run_441",
+        "run_447",
+        "run_454",
+        "run_487"
+    ],
+    "medium_train": [
+        "run_8",
+        "run_14",
+        "run_18",
+        "run_23",
+        "run_25",
+        "run_34",
+        "run_44",
+        "run_47",
+        "run_50",
+        "run_52",
+        "run_60",
+        "run_62",
+        "run_63",
+        "run_64",
+        "run_65",
+        "run_69",
+        "run_77",
+        "run_78",
+        "run_81",
+        "run_86",
+        "run_91",
+        "run_92",
+        "run_93",
+        "run_95",
+        "run_96",
+        "run_98",
+        "run_102",
+        "run_114",
+        "run_115",
+        "run_120",
+        "run_122",
+        "run_123",
+        "run_130",
+        "run_131",
+        "run_137",
+        "run_138",
+        "run_139",
+        "run_141",
+        "run_149",
+        "run_152",
+        "run_154",
+        "run_156",
+        "run_159",
+        "run_161",
+        "run_162",
+        "run_164",
+        "run_166",
+        "run_167",
+        "run_174",
+        "run_176",
+        "run_178",
+        "run_183",
+        "run_193",
+        "run_200",
+        "run_216",
+        "run_220",
+        "run_225",
+        "run_233",
+        "run_237",
+        "run_240",
+        "run_244",
+        "run_256",
+        "run_259",
+        "run_261",
+        "run_264",
+        "run_266",
+        "run_268",
+        "run_269",
+        "run_273",
+        "run_274",
+        "run_278",
+        "run_286",
+        "run_288",
+        "run_289",
+        "run_293",
+        "run_297",
+        "run_300",
+        "run_309",
+        "run_314",
+        "run_315",
+        "run_318",
+        "run_319",
+        "run_326",
+        "run_330",
+        "run_339",
+        "run_340",
+        "run_344",
+        "run_346",
+        "run_349",
+        "run_351",
+        "run_353",
+        "run_360",
+        "run_361",
+        "run_362",
+        "run_367",
+        "run_369",
+        "run_370",
+        "run_375",
+        "run_385",
+        "run_386",
+        "run_388",
+        "run_389",
+        "run_394",
+        "run_404",
+        "run_407",
+        "run_409",
+        "run_411",
+        "run_413",
+        "run_414",
+        "run_416",
+        "run_419",
+        "run_421",
+        "run_422",
+        "run_427",
+        "run_437",
+        "run_439",
+        "run_440",
+        "run_442",
+        "run_445",
+        "run_450",
+        "run_451",
+        "run_457",
+        "run_469",
+        "run_476",
+        "run_479",
+        "run_482",
+        "run_483",
+        "run_485",
+        "run_490",
+        "run_493",
+        "run_497",
+        "run_499",
+        "run_500"
+    ],
+    "medium_val": [
+        "run_24",
+        "run_26",
+        "run_29",
+        "run_38",
+        "run_41",
+        "run_55",
+        "run_59",
+        "run_104",
+        "run_108",
+        "run_124",
+        "run_133",
+        "run_142",
+        "run_158",
+        "run_173",
+        "run_180",
+        "run_188",
+        "run_196",
+        "run_197",
+        "run_199",
+        "run_205",
+        "run_207",
+        "run_210",
+        "run_222",
+        "run_226",
+        "run_230",
+        "run_258",
+        "run_263",
+        "run_280",
+        "run_281",
+        "run_284",
+        "run_290",
+        "run_291",
+        "run_295",
+        "run_304",
+        "run_312",
+        "run_337",
+        "run_350",
+        "run_354",
+        "run_363",
+        "run_372",
+        "run_387",
+        "run_403",
+        "run_405",
+        "run_424",
+        "run_428",
+        "run_429",
+        "run_432",
+        "run_455",
+        "run_472",
+        "run_494"
+    ],
+    "medium_test": [
+        "run_4",
+        "run_11",
+        "run_12",
+        "run_19",
+        "run_20",
+        "run_22",
+        "run_56",
+        "run_109",
+        "run_127",
+        "run_150",
+        "run_165",
+        "run_177",
+        "run_187",
+        "run_191",
+        "run_203",
+        "run_208",
+        "run_215",
+        "run_228",
+        "run_234",
+        "run_241",
+        "run_247",
+        "run_252",
+        "run_253",
+        "run_260",
+        "run_271",
+        "run_275",
+        "run_298",
+        "run_303",
+        "run_311",
+        "run_321",
+        "run_322",
+        "run_324",
+        "run_328",
+        "run_329",
+        "run_341",
+        "run_352",
+        "run_364",
+        "run_366",
+        "run_376",
+        "run_380",
+        "run_382",
+        "run_390",
+        "run_401",
+        "run_410",
+        "run_423",
+        "run_436",
+        "run_441",
+        "run_447",
+        "run_454",
+        "run_487"
+    ],
+    "scarce_train": [
+        "run_8",
+        "run_14",
+        "run_18",
+        "run_25",
+        "run_34",
+        "run_50",
+        "run_52",
+        "run_69",
+        "run_78",
+        "run_81",
+        "run_86",
+        "run_91",
+        "run_92",
+        "run_93",
+        "run_95",
+        "run_96",
+        "run_98",
+        "run_114",
+        "run_120",
+        "run_122",
+        "run_137",
+        "run_139",
+        "run_152",
+        "run_164",
+        "run_166",
+        "run_167",
+        "run_176",
+        "run_183",
+        "run_200",
+        "run_220",
+        "run_225",
+        "run_237",
+        "run_244",
+        "run_259",
+        "run_261",
+        "run_268",
+        "run_274",
+        "run_286",
+        "run_297",
+        "run_314",
+        "run_315",
+        "run_318",
+        "run_339",
+        "run_340",
+        "run_349",
+        "run_351",
+        "run_353",
+        "run_362",
+        "run_367",
+        "run_369",
+        "run_375",
+        "run_386",
+        "run_389",
+        "run_394",
+        "run_409",
+        "run_416",
+        "run_419",
+        "run_439",
+        "run_440",
+        "run_442",
+        "run_450",
+        "run_479",
+        "run_482",
+        "run_490",
+        "run_493",
+        "run_497",
+        "run_500"
+    ],
+    "scarce_val": [
+        "run_24",
+        "run_26",
+        "run_29",
+        "run_38",
+        "run_41",
+        "run_55",
+        "run_59",
+        "run_104",
+        "run_108",
+        "run_124",
+        "run_133",
+        "run_142",
+        "run_158",
+        "run_173",
+        "run_180",
+        "run_188",
+        "run_196",
+        "run_197",
+        "run_199",
+        "run_205",
+        "run_207",
+        "run_210",
+        "run_222",
+        "run_226",
+        "run_230",
+        "run_258",
+        "run_263",
+        "run_280",
+        "run_281",
+        "run_284",
+        "run_290",
+        "run_291",
+        "run_295",
+        "run_304",
+        "run_312",
+        "run_337",
+        "run_350",
+        "run_354",
+        "run_363",
+        "run_372",
+        "run_387",
+        "run_403",
+        "run_405",
+        "run_424",
+        "run_428",
+        "run_429",
+        "run_432",
+        "run_455",
+        "run_472",
+        "run_494"
+    ],
+    "scarce_test": [
+        "run_4",
+        "run_11",
+        "run_12",
+        "run_19",
+        "run_20",
+        "run_22",
+        "run_56",
+        "run_109",
+        "run_127",
+        "run_150",
+        "run_165",
+        "run_177",
+        "run_187",
+        "run_191",
+        "run_203",
+        "run_208",
+        "run_215",
+        "run_228",
+        "run_234",
+        "run_241",
+        "run_247",
+        "run_252",
+        "run_253",
+        "run_260",
+        "run_271",
+        "run_275",
+        "run_298",
+        "run_303",
+        "run_311",
+        "run_321",
+        "run_322",
+        "run_324",
+        "run_328",
+        "run_329",
+        "run_341",
+        "run_352",
+        "run_364",
+        "run_366",
+        "run_376",
+        "run_380",
+        "run_382",
+        "run_390",
+        "run_401",
+        "run_410",
+        "run_423",
+        "run_436",
+        "run_441",
+        "run_447",
+        "run_454",
+        "run_487"
+    ],
+    "super_scarce_train": [
+        "run_50",
+        "run_93",
+        "run_120",
+        "run_152",
+        "run_220",
+        "run_297",
+        "run_362",
+        "run_409",
+        "run_416",
+        "run_440",
+        "run_442"
+    ],
+    "super_scarce_val": [
+        "run_24",
+        "run_26",
+        "run_29",
+        "run_38",
+        "run_41",
+        "run_55",
+        "run_59",
+        "run_104",
+        "run_108",
+        "run_124",
+        "run_133",
+        "run_142",
+        "run_158",
+        "run_173",
+        "run_180",
+        "run_188",
+        "run_196",
+        "run_197",
+        "run_199",
+        "run_205",
+        "run_207",
+        "run_210",
+        "run_222",
+        "run_226",
+        "run_230",
+        "run_258",
+        "run_263",
+        "run_280",
+        "run_281",
+        "run_284",
+        "run_290",
+        "run_291",
+        "run_295",
+        "run_304",
+        "run_312",
+        "run_337",
+        "run_350",
+        "run_354",
+        "run_363",
+        "run_372",
+        "run_387",
+        "run_403",
+        "run_405",
+        "run_424",
+        "run_428",
+        "run_429",
+        "run_432",
+        "run_455",
+        "run_472",
+        "run_494"
+    ],
+    "super_scarce_test": [
+        "run_4",
+        "run_11",
+        "run_12",
+        "run_19",
+        "run_20",
+        "run_22",
+        "run_56",
+        "run_109",
+        "run_127",
+        "run_150",
+        "run_165",
+        "run_177",
+        "run_187",
+        "run_191",
+        "run_203",
+        "run_208",
+        "run_215",
+        "run_228",
+        "run_234",
+        "run_241",
+        "run_247",
+        "run_252",
+        "run_253",
+        "run_260",
+        "run_271",
+        "run_275",
+        "run_298",
+        "run_303",
+        "run_311",
+        "run_321",
+        "run_322",
+        "run_324",
+        "run_328",
+        "run_329",
+        "run_341",
+        "run_352",
+        "run_364",
+        "run_366",
+        "run_376",
+        "run_380",
+        "run_382",
+        "run_390",
+        "run_401",
+        "run_410",
+        "run_423",
+        "run_436",
+        "run_441",
+        "run_447",
+        "run_454",
+        "run_487"
+    ],
+    "high_drag_train": [
+        "run_1",
+        "run_2",
+        "run_3",
+        "run_5",
+        "run_7",
+        "run_10",
+        "run_11",
+        "run_15",
+        "run_16",
+        "run_17",
+        "run_19",
+        "run_20",
+        "run_21",
+        "run_22",
+        "run_24",
+        "run_25",
+        "run_27",
+        "run_28",
+        "run_29",
+        "run_31",
+        "run_32",
+        "run_33",
+        "run_35",
+        "run_37",
