Anto01
e3bdb08a22
- Add validation framework (config, model, results, study validators) - Add Claude Code skills (create-study, run-optimization, generate-report, troubleshoot, analyze-model) - Add Atomizer Dashboard (React frontend + FastAPI backend) - Reorganize docs into structured directories (00-09) - Add neural surrogate modules and training infrastructure - Add multi-objective optimization support 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
386 lines
12 KiB
Python
386 lines
12 KiB
Python
"""
|
|
Training Data Exporter for AtomizerField
|
|
|
|
This module exports training data from Atomizer optimization runs for AtomizerField neural network training.
|
|
It saves NX Nastran input (.dat) and output (.op2) files along with metadata for each trial.
|
|
|
|
Usage:
|
|
from optimization_engine.training_data_exporter import create_exporter_from_config
|
|
|
|
exporter = create_exporter_from_config(config)
|
|
if exporter:
|
|
exporter.export_trial(trial_number, design_vars, results, simulation_files)
|
|
exporter.finalize()
|
|
"""
|
|
|
|
import json
|
|
import shutil
|
|
from pathlib import Path
|
|
from datetime import datetime
|
|
from typing import Dict, Any, Optional, List
|
|
import logging
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
class TrainingDataExporter:
|
|
"""
|
|
Exports training data for AtomizerField neural network training.
|
|
|
|
After each FEA solve, saves:
|
|
- Input: NX Nastran .dat file (BDF format)
|
|
- Output: NX Nastran .op2 file (binary results)
|
|
- Metadata: JSON with design parameters, objectives, constraints
|
|
|
|
Directory structure:
|
|
{export_dir}/
|
|
├── trial_0001/
|
|
│ ├── input/
|
|
│ │ └── model.bdf
|
|
│ ├── output/
|
|
│ │ └── model.op2
|
|
│ └── metadata.json
|
|
├── trial_0002/
|
|
│ └── ...
|
|
├── study_summary.json
|
|
└── README.md
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
export_dir: Path,
|
|
study_name: str,
|
|
design_variable_names: List[str],
|
|
objective_names: List[str],
|
|
constraint_names: Optional[List[str]] = None,
|
|
metadata: Optional[Dict[str, Any]] = None
|
|
):
|
|
"""
|
|
Initialize the training data exporter.
|
|
|
|
Args:
|
|
export_dir: Root directory for exported training data
|
|
study_name: Name of the optimization study
|
|
design_variable_names: List of design variable names
|
|
objective_names: List of objective function names
|
|
constraint_names: List of constraint names (optional)
|
|
metadata: Additional study metadata (optional)
|
|
"""
|
|
self.export_dir = Path(export_dir)
|
|
self.study_name = study_name
|
|
self.design_variable_names = design_variable_names
|
|
self.objective_names = objective_names
|
|
self.constraint_names = constraint_names or []
|
|
self.study_metadata = metadata or {}
|
|
|
|
self.trial_count = 0
|
|
self.exported_trials: List[Dict[str, Any]] = []
|
|
|
|
# Create root export directory
|
|
self.export_dir.mkdir(parents=True, exist_ok=True)
|
|
|
|
logger.info(f"Training data exporter initialized: {self.export_dir}")
|
|
|
|
# Create README
|
|
self._create_readme()
|
|
|
|
def export_trial(
|
|
self,
|
|
trial_number: int,
|
|
design_variables: Dict[str, float],
|
|
results: Dict[str, Any],
|
|
simulation_files: Dict[str, Path]
|
|
) -> bool:
|
|
"""
|
|
Export training data for a single trial.
