Skip to content

README.md

Latest commit

 

History

History
72 lines (54 loc) · 2.87 KB

README.md

File metadata and controls

72 lines (54 loc) · 2.87 KB
Raw

Configuration Files

All configuration files for PyDimension should be maintained in this directory (pydimension/configs/).

Config Families

PyDimension carries three config families that correspond to different symmetry types and pipeline versions:

Config file Pipeline Symmetry type Status
config_synthetic.json Legacy 2.0 translational production
config_synthetic_with_noise.json Legacy 2.0 translational (5% noise) production
config_keyhole.json Legacy 2.0 translational (real data) production
config_translation.json 3.0 translational active
config_translation_v3.json 3.0 translational (alias) kept for backward compat
config_rotation.json 3.0 rotational scaffold
config_scaling.json 3.0 scaling scaffold

The 2.0 benchmark also has a frozen copy at legacy/pydimension_v2/configs/config_synthetic_v2.json.

Config Section Naming (3.0)

The 3.0 config uses section names that match the new module names:

{
  "DATA_GENERATION": { },
  "DATA_PREPROCESSING": { "preprocessing_method": "dimensional_analysis" },
  "INTRINSIC_COORDINATE": { "method": "pca_sir" },
  "SYMMETRY_DISCOVERY": { "symmetry_type": "translational", "encoder_name": "translational" },
  "OUTPUT": { "output_dir": "output_v3_translation" },
  "DATA_GENERATION_OUTPUT": { },
  "DATA_PREPROCESSING_OUTPUT": { },
  "DIMENSIONAL_ANALYSIS_OUTPUT": { },
  "INTRINSIC_COORDINATE_OUTPUT": { },
  "SYMMETRY_DISCOVERY_OUTPUT": { }
}

Legacy 2.0 sections (CONSTRAINT_FILTERING, OPTIMIZATION_DISCOVERY) are still accepted as fallbacks by the config loaders, so existing user configs continue to work.

Usage

# 3.0 translational pipeline
python run_pipeline.py --pipeline-version v3 --config pydimension/configs/config_translation.json

# Legacy 2.0 benchmark
python run_pipeline.py --pipeline-version v2 --config legacy/pydimension_v2/configs/config_synthetic_v2.json

# Individual modules
python generate_data.py --config pydimension/configs/config_synthetic.json --plot
python preprocess_data.py --config pydimension/configs/config_synthetic.json --plot
python discover_symmetry.py --config pydimension/configs/config_translation.json

Extending the Config

When adding a new symmetry type or module:

  1. Copy an existing config file (e.g., config_translation.json).
  2. Update SYMMETRY_DISCOVERY.symmetry_type and encoder_name.
  3. Update OUTPUT.output_dir to a new directory name.
  4. Add any module-specific parameters.
  5. Register the corresponding benchmark in pydimension/benchmarks/.

Config File Priority

When loading configs, the system checks in this order:

  1. Unified OUTPUT section for shared output paths.
  2. Module-specific output sections (e.g., SYMMETRY_DISCOVERY_OUTPUT) for filenames.
  3. Legacy fallback sections (e.g., OPTIMIZATION_DISCOVERY) for backward compatibility.