+        "run_39",
+        "run_40",
+        "run_42",
+        "run_43",
+        "run_44",
+        "run_45",
+        "run_46",
+        "run_48",
+        "run_50",
+        "run_51",
+        "run_53",
+        "run_55",
+        "run_56",
+        "run_57",
+        "run_59",
+        "run_61",
+        "run_62",
+        "run_64",
+        "run_65",
+        "run_67",
+        "run_68",
+        "run_69",
+        "run_70",
+        "run_71",
+        "run_72",
+        "run_73",
+        "run_75",
+        "run_77",
+        "run_78",
+        "run_81",
+        "run_84",
+        "run_85",
+        "run_86",
+        "run_87",
+        "run_89",
+        "run_90",
+        "run_91",
+        "run_92",
+        "run_93",
+        "run_95",
+        "run_97",
+        "run_99",
+        "run_100",
+        "run_101",
+        "run_102",
+        "run_105",
+        "run_106",
+        "run_108",
+        "run_109",
+        "run_110",
+        "run_111",
+        "run_113",
+        "run_114",
+        "run_115",
+        "run_116",
+        "run_117",
+        "run_118",
+        "run_119",
+        "run_121",
+        "run_123",
+        "run_124",
+        "run_125",
+        "run_126",
+        "run_127",
+        "run_128",
+        "run_129",
+        "run_131",
+        "run_132",
+        "run_134",
+        "run_136",
+        "run_139",
+        "run_140",
+        "run_142",
+        "run_143",
+        "run_144",
+        "run_145",
+        "run_146",
+        "run_147",
+        "run_149",
+        "run_150",
+        "run_151",
+        "run_154",
+        "run_155",
+        "run_156",
+        "run_157",
+        "run_158",
+        "run_159",
+        "run_160",
+        "run_162",
+        "run_163",
+        "run_164",
+        "run_165",
+        "run_166",
+        "run_167",
+        "run_169",
+        "run_170",
+        "run_172",
+        "run_174",
+        "run_175",
+        "run_176",
+        "run_177",
+        "run_179",
+        "run_181",
+        "run_182",
+        "run_183",
+        "run_184",
+        "run_186",
+        "run_187",
+        "run_188",
+        "run_189",
+        "run_192",
+        "run_193",
+        "run_194",
+        "run_195",
+        "run_196",
+        "run_198",
+        "run_199",
+        "run_200",
+        "run_202",
+        "run_204",
+        "run_206",
+        "run_207",
+        "run_208",
+        "run_209",
+        "run_210",
+        "run_211",
+        "run_212",
+        "run_213",
+        "run_214",
+        "run_218",
+        "run_219",
+        "run_221",
+        "run_224",
+        "run_225",
+        "run_226",
+        "run_227",
+        "run_229",
+        "run_230",
+        "run_231",
+        "run_232",
+        "run_233",
+        "run_234",
+        "run_235",
+        "run_237",
+        "run_238",
+        "run_239",
+        "run_240",
+        "run_241",
+        "run_245",
+        "run_247",
+        "run_249",
+        "run_251",
+        "run_252",
+        "run_255",
+        "run_256",
+        "run_257",
+        "run_258",
+        "run_259",
+        "run_261",
+        "run_263",
+        "run_265",
+        "run_267",
+        "run_268",
+        "run_269",
+        "run_270",
+        "run_271",
+        "run_272",
+        "run_273",
+        "run_276",
+        "run_278",
+        "run_279",
+        "run_281",
+        "run_282",
+        "run_285",
+        "run_286",
+        "run_288",
+        "run_289",
+        "run_291",
+        "run_294",
+        "run_296",
+        "run_298",
+        "run_299",
+        "run_300",
+        "run_301",
+        "run_302",
+        "run_303",
+        "run_304",
+        "run_307",
+        "run_311",
+        "run_312",
+        "run_313",
+        "run_316",
+        "run_317",
+        "run_321",
+        "run_322",
+        "run_324",
+        "run_325",
+        "run_326",
+        "run_327",
+        "run_329",
+        "run_330",
+        "run_332",
+        "run_334",
+        "run_335",
+        "run_338",
+        "run_343",
+        "run_344",
+        "run_345",
+        "run_346",
+        "run_347",
+        "run_348",
+        "run_350",
+        "run_352",
+        "run_353",
+        "run_354",
+        "run_355",
+        "run_356",
+        "run_358",
+        "run_359",
+        "run_360",
+        "run_361",
+        "run_363",
+        "run_365",
+        "run_366",
+        "run_367",
+        "run_368",
+        "run_369",
+        "run_370",
+        "run_371",
+        "run_373",
+        "run_374",
+        "run_375",
+        "run_377",
+        "run_378",
+        "run_379",
+        "run_380",
+        "run_381",
+        "run_382",
+        "run_384",
+        "run_385",
+        "run_386",
+        "run_387",
+        "run_389",
+        "run_390",
+        "run_391",
+        "run_393",
+        "run_395",
+        "run_396",
+        "run_397",
+        "run_398",
+        "run_399",
+        "run_400",
+        "run_401",
+        "run_402",
+        "run_403",
+        "run_405",
+        "run_406",
+        "run_407",
+        "run_408",
+        "run_409",
+        "run_410",
+        "run_412",
+        "run_414",
+        "run_416",
+        "run_417",
+        "run_418",
+        "run_419",
+        "run_422",
+        "run_424",
+        "run_425",
+        "run_427",
+        "run_428",
+        "run_429",
+        "run_430",
+        "run_431",
+        "run_432",
+        "run_433",
+        "run_434",
+        "run_435",
+        "run_436",
+        "run_437",
+        "run_438",
+        "run_439",
+        "run_440",
+        "run_441",
+        "run_442",
+        "run_445",
+        "run_446",
+        "run_447",
+        "run_448",
+        "run_449",
+        "run_450",
+        "run_452",
+        "run_457",
+        "run_458",
+        "run_460",
+        "run_462",
+        "run_463",
+        "run_464",
+        "run_465",
+        "run_466",
+        "run_469",
+        "run_470",
+        "run_472",
+        "run_473",
+        "run_474",
+        "run_475",
+        "run_476",
+        "run_477",
+        "run_478",
+        "run_479",
+        "run_481",
+        "run_483",
+        "run_485",
+        "run_486",
+        "run_487",
+        "run_488",
+        "run_489",
+        "run_492",
+        "run_493",
+        "run_494",
+        "run_495",
+        "run_496",
+        "run_497",
+        "run_498",
+        "run_500"
+    ],
+    "high_drag_val": [
+        "run_30",
+        "run_36",
+        "run_47",
+        "run_49",
+        "run_58",
+        "run_63",
+        "run_76",
+        "run_83",
+        "run_88",
+        "run_103",
+        "run_112",
+        "run_122",
+        "run_130",
+        "run_133",
+        "run_137",
+        "run_138",
+        "run_153",
+        "run_161",
+        "run_191",
+        "run_201",
+        "run_203",
+        "run_222",
+        "run_236",
+        "run_246",
+        "run_248",
+        "run_253",
+        "run_254",
+        "run_260",
+        "run_262",
+        "run_275",
+        "run_277",
+        "run_292",
+        "run_308",
+        "run_320",
+        "run_337",
+        "run_340",
+        "run_341",
+        "run_349",
+        "run_351",
+        "run_357",
+        "run_364",
+        "run_372",
+        "run_388",
+        "run_423",
+        "run_426",
+        "run_461",
+        "run_484",
+        "run_490",
+        "run_491",
+        "run_499"
+    ],
+    "high_drag_test": [
+        "run_4",
+        "run_6",
+        "run_8",
+        "run_9",
+        "run_12",
+        "run_13",
+        "run_14",
+        "run_18",
+        "run_23",
+        "run_26",
+        "run_34",
+        "run_38",
+        "run_41",
+        "run_52",
+        "run_54",
+        "run_60",
+        "run_66",
+        "run_74",
+        "run_79",
+        "run_80",
+        "run_82",
+        "run_94",
+        "run_96",
+        "run_98",
+        "run_104",
+        "run_107",
+        "run_120",
+        "run_135",
+        "run_141",
+        "run_148",
+        "run_152",
+        "run_168",
+        "run_171",
+        "run_173",
+        "run_178",
+        "run_180",
+        "run_185",
+        "run_190",
+        "run_197",
+        "run_205",
+        "run_215",
+        "run_216",
+        "run_217",
+        "run_220",
+        "run_223",
+        "run_228",
+        "run_242",
+        "run_243",
+        "run_244",
+        "run_250",
+        "run_264",
+        "run_266",
+        "run_274",
+        "run_280",
+        "run_283",
+        "run_284",
+        "run_287",
+        "run_290",
+        "run_293",
+        "run_295",
+        "run_297",
+        "run_305",
+        "run_306",
+        "run_309",
+        "run_310",
+        "run_314",
+        "run_315",
+        "run_318",
+        "run_319",
+        "run_323",
+        "run_328",
+        "run_331",
+        "run_333",
+        "run_336",
+        "run_339",
+        "run_342",
+        "run_362",
+        "run_376",
+        "run_383",
+        "run_392",
+        "run_394",
+        "run_404",
+        "run_411",
+        "run_413",
+        "run_415",
+        "run_420",
+        "run_421",
+        "run_443",
+        "run_444",
+        "run_451",
+        "run_453",
+        "run_454",
+        "run_455",
+        "run_456",
+        "run_459",
+        "run_467",
+        "run_468",
+        "run_471",
+        "run_480",
+        "run_482"
+    ],
+    "low_drag_train": [
+        "run_2",
+        "run_5",
+        "run_6",
+        "run_9",
+        "run_11",
+        "run_12",
+        "run_13",
+        "run_14",
+        "run_16",
+        "run_17",
+        "run_18",
+        "run_19",
+        "run_21",
+        "run_22",
+        "run_23",
+        "run_24",
+        "run_25",
+        "run_26",
+        "run_28",
+        "run_29",
+        "run_30",
+        "run_31",
+        "run_34",
+        "run_35",
+        "run_38",
+        "run_40",
+        "run_41",
+        "run_42",
+        "run_47",
+        "run_49",
+        "run_50",
+        "run_51",
+        "run_52",
+        "run_53",
+        "run_54",
+        "run_57",
+        "run_58",
+        "run_59",
+        "run_60",
+        "run_64",
+        "run_65",
+        "run_66",
+        "run_67",
+        "run_70",
+        "run_71",
+        "run_72",
+        "run_74",
+        "run_75",
+        "run_76",
+        "run_77",
+        "run_79",
+        "run_80",
+        "run_81",
+        "run_82",
+        "run_84",
+        "run_86",
+        "run_87",
+        "run_88",
+        "run_89",
+        "run_90",
+        "run_91",
+        "run_92",
+        "run_94",
+        "run_95",
+        "run_96",
+        "run_97",
+        "run_98",
+        "run_100",
+        "run_102",
+        "run_104",
+        "run_105",
+        "run_107",
+        "run_108",
+        "run_109",
+        "run_111",
+        "run_113",
+        "run_115",
+        "run_118",
+        "run_119",
+        "run_120",
+        "run_125",
+        "run_127",
+        "run_128",
+        "run_130",
+        "run_131",
+        "run_132",
+        "run_133",
+        "run_134",
+        "run_136",
+        "run_137",
+        "run_138",
+        "run_140",
+        "run_141",
+        "run_142",
+        "run_143",
+        "run_145",
+        "run_146",
+        "run_147",
+        "run_148",
+        "run_149",
+        "run_152",
+        "run_153",
+        "run_155",
+        "run_156",
+        "run_160",
+        "run_161",
+        "run_162",
+        "run_163",
+        "run_164",
+        "run_166",
+        "run_170",
+        "run_171",
+        "run_172",
+        "run_173",
+        "run_174",
+        "run_177",
+        "run_178",
+        "run_179",
+        "run_180",
+        "run_181",
+        "run_182",
+        "run_183",
+        "run_184",
+        "run_185",
+        "run_186",
+        "run_187",
+        "run_188",
+        "run_190",
+        "run_191",
+        "run_192",
+        "run_194",
+        "run_195",
+        "run_196",
+        "run_200",
+        "run_203",
+        "run_204",
+        "run_205",
+        "run_206",
+        "run_208",
+        "run_211",
+        "run_213",
+        "run_214",
+        "run_215",
+        "run_216",
+        "run_217",
+        "run_219",
+        "run_220",
+        "run_221",
+        "run_223",
+        "run_224",
+        "run_225",
+        "run_226",
+        "run_227",
+        "run_228",
+        "run_229",
+        "run_231",
+        "run_232",
+        "run_235",
+        "run_236",
+        "run_237",
+        "run_239",
+        "run_240",
+        "run_242",
+        "run_244",
+        "run_245",
+        "run_246",
+        "run_248",
+        "run_249",
+        "run_250",
+        "run_251",
+        "run_252",
+        "run_256",
+        "run_257",
+        "run_258",
+        "run_260",
+        "run_261",
+        "run_263",
+        "run_264",
+        "run_265",
+        "run_266",
+        "run_267",
+        "run_268",
+        "run_270",
+        "run_271",
+        "run_272",
+        "run_273",
+        "run_274",
+        "run_275",
+        "run_277",
+        "run_280",
+        "run_281",
+        "run_283",
+        "run_284",
+        "run_285",
+        "run_288",
+        "run_289",
+        "run_290",
+        "run_292",
+        "run_293",
+        "run_295",
+        "run_296",
+        "run_297",
+        "run_298",
+        "run_300",
+        "run_301",
+        "run_304",
+        "run_306",
+        "run_309",
+        "run_310",
+        "run_311",
+        "run_312",
+        "run_314",
+        "run_315",
+        "run_316",
+        "run_317",
+        "run_318",
+        "run_319",
+        "run_320",
+        "run_321",
+        "run_322",
+        "run_323",
+        "run_324",
+        "run_325",
+        "run_327",
+        "run_328",
+        "run_329",
+        "run_331",
+        "run_332",
+        "run_333",
+        "run_334",
+        "run_335",
+        "run_336",
+        "run_337",
+        "run_338",
+        "run_339",
+        "run_340",
+        "run_341",
+        "run_342",
+        "run_343",
+        "run_346",
+        "run_347",
+        "run_349",
+        "run_350",
+        "run_353",
+        "run_356",
+        "run_358",
+        "run_361",
+        "run_362",
+        "run_363",
+        "run_364",
+        "run_365",
+        "run_367",
+        "run_368",
+        "run_373",
+        "run_374",
+        "run_376",
+        "run_377",
+        "run_378",
+        "run_379",
+        "run_380",
+        "run_381",
+        "run_384",
+        "run_385",
+        "run_387",
+        "run_388",
+        "run_391",
+        "run_392",
+        "run_393",
+        "run_396",
+        "run_397",
+        "run_399",
+        "run_400",
+        "run_401",
+        "run_404",
+        "run_406",
+        "run_407",
+        "run_409",
+        "run_410",
+        "run_411",
+        "run_412",
+        "run_413",
+        "run_414",
+        "run_416",
+        "run_417",
+        "run_419",
+        "run_420",
+        "run_421",
+        "run_423",
+        "run_424",
+        "run_425",
+        "run_426",
+        "run_427",
+        "run_428",
+        "run_429",
+        "run_430",
+        "run_431",
+        "run_433",
+        "run_434",
+        "run_435",
+        "run_437",
+        "run_438",
+        "run_439",
+        "run_441",
+        "run_443",
+        "run_444",
+        "run_445",
+        "run_446",
+        "run_447",
+        "run_448",
+        "run_449",
+        "run_450",
+        "run_451",
+        "run_452",
+        "run_453",
+        "run_454",
+        "run_455",
+        "run_456",
+        "run_457",
+        "run_458",
+        "run_459",
+        "run_461",
+        "run_462",
+        "run_464",
+        "run_465",
+        "run_467",
+        "run_468",
+        "run_469",
+        "run_470",
+        "run_471",
+        "run_472",
+        "run_473",
+        "run_474",
+        "run_475",
+        "run_476",
+        "run_477",
+        "run_479",
+        "run_480",
+        "run_481",
+        "run_482",
+        "run_483",
+        "run_487",
+        "run_491",
+        "run_492",
+        "run_493",
+        "run_494",
+        "run_495",
+        "run_496",
+        "run_497",
+        "run_498",
+        "run_499"
+    ],
+    "low_drag_val": [
+        "run_4",
+        "run_7",
+        "run_8",
+        "run_15",
+        "run_37",
+        "run_39",
+        "run_55",
+        "run_56",
+        "run_61",
+        "run_73",
+        "run_101",
+        "run_106",
+        "run_135",
+        "run_159",
+        "run_165",
+        "run_168",
+        "run_169",
+        "run_175",
+        "run_197",
+        "run_207",
+        "run_209",
+        "run_212",
+        "run_222",
+        "run_230",
+        "run_243",
+        "run_254",
+        "run_255",
+        "run_278",
+        "run_287",
+        "run_294",
+        "run_305",
+        "run_344",
+        "run_351",
+        "run_366",
+        "run_370",
+        "run_371",
+        "run_382",
+        "run_383",
+        "run_389",
+        "run_394",
+        "run_402",
+        "run_403",
+        "run_415",
+        "run_432",
+        "run_460",
+        "run_484",
+        "run_485",
+        "run_489",
+        "run_490",
+        "run_500"
+    ],
+    "low_drag_test": [
+        "run_1",
+        "run_3",
+        "run_10",
+        "run_20",
+        "run_27",
+        "run_32",
+        "run_33",
+        "run_36",
+        "run_43",
+        "run_44",
+        "run_45",
+        "run_46",
+        "run_48",
+        "run_62",
+        "run_63",
+        "run_68",
+        "run_69",
+        "run_78",
+        "run_83",
+        "run_85",
+        "run_93",
+        "run_99",
+        "run_103",
+        "run_110",
+        "run_112",
+        "run_114",
+        "run_116",
+        "run_117",
+        "run_121",
+        "run_122",
+        "run_123",
+        "run_124",
+        "run_126",
+        "run_129",
+        "run_139",
+        "run_144",
+        "run_150",
+        "run_151",
+        "run_154",
+        "run_157",
+        "run_158",
+        "run_167",
+        "run_176",
+        "run_189",
+        "run_193",
+        "run_198",
+        "run_199",
+        "run_201",
+        "run_202",
+        "run_210",
+        "run_218",
+        "run_233",
+        "run_234",
+        "run_238",
+        "run_241",
+        "run_247",
+        "run_253",
+        "run_259",
+        "run_262",
+        "run_269",
+        "run_276",
+        "run_279",
+        "run_282",
+        "run_286",
+        "run_291",
+        "run_299",
+        "run_302",
+        "run_303",
+        "run_307",
+        "run_308",
+        "run_313",
+        "run_326",
+        "run_330",
+        "run_345",
+        "run_348",
+        "run_352",
+        "run_354",
+        "run_355",
+        "run_357",
+        "run_359",
+        "run_360",
+        "run_369",
+        "run_372",
+        "run_375",
+        "run_386",
+        "run_390",
+        "run_395",
+        "run_398",
+        "run_405",
+        "run_408",
+        "run_418",
+        "run_422",
+        "run_436",
+        "run_440",
+        "run_442",
+        "run_463",
+        "run_466",
+        "run_478",
+        "run_486",
+        "run_488"
+    ],
+    "geometry_train": [
+        "run_1",
+        "run_2",
+        "run_3",
+        "run_4",
+        "run_5",
+        "run_6",
+        "run_8",
+        "run_9",
+        "run_13",
+        "run_15",
+        "run_16",
+        "run_18",
+        "run_19",
+        "run_20",
+        "run_21",
+        "run_22",
+        "run_24",
+        "run_25",
+        "run_26",
+        "run_27",
+        "run_29",
+        "run_30",
+        "run_32",
+        "run_33",
+        "run_34",
+        "run_37",
+        "run_39",
+        "run_40",
+        "run_41",
+        "run_42",
+        "run_43",
+        "run_44",
+        "run_45",
+        "run_46",
+        "run_47",
+        "run_51",
+        "run_53",
+        "run_57",
+        "run_58",
+        "run_64",
+        "run_66",
+        "run_68",
+        "run_70",
+        "run_71",
+        "run_72",
+        "run_73",
+        "run_74",
+        "run_75",
+        "run_76",
+        "run_77",
+        "run_78",
+        "run_79",
+        "run_80",
+        "run_82",
+        "run_83",
+        "run_84",
+        "run_85",
+        "run_86",
+        "run_87",
+        "run_88",
+        "run_89",
+        "run_93",
+        "run_94",
+        "run_95",
+        "run_99",
+        "run_100",
+        "run_102",
+        "run_103",
+        "run_105",
+        "run_106",
+        "run_107",
+        "run_108",
+        "run_110",
+        "run_111",
+        "run_112",
+        "run_113",
+        "run_116",
+        "run_117",
+        "run_118",
+        "run_119",
+        "run_120",
+        "run_121",
+        "run_123",
+        "run_124",
+        "run_125",
+        "run_126",
+        "run_127",
+        "run_129",
+        "run_130",
+        "run_131",
+        "run_132",
+        "run_133",
+        "run_134",
+        "run_136",
+        "run_137",
+        "run_138",
+        "run_139",
+        "run_140",
+        "run_141",
+        "run_143",
+        "run_144",
+        "run_146",
+        "run_147",
+        "run_148",
+        "run_149",
+        "run_150",
+        "run_151",
+        "run_154",
+        "run_155",
+        "run_157",
+        "run_159",
+        "run_160",
+        "run_161",
+        "run_163",
+        "run_165",
+        "run_166",
+        "run_167",
+        "run_168",
+        "run_170",
+        "run_171",
+        "run_172",
+        "run_173",
+        "run_174",
+        "run_175",
+        "run_179",
+        "run_180",
+        "run_181",
+        "run_184",
+        "run_185",
+        "run_186",
+        "run_187",
+        "run_188",
+        "run_189",
+        "run_190",
+        "run_191",
+        "run_192",
+        "run_193",
+        "run_194",
+        "run_195",
+        "run_196",
+        "run_197",
+        "run_198",
+        "run_199",
+        "run_201",
+        "run_202",
+        "run_204",
+        "run_207",
+        "run_208",
+        "run_209",
+        "run_210",
+        "run_212",
+        "run_213",
+        "run_214",
+        "run_215",
+        "run_217",
+        "run_218",
+        "run_219",
+        "run_220",
+        "run_221",
+        "run_224",
+        "run_226",
+        "run_227",
+        "run_228",
+        "run_231",
+        "run_232",
+        "run_234",
+        "run_235",
+        "run_236",
+        "run_237",
+        "run_238",
+        "run_239",
+        "run_241",
+        "run_245",
+        "run_246",
+        "run_247",
+        "run_248",
+        "run_250",
+        "run_251",
+        "run_252",
+        "run_255",
+        "run_256",
+        "run_257",
+        "run_258",
+        "run_259",
+        "run_260",
+        "run_261",
+        "run_262",
+        "run_264",
+        "run_265",
+        "run_267",
+        "run_268",
+        "run_269",
+        "run_270",
+        "run_271",
+        "run_272",
+        "run_274",
+        "run_275",
+        "run_276",
+        "run_277",
+        "run_278",
+        "run_279",
+        "run_280",
+        "run_281",
+        "run_282",
+        "run_283",
+        "run_284",
+        "run_286",
+        "run_287",
+        "run_289",
+        "run_290",
+        "run_292",
+        "run_293",
+        "run_294",
+        "run_298",
+        "run_299",
+        "run_300",
+        "run_302",
+        "run_303",
+        "run_305",
+        "run_306",
+        "run_308",
+        "run_310",
+        "run_311",
+        "run_312",
+        "run_316",
+        "run_317",
+        "run_320",
+        "run_324",
+        "run_325",
+        "run_326",
+        "run_327",
+        "run_328",
+        "run_329",
+        "run_330",
+        "run_331",
+        "run_332",
+        "run_336",
+        "run_337",
+        "run_338",
+        "run_340",
+        "run_341",
+        "run_342",
+        "run_344",
+        "run_345",
+        "run_346",
+        "run_347",
+        "run_348",
+        "run_350",
+        "run_351",
+        "run_352",
+        "run_353",
+        "run_354",
+        "run_356",
+        "run_358",
+        "run_359",
+        "run_360",
+        "run_363",
+        "run_365",
+        "run_366",
+        "run_368",
+        "run_369",
+        "run_370",
+        "run_371",
+        "run_372",
+        "run_373",
+        "run_374",
+        "run_377",
+        "run_378",
+        "run_379",
+        "run_380",
+        "run_383",
+        "run_384",
+        "run_385",
+        "run_386",
+        "run_388",
+        "run_389",
+        "run_391",
+        "run_392",
+        "run_393",
+        "run_395",
+        "run_396",
+        "run_397",
+        "run_398",
+        "run_399",
+        "run_400",
+        "run_401",
+        "run_402",
+        "run_403",
+        "run_405",
+        "run_406",
+        "run_407",
+        "run_408",
+        "run_412",
+        "run_414",
+        "run_417",
+        "run_418",
+        "run_419",
+        "run_420",
+        "run_422",
+        "run_423",
+        "run_424",
+        "run_427",
+        "run_429",
+        "run_430",
+        "run_432",
+        "run_434",
+        "run_435",
+        "run_436",
+        "run_437",
+        "run_438",
+        "run_439",
+        "run_440",
+        "run_441",
+        "run_442",
+        "run_443",
+        "run_445",
+        "run_448",
+        "run_449",
+        "run_451",
+        "run_452",
+        "run_457",
+        "run_459",
+        "run_463",
+        "run_464",
+        "run_465",
+        "run_466",
+        "run_467",
+        "run_469",
+        "run_470",
+        "run_471",
+        "run_472",
+        "run_473",
+        "run_476",
+        "run_478",
+        "run_479",
+        "run_481",
+        "run_483",
+        "run_484",
+        "run_486",
+        "run_487",
+        "run_488",
+        "run_490",
+        "run_492",
+        "run_493",
+        "run_494",
+        "run_495",
+        "run_496",
+        "run_498",
+        "run_499",
+        "run_500"
+    ],
+    "geometry_val": [
+        "run_7",
+        "run_10",
+        "run_23",
+        "run_28",
+        "run_31",
+        "run_35",
+        "run_36",
+        "run_48",
+        "run_49",
+        "run_67",
+        "run_69",
+        "run_81",
+        "run_90",
+        "run_97",
+        "run_101",
+        "run_114",
+        "run_122",
+        "run_128",
+        "run_142",
+        "run_156",
+        "run_158",
+        "run_169",
+        "run_177",
+        "run_178",
+        "run_203",
+        "run_205",
+        "run_222",
+        "run_230",
+        "run_233",
+        "run_254",
+        "run_263",
+        "run_285",
+        "run_304",
+        "run_307",
+        "run_313",
+        "run_321",
+        "run_322",
+        "run_367",
+        "run_376",
+        "run_381",
+        "run_387",
+        "run_390",
+        "run_416",
+        "run_426",
+        "run_447",
+        "run_453",
+        "run_460",
+        "run_461",
+        "run_475",
+        "run_485"
+    ],
+    "geometry_test": [
+        "run_11",
+        "run_12",
+        "run_14",
+        "run_17",
+        "run_38",
+        "run_50",
+        "run_52",
+        "run_54",
+        "run_55",
+        "run_56",
+        "run_59",
+        "run_60",
+        "run_61",
+        "run_62",
+        "run_63",
+        "run_65",
+        "run_91",
+        "run_92",
+        "run_96",
+        "run_98",
+        "run_104",
+        "run_109",
+        "run_115",
+        "run_135",
+        "run_145",
+        "run_152",
+        "run_153",
+        "run_162",
+        "run_164",
+        "run_176",
+        "run_182",
+        "run_183",
+        "run_200",
+        "run_206",
+        "run_211",
+        "run_216",
+        "run_223",
+        "run_225",
+        "run_229",
+        "run_240",
+        "run_242",
+        "run_243",
+        "run_244",
+        "run_249",
+        "run_253",
+        "run_266",
+        "run_273",
+        "run_288",
+        "run_291",
+        "run_295",
+        "run_296",
+        "run_297",
+        "run_301",
+        "run_309",
+        "run_314",
+        "run_315",
+        "run_318",
+        "run_319",
+        "run_323",
+        "run_333",
+        "run_334",
+        "run_335",
+        "run_339",
+        "run_343",
+        "run_349",
+        "run_355",
+        "run_357",
+        "run_361",
+        "run_362",
+        "run_364",
+        "run_375",
+        "run_382",
+        "run_394",
+        "run_404",
+        "run_409",
+        "run_410",
+        "run_411",
+        "run_413",
+        "run_415",
+        "run_421",
+        "run_425",
+        "run_428",
+        "run_431",
+        "run_433",
+        "run_444",
+        "run_446",
+        "run_450",
+        "run_454",
+        "run_455",
+        "run_456",
+        "run_458",
+        "run_462",
+        "run_468",
+        "run_474",
+        "run_477",
+        "run_480",
+        "run_482",
+        "run_489",
+        "run_491",
+        "run_497"
+    ],
+    "image_wake_train": [
+        "run_2",
+        "run_3",
+        "run_4",
+        "run_7",
+        "run_9",
+        "run_10",
+        "run_11",
+        "run_15",
+        "run_16",
+        "run_17",
+        "run_19",
+        "run_20",
+        "run_22",
+        "run_24",
+        "run_27",
+        "run_28",
+        "run_29",
+        "run_30",
+        "run_32",
+        "run_33",
+        "run_35",
+        "run_36",
+        "run_37",
+        "run_38",
+        "run_39",
+        "run_40",
+        "run_42",
+        "run_43",
+        "run_44",
+        "run_45",
+        "run_46",
+        "run_47",
+        "run_49",
+        "run_50",
+        "run_51",
+        "run_52",
+        "run_53",
+        "run_54",
+        "run_56",
+        "run_57",
+        "run_58",
+        "run_59",
+        "run_61",
+        "run_62",
+        "run_66",
+        "run_67",
+        "run_68",
+        "run_70",
+        "run_72",
+        "run_77",
+        "run_78",
+        "run_79",
+        "run_80",
+        "run_83",
+        "run_84",
+        "run_86",
+        "run_87",
+        "run_88",
+        "run_91",
+        "run_92",
+        "run_93",
+        "run_95",
+        "run_96",
+        "run_97",
+        "run_98",
+        "run_99",
+        "run_101",
+        "run_102",
+        "run_103",
+        "run_104",
+        "run_105",
+        "run_106",
+        "run_109",
+        "run_110",
+        "run_111",
+        "run_112",
+        "run_113",
+        "run_114",
+        "run_116",
+        "run_117",
+        "run_118",
+        "run_119",
+        "run_120",
+        "run_121",
+        "run_122",
+        "run_123",
+        "run_124",
+        "run_125",
+        "run_126",
+        "run_127",
+        "run_128",
+        "run_129",
+        "run_131",
+        "run_133",
+        "run_134",
+        "run_135",
+        "run_136",
+        "run_137",
+        "run_139",
+        "run_140",
+        "run_142",
+        "run_143",
+        "run_144",
+        "run_145",
+        "run_146",
+        "run_147",
+        "run_149",
+        "run_150",
+        "run_151",
+        "run_153",
+        "run_154",
+        "run_155",
+        "run_156",
+        "run_157",
+        "run_159",
+        "run_160",
+        "run_161",
+        "run_163",
+        "run_165",
+        "run_166",
+        "run_167",
+        "run_168",
+        "run_169",
+        "run_170",
+        "run_172",
+        "run_175",
+        "run_176",
+        "run_177",
+        "run_180",
+        "run_181",
+        "run_182",
+        "run_184",
+        "run_185",
+        "run_186",
+        "run_188",
+        "run_189",
+        "run_192",
+        "run_193",
+        "run_194",
+        "run_199",
+        "run_200",
+        "run_201",
+        "run_202",
+        "run_203",
+        "run_206",
+        "run_208",
+        "run_210",
+        "run_211",
+        "run_212",
+        "run_213",
+        "run_214",
+        "run_215",
+        "run_219",
+        "run_221",
+        "run_222",
+        "run_223",
+        "run_224",
+        "run_225",
+        "run_226",
+        "run_229",
+        "run_230",
+        "run_231",
+        "run_233",
+        "run_234",
+        "run_235",
+        "run_237",
+        "run_238",
+        "run_239",
+        "run_240",
+        "run_241",
+        "run_242",
+        "run_244",
+        "run_245",
+        "run_246",
+        "run_247",
+        "run_248",
+        "run_251",
+        "run_252",
+        "run_253",
+        "run_257",
+        "run_258",
+        "run_259",
+        "run_260",
+        "run_261",
+        "run_262",
+        "run_263",
+        "run_265",
+        "run_266",
+        "run_267",
+        "run_268",
+        "run_270",
+        "run_271",
+        "run_272",
+        "run_273",
+        "run_275",
+        "run_276",
+        "run_277",
+        "run_278",
+        "run_279",
+        "run_282",
+        "run_285",
+        "run_286",
+        "run_287",
+        "run_288",
+        "run_290",
+        "run_291",
+        "run_292",
+        "run_293",
+        "run_294",
+        "run_296",
+        "run_297",
+        "run_298",
+        "run_299",
+        "run_300",
+        "run_301",
+        "run_302",
+        "run_303",
+        "run_304",
+        "run_307",
+        "run_308",
+        "run_312",
+        "run_313",
+        "run_314",
+        "run_317",
+        "run_318",
+        "run_319",
+        "run_320",
+        "run_322",
+        "run_323",
+        "run_324",
+        "run_325",
+        "run_326",
+        "run_328",
+        "run_329",
+        "run_330",
+        "run_332",
+        "run_333",
+        "run_334",
+        "run_335",
+        "run_337",
+        "run_338",
+        "run_340",
+        "run_341",
+        "run_342",
+        "run_343",
+        "run_344",
+        "run_346",
+        "run_348",
+        "run_349",
+        "run_350",
+        "run_352",
+        "run_355",
+        "run_357",
+        "run_358",
+        "run_359",
+        "run_360",
+        "run_361",
+        "run_363",
+        "run_364",
+        "run_366",
+        "run_368",
+        "run_369",
+        "run_370",
+        "run_371",
+        "run_372",
+        "run_373",
+        "run_374",
+        "run_375",
+        "run_376",
+        "run_379",
+        "run_380",
+        "run_381",
+        "run_384",
+        "run_386",
+        "run_387",
+        "run_389",
+        "run_390",
+        "run_391",
+        "run_392",
+        "run_393",
+        "run_394",
+        "run_395",
+        "run_396",
+        "run_398",
+        "run_401",
+        "run_402",
+        "run_403",
+        "run_404",
+        "run_405",
+        "run_406",
+        "run_408",
+        "run_409",
+        "run_411",
+        "run_412",
+        "run_413",
+        "run_415",
+        "run_416",
+        "run_417",
+        "run_418",
+        "run_419",
+        "run_422",
+        "run_423",
+        "run_425",
+        "run_426",
+        "run_427",
+        "run_429",
+        "run_431",
+        "run_432",
+        "run_433",
+        "run_434",
+        "run_436",
+        "run_439",
+        "run_440",
+        "run_441",
+        "run_442",
+        "run_444",
+        "run_445",
+        "run_446",
+        "run_450",
+        "run_451",
+        "run_454",
+        "run_456",
+        "run_458",
+        "run_461",
+        "run_463",
+        "run_464",
+        "run_466",
+        "run_467",
+        "run_469",
+        "run_471",
+        "run_472",
+        "run_473",
+        "run_474",
+        "run_475",
+        "run_476",
+        "run_477",
+        "run_478",
+        "run_479",
+        "run_480",
+        "run_484",
+        "run_485",
+        "run_486",
+        "run_487",
+        "run_488",
+        "run_490",
+        "run_491",
+        "run_493",
+        "run_496",
+        "run_498",
+        "run_499"
+    ],
+    "image_wake_val": [
+        "run_1",
+        "run_5",
+        "run_8",
+        "run_25",
+        "run_31",
+        "run_48",
+        "run_63",
+        "run_64",
+        "run_65",
+        "run_85",
+        "run_90",
+        "run_100",
+        "run_108",
+        "run_130",
+        "run_132",
+        "run_138",
+        "run_158",
+        "run_162",
+        "run_164",
+        "run_179",
+        "run_183",
+        "run_195",
+        "run_198",
+        "run_207",
+        "run_209",
+        "run_218",
+        "run_227",
+        "run_243",
+        "run_249",
+        "run_254",
+        "run_255",
+        "run_256",
+        "run_269",
+        "run_281",
+        "run_310",
+        "run_345",
+        "run_347",
+        "run_377",
+        "run_385",
+        "run_397",
+        "run_410",
+        "run_430",
+        "run_437",
+        "run_438",
+        "run_449",
+        "run_452",
+        "run_468",
+        "run_481",
+        "run_482",
+        "run_489"
+    ],
+    "image_wake_test": [
+        "run_6",
+        "run_12",
+        "run_13",
+        "run_14",
+        "run_18",
+        "run_21",
+        "run_23",
+        "run_26",
+        "run_34",
+        "run_41",
+        "run_55",
+        "run_60",
+        "run_69",
+        "run_71",
+        "run_73",
+        "run_74",
+        "run_75",
+        "run_76",
+        "run_81",
+        "run_82",
+        "run_89",
+        "run_94",
+        "run_107",
+        "run_115",
+        "run_141",
+        "run_148",
+        "run_152",
+        "run_171",
+        "run_173",
+        "run_174",
+        "run_178",
+        "run_187",
+        "run_190",
+        "run_191",
+        "run_196",
+        "run_197",
+        "run_204",
+        "run_205",
+        "run_216",
+        "run_217",
+        "run_220",
+        "run_228",
+        "run_232",
+        "run_236",
+        "run_250",
+        "run_264",
+        "run_274",
+        "run_280",
+        "run_283",
+        "run_284",
+        "run_289",
+        "run_295",
+        "run_305",
+        "run_306",
+        "run_309",
+        "run_311",
+        "run_315",
+        "run_316",
+        "run_321",
+        "run_327",
+        "run_331",
+        "run_336",
+        "run_339",
+        "run_351",
+        "run_353",
+        "run_354",
+        "run_356",
+        "run_362",
+        "run_365",
+        "run_367",
+        "run_378",
+        "run_382",
+        "run_383",
+        "run_388",
+        "run_399",
+        "run_400",
+        "run_407",
+        "run_414",
+        "run_420",
+        "run_421",
+        "run_424",
+        "run_428",
+        "run_435",
+        "run_443",
+        "run_447",
+        "run_448",
+        "run_453",
+        "run_455",
+        "run_457",
+        "run_459",
+        "run_460",
+        "run_462",
+        "run_465",
+        "run_470",
+        "run_483",
+        "run_492",
+        "run_494",
+        "run_495",
+        "run_497",
+        "run_500"
+    ]
+}

splits/parameter_geometry_metrics.csv

@@ -0,0 +1,501 @@
+run,geometry_observed,ood_score,mean_10_nn_parameter_distance,mean_all_parameter_distance
+1,true,1.5371733399829923,1.5371733399829923,4.199731382417224
+2,true,1.4457567606929893,1.4457567606929893,3.643387473648818
+3,true,1.3189681131530917,1.3189681131530917,3.35612935034657
+4,true,1.4637477105727439,1.4637477105727439,3.605043103191509
+5,true,1.4537520206700978,1.4537520206700978,3.974757639523387
+6,true,1.365176865740792,1.365176865740792,3.8453861852612605
+7,true,1.2694473876511527,1.2694473876511527,3.2423345370647625
+8,true,1.5092872398730628,1.5092872398730628,4.034250219149982
+9,true,1.4984929417418336,1.4984929417418336,3.7149773093263545
+10,true,1.4112894153192828,1.4112894153192828,4.013813504220453
+11,true,1.8251202002021731,1.8251202002021731,4.532237727003187
+12,true,1.7712793033405727,1.7712793033405727,3.930804887593982
+13,true,1.3728479009245877,1.3728479009245877,3.3227738862855456
+14,true,1.968277047586993,1.968277047586993,4.277672628834627
+15,true,1.51700391755795,1.51700391755795,3.860611458610603
+16,true,1.491179657734723,1.491179657734723,3.5705746292863356
+17,true,1.8151770518836585,1.8151770518836585,4.454243690420919
+18,true,1.2425136674783155,1.2425136674783155,3.7099203668773617
+19,true,1.9366994787623184,1.9366994787623184,4.300523616915295
+20,true,1.6541944498218613,1.6541944498218613,3.808239982221566
+21,true,1.4989923349794088,1.4989923349794088,3.617694239894782
+22,true,1.5885224877940947,1.5885224877940947,4.683458684595786
+23,true,1.4953705034562617,1.4953705034562617,3.518936957690446
+24,true,1.487681441561694,1.487681441561694,3.4777791385117873
+25,true,1.8454082571389812,1.8454082571389812,4.254133885400819
+26,true,1.5365936800782911,1.5365936800782911,3.3199332227966383
+27,true,1.3970045311559727,1.3970045311559727,3.7933758739257315
+28,true,1.4745004264323,1.4745004264323,3.5335530641974446
+29,true,1.3761190025066437,1.3761190025066437,3.3004267985172113
+30,true,1.6893560216752586,1.6893560216752586,3.9962169886671584
+31,true,1.545621702962044,1.545621702962044,3.5267752580256215
+32,true,1.6889309056680966,1.6889309056680966,3.760402654267447
+33,true,1.4310336992451789,1.4310336992451789,3.587550074637966
+34,true,1.7397834960200353,1.7397834960200353,3.8012203702726968
+35,true,1.33856233573376,1.33856233573376,3.2424276247642267
+36,true,1.6189911615062065,1.6189911615062065,3.978705649462621
+37,true,1.4139180883261935,1.4139180883261935,3.2499973775608963
+38,true,1.6770964866407307,1.6770964866407307,3.821484950355596
+39,true,1.3548187132170022,1.3548187132170022,3.406777441274412
+40,true,1.6557060501522805,1.6557060501522805,3.407621123089733
+41,true,1.4376302338623428,1.4376302338623428,3.412314420814256
+42,true,1.6989847984410862,1.6989847984410862,4.193548468789302
+43,true,1.4477911205941505,1.4477911205941505,3.9458349095031995
+44,true,1.4744137561868589,1.4744137561868589,4.009791043161175
+45,true,1.754005359467222,1.754005359467222,3.706431531481122
+46,true,1.823165410165522,1.823165410165522,4.310895435146352
+47,true,1.6916700526409845,1.6916700526409845,3.704412318413234
+48,true,1.4391015095848254,1.4391015095848254,3.6568167122340443
+49,true,1.3956524392748952,1.3956524392748952,3.4060303385280606
+50,true,1.8907191285272638,1.8907191285272638,4.720967852857075
+51,true,1.7893969917047123,1.7893969917047123,3.7488823164207283
+52,true,2.309282868545322,2.309282868545322,4.768504835949298
+53,true,1.6497577994800028,1.6497577994800028,4.207021978009715
+54,true,1.9264315428222236,1.9264315428222236,4.375363995160352
+55,true,1.4602410845533815,1.4602410845533815,4.296239310035959
+56,true,2.00997154823524,2.00997154823524,4.3028316342666155
+57,true,1.682830682672233,1.682830682672233,3.788080553087215
+58,true,1.620887287124845,1.620887287124845,4.049923317465914
+59,true,1.624696725696135,1.624696725696135,4.479079928660714
+60,true,1.8290648944465844,1.8290648944465844,3.938631142580382
+61,true,1.5688593528957715,1.5688593528957715,3.933310793887141
+62,true,1.9165782635278668,1.9165782635278668,4.123295657796598
+63,true,1.6877121111162885,1.6877121111162885,3.9226317047368986
+64,true,1.8509594776263154,1.8509594776263154,4.377195316735396
+65,true,1.5709692377174886,1.5709692377174886,4.025412137718886
+66,true,1.5718055414874172,1.5718055414874172,3.9417203385067787
+67,true,1.3692193227006528,1.3692193227006528,3.099839273305964
+68,true,1.5374225220524156,1.5374225220524156,3.90330701058173
+69,true,1.3286517771489355,1.3286517771489355,3.562948488606504
+70,true,1.4225706735091213,1.4225706735091213,3.761488447899334
+71,true,1.083451807365487,1.083451807365487,3.406889003487848
+72,true,1.3799166377376035,1.3799166377376035,3.8555909942172644
+73,true,1.3741014027201444,1.3741014027201444,3.173782952181643
+74,true,1.5053233531809995,1.5053233531809995,4.1250857969598655
+75,true,1.7506473635808342,1.7506473635808342,4.116616578952039
+76,true,1.458204620468242,1.458204620468242,3.2634205245687276
+77,true,1.8185581729849627,1.8185581729849627,3.8328932727908347
+78,true,1.5531491381911433,1.5531491381911433,3.6420691452681724
+79,true,1.4641335851408752,1.4641335851408752,3.4385044109216545
+80,true,1.5853516582137321,1.5853516582137321,3.4828392403975132
+81,true,1.5016173850537036,1.5016173850537036,4.126321204830208
+82,true,1.231829270569848,1.231829270569848,3.8331919954302602
+83,true,1.7429301423139887,1.7429301423139887,4.3377907657112695
+84,true,1.4569497478590012,1.4569497478590012,3.694709947440542
+85,true,1.907333589286227,1.907333589286227,4.385421965606942
+86,true,1.5990679910368417,1.5990679910368417,4.0436722271511165
+87,true,1.3852075134203237,1.3852075134203237,4.010298556134474
+88,true,1.301818625756542,1.301818625756542,3.379433395623079
+89,true,1.2342852884850912,1.2342852884850912,3.6087925765451954
+90,true,1.5162832449183785,1.5162832449183785,3.9262912924427162
+91,true,1.6079192889485596,1.6079192889485596,4.022377516778384
+92,true,1.6336931969081878,1.6336931969081878,4.285207943492267
+93,true,2.2213372481698133,2.2213372481698133,5.071586810401775
+94,true,1.4467328890796982,1.4467328890796982,3.690955465678884
+95,true,1.6308090897689795,1.6308090897689795,4.164440447719747
+96,true,2.0062888892131014,2.0062888892131014,4.2199078502423575
+97,true,1.4734040300026445,1.4734040300026445,3.6031866346525523
+98,true,1.7609987934622708,1.7609987934622708,3.5560480013947338
+99,true,1.2920265450446036,1.2920265450446036,3.6396381560843185
+100,true,1.5577806476755307,1.5577806476755307,3.9738473715604945
+101,true,1.5666692997826677,1.5666692997826677,3.827994554172128
+102,true,1.503418391611525,1.503418391611525,3.9232942943165448
+103,true,1.866972530233232,1.866972530233232,4.516924538014431
+104,true,1.329272387419678,1.329272387419678,3.7686685166402043
+105,true,1.7659789516467626,1.7659789516467626,4.603390267694349
+106,true,1.513333885908281,1.513333885908281,3.8701484435678037
+107,true,1.4019711644283679,1.4019711644283679,3.1959798186124897
+108,true,1.384244101952318,1.384244101952318,3.5136350071822817
+109,true,1.598152773354302,1.598152773354302,4.297062213695695
+110,true,1.5537827299165634,1.5537827299165634,3.795912198655318
+111,true,1.4990687886531358,1.4990687886531358,3.662523668905521
+112,true,1.7059166438732767,1.7059166438732767,4.489260535982661
+113,true,1.416561609923958,1.416561609923958,3.697305251714732
+114,true,1.3424313110870458,1.3424313110870458,3.8253418961071306
+115,true,1.9123890123640013,1.9123890123640013,4.035519602925534
+116,true,1.773010513727894,1.773010513727894,4.518845115013138
+117,true,1.8510929041365287,1.8510929041365287,4.258679591411199
+118,true,1.5086183142581073,1.5086183142581073,4.050039795106349
+119,true,1.4168010613617503,1.4168010613617503,3.570965617638404
+120,true,1.8150002778042265,1.8150002778042265,4.294952446745636
+121,true,1.6981522445303685,1.6981522445303685,4.106364615592783
+122,true,1.8868375002086952,1.8868375002086952,4.38873749475354
+123,true,1.6569146885490185,1.6569146885490185,4.107576147721832
+124,true,1.6974542933906733,1.6974542933906733,4.066510216556818
+125,true,1.5491074453582854,1.5491074453582854,4.011479978981576
+126,true,1.5384045239015844,1.5384045239015844,3.82230289186579
+127,true,1.4903642596825828,1.4903642596825828,3.32856181624478
+128,true,1.6183935445201283,1.6183935445201283,3.9026798883704372
+129,true,1.430622492729711,1.430622492729711,3.8516607805356546
+130,true,1.643520267904035,1.643520267904035,4.264588000757018
+131,true,1.8186863015363244,1.8186863015363244,4.388337954474262
+132,true,1.3417771058965968,1.3417771058965968,4.084096813373756
+133,true,1.5855573278009807,1.5855573278009807,3.9888205324798878
+134,true,1.3160390736685117,1.3160390736685117,3.3206357088486844
+135,true,1.9237306666996237,1.9237306666996237,4.781593657086239
+136,true,1.3387125083778972,1.3387125083778972,3.524712033994034
+137,true,1.4206716301610673,1.4206716301610673,3.1473648200263407
+138,true,1.5661818161732375,1.5661818161732375,3.9958459703812617
+139,true,1.8776698909096872,1.8776698909096872,4.513687443367072
+140,true,1.6772516821164896,1.6772516821164896,3.681922263980017
+141,true,1.4230269799494033,1.4230269799494033,3.550802190367728
+142,true,1.4621891124779869,1.4621891124779869,3.2878100520751
+143,true,1.434927239029477,1.434927239029477,3.1303910146268374
+144,true,1.7492852055084573,1.7492852055084573,4.266224692621611
+145,true,1.8142617004317614,1.8142617004317614,4.05907918866984
+146,true,1.4361906507815738,1.4361906507815738,3.149397725635229
+147,true,1.3405042799540927,1.3405042799540927,3.617573082841226
+148,true,1.4960221740966344,1.4960221740966344,3.3071721483419756
+149,true,1.587769545464882,1.587769545464882,3.762286353941433
+150,true,1.5328195510695175,1.5328195510695175,3.7762772967098743
+151,true,1.525584596798803,1.525584596798803,3.506494477078765
+152,true,2.0290502850557472,2.0290502850557472,4.59474056522266
+153,true,1.674808809207961,1.674808809207961,4.423064271390135
+154,true,1.660721583580795,1.660721583580795,4.093659768445963
+155,true,1.7033023050179217,1.7033023050179217,4.576473858905595
+156,true,1.8431260481805576,1.8431260481805576,4.059210340326662
+157,true,1.6617799057996314,1.6617799057996314,4.200185091833271
+158,true,1.751472749570747,1.751472749570747,4.351711379334925
+159,true,1.630941123885798,1.630941123885798,4.038355072561369
+160,true,1.477996981966692,1.477996981966692,3.5577895723337556
+161,true,1.6091362248455805,1.6091362248455805,3.777465268627914
+162,true,1.9023111149479917,1.9023111149479917,4.449125378971031
+163,true,1.5692527278034554,1.5692527278034554,3.943460799498046
+164,true,2.0165650348331723,2.0165650348331723,4.827079504989251
+165,true,1.4756260409626782,1.4756260409626782,3.4790884789314838
+166,true,1.765067819369215,1.765067819369215,4.3575809754969566
+167,true,1.968666423129207,1.968666423129207,4.806530587116946
+168,true,1.441254724538593,1.441254724538593,3.3571999323196833
+169,true,1.5294551218558556,1.5294551218558556,3.8964200472634847
+170,true,1.632686892807427,1.632686892807427,4.1451942488544695
+171,true,1.2764347510860659,1.2764347510860659,3.9586919360452653
+172,true,1.5118139054815074,1.5118139054815074,3.650379136356431
+173,true,1.3992759846553315,1.3992759846553315,3.7369280772726827
+174,true,1.3266707170166816,1.3266707170166816,3.7848426023705275
+175,true,1.488498974885008,1.488498974885008,3.6993479123093698
+176,true,1.7686188210820286,1.7686188210820286,4.152237067358615
+177,true,1.4336569886253003,1.4336569886253003,3.3859366885876008
+178,true,1.4774675288050207,1.4774675288050207,3.7722946674496045
+179,true,1.2598843489656582,1.2598843489656582,3.3668756694499806
+180,true,1.9511579018962535,1.9511579018962535,4.119390079632346
+181,true,1.6142815834751147,1.6142815834751147,4.162612081542571
+182,true,1.6339558613447207,1.6339558613447207,3.771458579031521
+183,true,1.691462833025458,1.691462833025458,4.272925601063628
+184,true,1.6730207810808544,1.6730207810808544,3.880910999149458
+185,true,1.580686358708466,1.580686358708466,3.94689074807866
+186,true,1.703984694626201,1.703984694626201,4.500703769831264
+187,true,1.39017741488616,1.39017741488616,3.275242323805824
+188,true,1.5255931184870264,1.5255931184870264,3.9408541316271517
+189,true,1.2731759202107171,1.2731759202107171,3.4274090911942574
+190,true,1.4450729527376085,1.4450729527376085,4.257778215105246
+191,true,1.3663790649339724,1.3663790649339724,3.2643450502058022
+192,true,1.5977635909028203,1.5977635909028203,3.846389265143639
+193,true,1.6187821445722879,1.6187821445722879,4.218166013830504
+194,true,1.2768093904995577,1.2768093904995577,3.9482100817754326
+195,true,1.3302452008629913,1.3302452008629913,3.5301269605395857
+196,true,1.3768259080530505,1.3768259080530505,3.34753222820955
+197,true,1.6308037461313194,1.6308037461313194,3.597480747663258
+198,true,1.4473173876681946,1.4473173876681946,4.160570982383998
+199,true,1.698859945411184,1.698859945411184,4.046620143608707
+200,true,2.092319445083557,2.092319445083557,4.706548868021713
+201,true,1.2878642769225557,1.2878642769225557,3.6135827076033644
+202,true,1.5046112136192584,1.5046112136192584,3.4910459705936447
+203,true,1.5784028711775087,1.5784028711775087,4.142574065458008
+204,true,1.449223993374568,1.449223993374568,3.7066178941635473
+205,true,1.4023548643502521,1.4023548643502521,4.1950033991508064
+206,true,1.7659293532361748,1.7659293532361748,4.369807963744033
+207,true,1.5795932756129072,1.5795932756129072,3.5126272416649815
+208,true,1.4920479895665453,1.4920479895665453,3.6686007860465573
+209,true,1.430814315188954,1.430814315188954,3.8971423613988754
+210,true,1.4032800786398665,1.4032800786398665,3.9521825042895444
+211,true,1.7241113150382485,1.7241113150382485,4.570427170986842
+212,true,1.331555746545235,1.331555746545235,3.1786058575244023
+213,true,1.4706606955700192,1.4706606955700192,3.44419218568043
+214,true,1.5310881640190934,1.5310881640190934,3.8132245024288878
+215,true,1.7213387359046322,1.7213387359046322,4.578378367913626
+216,true,1.531687980500113,1.531687980500113,3.375077593523281
+217,true,1.4646845206652388,1.4646845206652388,3.3473789469031967
+218,true,1.502754253048365,1.502754253048365,3.578138938479329
+219,true,1.2666702567098516,1.2666702567098516,3.1401111922234892
+220,true,1.6346609392485156,1.6346609392485156,3.651500420376014
+221,true,1.6917388993139255,1.6917388993139255,4.187951283343251
+222,true,1.762639043619157,1.762639043619157,4.035365349386163
+223,true,1.630651795962183,1.630651795962183,4.077008159680319
+224,true,1.4186918970467592,1.4186918970467592,3.3371086661654816
+225,true,1.9860321720179868,1.9860321720179868,4.402211078673647
+226,true,1.4193351671580323,1.4193351671580323,3.595522326343532
+227,true,1.4788378987303592,1.4788378987303592,3.8478466865948375
+228,true,1.4789517578022864,1.4789517578022864,3.6494468663007282
+229,true,1.3196590656353597,1.3196590656353597,4.182756735713919
+230,true,1.2594321337975736,1.2594321337975736,3.7683955563516034
+231,true,1.4418883948589767,1.4418883948589767,3.612297640443388
+232,true,1.7810911379943657,1.7810911379943657,3.891206391345554
+233,true,1.6388770901005458,1.6388770901005458,3.9721706158291514
+234,true,1.3341526863053559,1.3341526863053559,3.601126813281968
+235,true,1.5551106867702988,1.5551106867702988,3.6405799038860764
+236,true,1.4773490485819527,1.4773490485819527,3.9761706448135774
+237,true,1.5127150599732122,1.5127150599732122,3.683908548054631
+238,true,1.5193867931045688,1.5193867931045688,3.9843098555583865
+239,true,1.6474960940940815,1.6474960940940815,4.081018636595739
+240,true,1.4787627784876005,1.4787627784876005,3.421542389176913
+241,true,1.4466690787794565,1.4466690787794565,3.7867830158211837
+242,true,1.4742719870183278,1.4742719870183278,3.9760727200506776
+243,true,1.6184559704501111,1.6184559704501111,4.045217072412992
+244,true,2.012624630146699,2.012624630146699,4.7698662423591225
+245,true,1.3666806897391455,1.3666806897391455,4.087180572283866
+246,true,1.2231877372146116,1.2231877372146116,4.02681623795392
+247,true,1.4281133372401829,1.4281133372401829,4.096414598565276
+248,true,1.5118507266175427,1.5118507266175427,3.7869217597047586
+249,true,1.8646922827742967,1.8646922827742967,4.149717790233465
+250,true,1.4568880969468727,1.4568880969468727,3.5275715972428068
+251,true,1.7885070449424112,1.7885070449424112,3.6202985090741615
+252,true,1.5761604217960012,1.5761604217960012,3.979728813912061
+253,true,1.8884535604219441,1.8884535604219441,4.756034168891553
+254,true,1.3854725431786492,1.3854725431786492,3.541793180953965
+255,true,1.5634092293390327,1.5634092293390327,4.058433206016442
+256,true,1.712897059583618,1.712897059583618,3.7837848609514264
+257,true,1.3453022814819469,1.3453022814819469,3.4030472637086455
+258,true,1.7506886682973222,1.7506886682973222,3.795711837933805
+259,true,1.573909900272678,1.573909900272678,3.9499565254852556
+260,true,1.4255626157986476,1.4255626157986476,3.7606668771138145
+261,true,1.5640531668544082,1.5640531668544082,4.477926636146256
+262,true,1.2929072929701366,1.2929072929701366,3.4005631775722076
+263,true,1.5852808264354339,1.5852808264354339,3.2123028035211783
+264,true,1.415264100409358,1.415264100409358,3.1599084329613687
+265,true,1.4912956043498158,1.4912956043498158,3.554678125546315
+266,true,1.9635891442248485,1.9635891442248485,4.134938876452722
+267,true,1.543160821241698,1.543160821241698,3.616506496643573
+268,true,1.7992999718217568,1.7992999718217568,4.133794004328868
+269,true,1.9783090829105816,1.9783090829105816,4.573608357187377
+270,true,1.2324962917335254,1.2324962917335254,2.963512871133029
+271,true,2.0658388549421938,2.0658388549421938,4.457116840444852
+272,true,1.3702147171394319,1.3702147171394319,3.2439214359484354
+273,true,1.929893680254637,1.929893680254637,4.535385752519869
+274,true,1.6640089097356998,1.6640089097356998,3.7073982460350683
+275,true,1.3756489842048603,1.3756489842048603,3.841503261321753
+276,true,1.5147437740285603,1.5147437740285603,3.840453323829047
+277,true,1.5195497028918563,1.5195497028918563,3.698763544825648
+278,true,1.5621248637731227,1.5621248637731227,3.3913350038075096
+279,true,1.6099655068549867,1.6099655068549867,3.8716815897546764
+280,true,1.190512599182044,1.190512599182044,3.880039425028463
+281,true,1.4112553725167563,1.4112553725167563,3.758313929356346
+282,true,1.7438215873241525,1.7438215873241525,4.145734964247106
+283,true,1.3920747158303193,1.3920747158303193,3.719161667219829
+284,true,1.3451803007960965,1.3451803007960965,3.767064094148237
+285,true,1.2813724076867763,1.2813724076867763,3.754617620760507
+286,true,1.7514658257955276,1.7514658257955276,4.321965454972766
+287,true,1.4505832791562818,1.4505832791562818,3.3159488047893664
+288,true,1.964490387810811,1.964490387810811,4.223261710404559
+289,true,1.5869095951440357,1.5869095951440357,3.41881133235518
+290,true,1.4737386431837085,1.4737386431837085,4.022346415508156
+291,true,1.8319195189132926,1.8319195189132926,4.707881097803055
+292,true,1.3729848575432864,1.3729848575432864,3.6173533999669227
+293,true,1.9580605771396684,1.9580605771396684,4.282016383370361
+294,true,1.4275329412281363,1.4275329412281363,3.4942892826036474
+295,true,1.695957696269208,1.695957696269208,3.89279973836442
+296,true,1.7175602917835158,1.7175602917835158,4.049120115914259
+297,true,2.0393198532739207,2.0393198532739207,4.5581734707314085
+298,true,1.2575670744252732,1.2575670744252732,3.059770166100042
+299,true,1.489515090178933,1.489515090178933,3.9671771007957988
+300,true,1.5813971038974706,1.5813971038974706,3.9419941916705343
+301,true,1.7418082847659027,1.7418082847659027,4.038183804356725
+302,true,1.5395003255250306,1.5395003255250306,4.006495391399913
+303,true,1.7583540827675033,1.7583540827675033,4.113066838494202
+304,true,1.272690875909428,1.272690875909428,3.6802763203842392
+305,true,1.3854950787338907,1.3854950787338907,3.661869524964353
+306,true,1.6006793053903672,1.6006793053903672,3.383612612826422
+307,true,1.5991364980061227,1.5991364980061227,3.777100586711148
+308,true,2.0153575680600353,2.0153575680600353,4.583708518098553
+309,true,1.9493949695757788,1.9493949695757788,4.238552987356972
+310,true,1.5919967705376323,1.5919967705376323,3.709109320872328
+311,true,1.5843548914124965,1.5843548914124965,3.5456192412593674
+312,true,1.413238410607534,1.413238410607534,3.672465309773166
+313,true,1.4547576588156674,1.4547576588156674,4.054025187867071
+314,true,1.699124039295338,1.699124039295338,4.594606060052904
+315,true,1.3650155412026483,1.3650155412026483,3.496545069908438
+316,true,1.5411591174365182,1.5411591174365182,3.749699747609708
+317,true,1.238298340180521,1.238298340180521,3.051113755847939
+318,true,1.9144442966848891,1.9144442966848891,4.692673960932498
+319,true,1.5031449116504683,1.5031449116504683,3.804859532752375
+320,true,1.672615864995017,1.672615864995017,3.8698779186503565
+321,true,1.457937956176289,1.457937956176289,3.4518343257370927
+322,true,1.756874239922358,1.756874239922358,4.01773954670625
+323,true,1.6161645552460544,1.6161645552460544,3.7236829087213414
+324,true,1.5365997225832653,1.5365997225832653,4.10655000002635
+325,true,1.4836262568469911,1.4836262568469911,3.419566387423208
+326,true,1.5983955610940046,1.5983955610940046,3.51439331100331
+327,true,1.2170393770048604,1.2170393770048604,3.3612497216087562
+328,true,1.543614196507575,1.543614196507575,3.541742102411004
+329,true,1.6944332843856387,1.6944332843856387,3.8278883974778006
+330,true,1.6709576589374202,1.6709576589374202,4.0586506977054455
+331,true,1.3439577096720214,1.3439577096720214,3.66945558844034
+332,true,1.551822825936339,1.551822825936339,3.64327766035966
+333,true,1.6346947831930727,1.6346947831930727,3.8717608951284195
+334,true,1.5626820495558313,1.5626820495558313,4.323958498765296
+335,true,1.7473387079024036,1.7473387079024036,4.295609619453712
+336,true,1.3495188220510645,1.3495188220510645,3.674924462798825
+337,true,1.4402430629822165,1.4402430629822165,3.85193183034893
+338,true,1.4357319429076596,1.4357319429076596,3.740682727329855
+339,true,1.4306357398664542,1.4306357398664542,4.464331225078595
+340,true,1.7323377865037688,1.7323377865037688,3.943682538474043
+341,true,1.881865562203058,1.881865562203058,4.468956904297487
+342,true,1.3966851227151482,1.3966851227151482,4.232846913366065
+343,true,1.466135840254499,1.466135840254499,3.7336052566594975
+344,true,1.3428766419182,1.3428766419182,3.586209580918887
+345,true,1.4571093716544543,1.4571093716544543,3.820494908077694
+346,true,1.3136272149577801,1.3136272149577801,3.040564541627838
+347,true,1.5813477806999459,1.5813477806999459,3.589502100570865
+348,true,1.4841310151695728,1.4841310151695728,3.7095789208807184
+349,true,1.8245905716610071,1.8245905716610071,3.718362766130906
+350,true,1.4477731522050523,1.4477731522050523,3.5812517688974457
+351,true,1.6227670757916273,1.6227670757916273,3.800225239950953
+352,true,1.5323896629815077,1.5323896629815077,4.045390184855502
+353,true,1.7176117367433907,1.7176117367433907,3.8697912012784688
+354,true,1.4848741087817285,1.4848741087817285,3.7484570160182797
+355,true,1.7737307338837875,1.7737307338837875,4.335856486898491
+356,true,1.3314859062903628,1.3314859062903628,3.4913071666554067
+357,true,1.673372611913874,1.673372611913874,3.984368638497204
+358,true,1.871755412653434,1.871755412653434,4.228374555412362
+359,true,1.3299584054327251,1.3299584054327251,3.341026386926293
+360,true,1.4529611071570252,1.4529611071570252,3.788896632562718
+361,true,1.6467965299702776,1.6467965299702776,3.8600698707056282
+362,true,1.6100841470105653,1.6100841470105653,4.15870499554287
+363,true,1.5234493441783155,1.5234493441783155,3.9808699396158556
+364,true,1.6527619700035985,1.6527619700035985,3.991246289225501
+365,true,1.2012774721668553,1.2012774721668553,3.4315691047053107
+366,true,1.4713041432864058,1.4713041432864058,3.507183392498813
+367,true,1.425264960859607,1.425264960859607,3.674294414339645
+368,true,1.4551655286106935,1.4551655286106935,3.0654153062832483
+369,true,1.4473560682280495,1.4473560682280495,3.845383263340203
+370,true,1.8501507769766092,1.8501507769766092,4.058666567582153
+371,true,1.502167282389066,1.502167282389066,3.4532832748531224
+372,true,1.5621700059656802,1.5621700059656802,4.007337916258039
+373,true,1.3762907321569164,1.3762907321569164,3.432453142400082
+374,true,1.7709283226857664,1.7709283226857664,3.884978623871526
+375,true,1.811829528568731,1.811829528568731,4.210506981549323
+376,true,1.8197610498760457,1.8197610498760457,3.8915658578451073
+377,true,1.4535677426143327,1.4535677426143327,3.6753161653703907
+378,true,1.6166247663712874,1.6166247663712874,3.3401366068834997
+379,true,1.4869278540914936,1.4869278540914936,3.9676570441120576
+380,true,1.4143944199154839,1.4143944199154839,4.385777496366048
+381,true,1.4389711839784236,1.4389711839784236,3.4269175693720815
+382,true,1.8941894462583668,1.8941894462583668,4.218332727141845
+383,true,1.2778808626016882,1.2778808626016882,3.3212462646627925
+384,true,1.480054795875954,1.480054795875954,3.777936684488486
+385,true,1.8934845743509516,1.8934845743509516,4.613453895083935
+386,true,1.7003360496336661,1.7003360496336661,4.365345618721074
+387,true,1.4469271128664167,1.4469271128664167,3.1230290964554817
+388,true,1.5113472196465378,1.5113472196465378,3.648544829453335
+389,true,1.5686950883401258,1.5686950883401258,4.028431350700472
+390,true,1.7036336347252465,1.7036336347252465,4.227966867251441
+391,true,1.5814976361432738,1.5814976361432738,3.6783022551220106
+392,true,1.5499234635378625,1.5499234635378625,3.651217841421992
+393,true,1.5937259642087942,1.5937259642087942,3.5202730753887175
+394,true,1.8163123139438966,1.8163123139438966,4.215568027224527
+395,true,1.2944491728548218,1.2944491728548218,3.3499974337053
+396,true,1.5708410422660564,1.5708410422660564,3.708484694073777
+397,true,1.5169390640461131,1.5169390640461131,3.6783535594430825
+398,true,1.3570970225944075,1.3570970225944075,3.4078188487724876
+399,true,1.5274482177392728,1.5274482177392728,4.212579218605488
+400,true,1.44405400717944,1.44405400717944,3.1446663928383263
+401,true,1.2241571983830084,1.2241571983830084,3.7872859277955233
+402,true,1.3194787018301852,1.3194787018301852,4.034832891896527
+403,true,1.253364519447087,1.253364519447087,3.0100240806411405
+404,true,1.7684924597456786,1.7684924597456786,3.9665236881171113
+405,true,1.6420646863200268,1.6420646863200268,4.038661291510588
+406,true,1.2494041360320156,1.2494041360320156,3.1862494890526665
+407,true,1.3039157175562313,1.3039157175562313,3.5152490059894164
+408,true,1.3983811053054258,1.3983811053054258,3.6365175300419055
+409,true,1.7894967581705998,1.7894967581705998,4.527672613230798
+410,true,1.6174344360754866,1.6174344360754866,4.143823407993336
+411,true,1.9104325845813581,1.9104325845813581,4.328909400058235
+412,true,1.5052970034886308,1.5052970034886308,3.79517993210204
+413,true,1.6843204663658462,1.6843204663658462,3.7291704756342074
+414,true,1.3711106652413387,1.3711106652413387,3.3614977453781356
+415,true,1.3265883522853787,1.3265883522853787,3.954490868557078
+416,true,1.714784060847229,1.714784060847229,4.093960951396613
+417,true,1.2762124940746433,1.2762124940746433,3.4157915763084183
+418,true,1.5772513514086177,1.5772513514086177,4.090482723350529
+419,true,1.6557017835721812,1.6557017835721812,4.070424503584702
+420,true,1.3803111943084654,1.3803111943084654,2.9027988031175127
+421,true,1.471043510745189,1.471043510745189,3.363027124937417
+422,true,1.9058825359917655,1.9058825359917655,4.541633127442881
+423,true,1.7759746413316475,1.7759746413316475,4.161860534341669
+424,true,1.4970796969341014,1.4970796969341014,3.3457092914191073
+425,true,1.604821914372458,1.604821914372458,4.163129597410047
+426,true,1.3434640465023029,1.3434640465023029,3.503183384020432
+427,true,1.7909283250908867,1.7909283250908867,4.432529711520907
+428,true,1.4305848548281015,1.4305848548281015,4.195801250270497
+429,true,1.6605872449592112,1.6605872449592112,3.763321063454855
+430,true,1.62734775393244,1.62734775393244,4.28413142439014
+431,true,1.6746881091338206,1.6746881091338206,4.256537428696876
+432,true,1.9589219363272359,1.9589219363272359,4.326568498709326
+433,true,1.3108825360670564,1.3108825360670564,4.009813966445045
+434,true,1.779504068348248,1.779504068348248,4.372322060148389
+435,true,1.6038841723346688,1.6038841723346688,3.618920113664016
+436,true,1.678092298868299,1.678092298868299,4.067670743448085
+437,true,1.5746553695993313,1.5746553695993313,4.056648307546258
+438,true,1.9596110339421835,1.9596110339421835,4.415420741474174
+439,true,1.495554959065674,1.495554959065674,3.3940441996234862
+440,true,1.913961770672302,1.913961770672302,4.106746031964286
+441,true,1.461464772827889,1.461464772827889,3.6542318604654933
+442,true,1.635348397554095,1.635348397554095,4.249447678472901
+443,true,1.2950917733519398,1.2950917733519398,3.412594652155775
+444,true,1.635286942730621,1.635286942730621,4.103960566287238
+445,true,1.4686600363315807,1.4686600363315807,3.833666096754398
+446,true,1.7457794936490771,1.7457794936490771,4.140389783324262
+447,true,1.4192035489510444,1.4192035489510444,3.4005922109735436
+448,true,1.4497408750842735,1.4497408750842735,3.573785168875676
+449,true,1.3444367828555284,1.3444367828555284,3.30715059102383
+450,true,2.013158797782876,2.013158797782876,4.781842804774659
+451,true,1.5275066868860012,1.5275066868860012,3.833606172043597
+452,true,1.4574025725300443,1.4574025725300443,3.8680726487954726
+453,true,1.609782806681718,1.609782806681718,3.5168056761934223
+454,true,1.7319480829873648,1.7319480829873648,3.8170177774318796
+455,true,1.89925704317932,1.89925704317932,4.49547905196923
+456,true,1.5736728381265512,1.5736728381265512,4.122117079187223
+457,true,1.4300190066152716,1.4300190066152716,3.388253713848369
+458,true,1.6597934478648404,1.6597934478648404,4.042290607193892
+459,true,1.4056803744462627,1.4056803744462627,3.727881645754546
+460,true,1.328296727289158,1.328296727289158,3.4518036375603307
+461,true,1.4930137915404198,1.4930137915404198,3.5642820409055007
+462,true,1.7109121661225775,1.7109121661225775,4.228514239163537
+463,true,1.7859262258814865,1.7859262258814865,4.180430220841022
+464,true,1.4534157795067315,1.4534157795067315,3.1388569753476006
+465,true,1.413941898026954,1.413941898026954,3.4979545049556617
+466,true,1.4053085900716085,1.4053085900716085,3.803703673991416
+467,true,1.5243752203796765,1.5243752203796765,3.568024366328497
+468,true,1.6494719034563352,1.6494719034563352,3.9820391957635413
+469,true,1.7363304779194533,1.7363304779194533,4.693771323526073
+470,true,1.535983838118352,1.535983838118352,3.664451845996632
+471,true,1.7322816288886287,1.7322816288886287,4.041422221416481
+472,true,1.6764783552529234,1.6764783552529234,3.7427580483090086
+473,true,1.4981753892355487,1.4981753892355487,3.3896487499896684
+474,true,1.5925453685628301,1.5925453685628301,4.351462489850207
+475,true,1.3535346170154001,1.3535346170154001,3.357198167835566
+476,true,1.3908940911453165,1.3908940911453165,3.4176226881029894
+477,true,1.510520829845718,1.510520829845718,4.1248061864089065
+478,true,1.4412958733106482,1.4412958733106482,4.057489956652196
+479,true,2.1746314407306135,2.1746314407306135,4.6475629777536716
+480,true,1.761378384913052,1.761378384913052,4.063896525615353
+481,true,1.4438962242706141,1.4438962242706141,3.3074847768564855
+482,true,1.8268746756989522,1.8268746756989522,3.957924449654829
+483,true,1.6500362777622957,1.6500362777622957,3.664166849827382
+484,true,1.4531725996615523,1.4531725996615523,3.737860985941908
+485,true,1.2017246881990924,1.2017246881990924,4.044780875119455
+486,true,1.6800525678303817,1.6800525678303817,4.249595529311262
+487,true,1.2499329318313734,1.2499329318313734,3.7703529900078094
+488,true,1.3022251030987415,1.3022251030987415,3.5549949726639984
+489,true,1.6881535503288112,1.6881535503288112,4.134054694583777
+490,true,1.7298195286626996,1.7298195286626996,3.7209786353876897
+491,true,1.6116560185009312,1.6116560185009312,3.8661236341403344
+492,true,1.6560390955768471,1.6560390955768471,3.764657511877344
+493,true,1.4139685065887613,1.4139685065887613,4.327030500897412
+494,true,1.5392749522906342,1.5392749522906342,3.561896668018823
+495,true,1.6070136235872774,1.6070136235872774,3.920521288603782
+496,true,1.488061167974995,1.488061167974995,4.03132040145444
+497,true,1.7973129429053742,1.7973129429053742,3.989903789473352
+498,true,1.7055820694713124,1.7055820694713124,4.238191392423554
+499,true,1.8824253602169427,1.8824253602169427,4.050748036485077
+500,false,1.3208755672750174,1.3208755672750174,2.769966467842186

splits/visualize_splits.py

@@ -0,0 +1,113 @@
+"""Create AhmedML split diagnostic plots."""
+
+from __future__ import annotations
+
+import csv
+import json
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+from matplotlib.lines import Line2D
+import numpy as np
+
+from generate_splits import load_force_mom, run_id
+
+
+SCRIPT_DIR = Path(__file__).resolve().parent
+PACKAGE_ROOT = SCRIPT_DIR
+DATA_DIR = SCRIPT_DIR
+DOCS_DIR = SCRIPT_DIR
+SPLITS_DIR = SCRIPT_DIR
+MANIFEST = SPLITS_DIR / "manifest.json"
+CHAMFER = DATA_DIR / "chamfer_metrics.csv"
+IMAGE = DATA_DIR / "image_metrics.csv"
+OUT = DOCS_DIR / "split_diagnostics.png"
+
+
+def _ids(manifest: dict[str, list[str]], key: str) -> set[int]:
+    return {run_id(case_id) for case_id in manifest[key]}
+
+
+def _load_scores(path: Path, column: str) -> dict[int, float]:
+    with path.open(encoding="utf-8", newline="") as f:
+        return {int(row["run"]): float(row[column]) for row in csv.DictReader(f) if row.get(column, "") != ""}
+
+
+def _arrays(values: dict[int, float]) -> tuple[np.ndarray, np.ndarray]:
+    runs = np.asarray(sorted(values))
+    arr = np.asarray([values[int(run)] for run in runs], dtype=float)
+    return runs, arr
+
+
+def _plot_partitioned(
+    ax,
+    runs: np.ndarray,
+    values: np.ndarray,
+    train_ids: set[int],
+    val_ids: set[int],
+    test_ids: set[int],
+    *,
+    title: str,
+    ylabel: str,
+    colors: dict[str, str],
+) -> None:
+    train = np.asarray([int(run) in train_ids for run in runs])
+    val = np.asarray([int(run) in val_ids for run in runs])
+    test = np.asarray([int(run) in test_ids for run in runs])
+    ax.scatter(runs[train], values[train], s=30, color=colors["train"], linewidth=0, alpha=0.58)
+    ax.scatter(runs[val], values[val], s=46, color=colors["val"], linewidth=0, alpha=0.95)
+    ax.scatter(runs[test], values[test], s=46, color=colors["test"], linewidth=0, alpha=0.95)
+    ax.set_title(title)
+    ax.set_xlabel("run")
+    ax.set_ylabel(ylabel)
+    ax.grid(True, color="#e1e6eb", lw=0.7)
+    ax.spines["top"].set_visible(False)
+    ax.spines["right"].set_visible(False)
+
+
+def main() -> None:
+    manifest = json.loads(MANIFEST.read_text(encoding="utf-8"))
+    force_records, _ = load_force_mom()
+    geometry_scores = _load_scores(CHAMFER, "ood_score")
+    image_scores = _load_scores(IMAGE, "image_wake_score")
+
+    runs, cd = _arrays({rid: values["cd"] for rid, values in force_records.items()})
+    _, geometry = _arrays(geometry_scores)
+    _, image_wake = _arrays(image_scores)
+
+    colors = {"train": "#cfd5dc", "val": "#c28f22", "test": "#2f8f61"}
+    fig, axes = plt.subplots(2, 3, figsize=(14.0, 8.0), constrained_layout=True)
+    panels = [
+        ("full", cd, "Full random baseline", "Cd"),
+        ("high_drag", cd, "High-drag holdout", "Cd"),
+        ("low_drag", cd, "Low-drag holdout", "Cd"),
+        ("geometry", geometry, "STL-Chamfer geometry holdout", "mean 10-NN Chamfer"),
+        ("image_wake", image_wake, "Image wake holdout", "UxMean image wake score"),
+        ("image_wake", cd, "Image wake holdout on Cd", "Cd"),
+    ]
+    for ax, (prefix, values, title, ylabel) in zip(axes.flat, panels):
+        _plot_partitioned(
+            ax,
+            runs,
+            values,
+            _ids(manifest, f"{prefix}_train"),
+            _ids(manifest, f"{prefix}_val"),
+            _ids(manifest, f"{prefix}_test"),
+            title=title,
+            ylabel=ylabel,
+            colors=colors,
+        )
+    legend_handles = [
+        Line2D([0], [0], marker="o", color="none", markerfacecolor=colors["train"], markeredgewidth=0, markersize=8, label="train"),
+        Line2D([0], [0], marker="o", color="none", markerfacecolor=colors["val"], markeredgewidth=0, markersize=8, label="val"),
+        Line2D([0], [0], marker="o", color="none", markerfacecolor=colors["test"], markeredgewidth=0, markersize=8, label="test"),
+    ]
+    fig.legend(handles=legend_handles, frameon=False, loc="upper center", ncol=3, bbox_to_anchor=(0.5, 0.99))
+    fig.suptitle("AhmedML split diagnostics", fontsize=13)
+    DOCS_DIR.mkdir(parents=True, exist_ok=True)
+    fig.savefig(OUT, dpi=180)
+    print(f"Wrote {OUT}")
+
+
+if __name__ == "__main__":
+    main()