|
|
|
|
Args:
|
|
trial_number: Optuna trial number
|
|
design_variables: Dict of design parameter names and values
|
|
results: Dict containing objectives, constraints, and other results
|
|
simulation_files: Dict with paths to 'dat_file' and 'op2_file'
|
|
|
|
Returns:
|
|
True if export successful, False otherwise
|
|
"""
|
|
try:
|
|
# Create trial directory
|
|
trial_dir = self.export_dir / f"trial_{trial_number:04d}"
|
|
trial_dir.mkdir(parents=True, exist_ok=True)
|
|
|
|
input_dir = trial_dir / "input"
|
|
output_dir = trial_dir / "output"
|
|
input_dir.mkdir(exist_ok=True)
|
|
output_dir.mkdir(exist_ok=True)
|
|
|
|
# Copy .dat file (Nastran input deck)
|
|
dat_file = simulation_files.get('dat_file')
|
|
if dat_file and Path(dat_file).exists():
|
|
shutil.copy2(dat_file, input_dir / "model.bdf")
|
|
logger.debug(f"Copied .dat file: {dat_file} -> {input_dir / 'model.bdf'}")
|
|
else:
|
|
logger.warning(f"Trial {trial_number}: .dat file not found at {dat_file}")
|
|
return False
|
|
|
|
# Copy .op2 file (Nastran binary results)
|
|
op2_file = simulation_files.get('op2_file')
|
|
if op2_file and Path(op2_file).exists():
|
|
shutil.copy2(op2_file, output_dir / "model.op2")
|
|
logger.debug(f"Copied .op2 file: {op2_file} -> {output_dir / 'model.op2'}")
|
|
else:
|
|
logger.warning(f"Trial {trial_number}: .op2 file not found at {op2_file}")
|
|
return False
|
|
|
|
# Create metadata.json
|
|
metadata = self._create_trial_metadata(
|
|
trial_number,
|
|
design_variables,
|
|
results
|
|
)
|
|
|
|
metadata_path = trial_dir / "metadata.json"
|
|
with open(metadata_path, 'w') as f:
|
|
json.dump(metadata, f, indent=2)
|
|
|
|
logger.info(f"Exported training data for trial {trial_number}")
|
|
|
|
self.trial_count += 1
|
|
self.exported_trials.append(metadata)
|
|
|
|
return True
|
|
|
|
except Exception as e:
|
|
logger.error(f"Failed to export trial {trial_number}: {e}", exc_info=True)
|
|
return False
|
|
|
|
def _create_trial_metadata(
|
|
self,
|
|
trial_number: int,
|
|
design_variables: Dict[str, float],
|
|
results: Dict[str, Any]
|
|
) -> Dict[str, Any]:
|
|
"""
|
|
Create metadata dictionary for a trial.
|
|
|
|
Args:
|
|
trial_number: Optuna trial number
|
|
design_variables: Design parameter values
|
|
results: Optimization results
|
|
|
|
Returns:
|
|
Metadata dictionary
|
|
"""
|
|
metadata = {
|
|
"trial_number": trial_number,
|
|
"timestamp": datetime.now().isoformat(),
|
|
"atomizer_study": self.study_name,
|
|
"design_parameters": design_variables,
|
|
"results": {}
|
|
}
|
|
|
|
# Extract objectives
|
|
if "objectives" in results:
|
|
metadata["results"]["objectives"] = results["objectives"]
|
|
|
|
# Extract constraints
|
|
if "constraints" in results:
|
|
metadata["results"]["constraints"] = results["constraints"]
|
|
|
|
# Extract any scalar results (max_stress, max_displacement, etc.)
|
|
for key in ["max_stress", "max_displacement", "mass", "volume"]:
|
|
if key in results:
|
|
metadata["results"][key] = results[key]
|
|
|
|
# Add any additional result fields
|
|
for key, value in results.items():
|
|
if key not in ["objectives", "constraints"] and isinstance(value, (int, float, str, bool)):
|
|
metadata["results"][key] = value
|
|
|
|
return metadata
|
|
|
|
def finalize(self) -> None:
|
|
"""
|
|
Finalize the training data export.
|
|
|
|
Creates study_summary.json with metadata about the entire study.
|
|
"""
|
|
try:
|
|
summary = {
|
|
"study_name": self.study_name,
|
|
"total_trials": self.trial_count,
|
|
"design_variables": self.design_variable_names,
|
|
"objectives": self.objective_names,
|
|
"constraints": self.constraint_names,
|
|
"export_timestamp": datetime.now().isoformat(),
|
|
"metadata": self.study_metadata
|
|
}
|
|
|
|
summary_path = self.export_dir / "study_summary.json"
|
|
with open(summary_path, 'w') as f:
|
|
json.dump(summary, f, indent=2)
|
|
|
|
logger.info(f"Training data export finalized: {self.trial_count} trials exported")
|
|
logger.info(f"Summary saved to: {summary_path}")
|
|
|
|
except Exception as e:
|
|
logger.error(f"Failed to finalize training data export: {e}", exc_info=True)
|
|
|
|
def _create_readme(self) -> None:
|
|
"""Create README.md explaining the training data structure."""
|
|
readme_content = f"""# AtomizerField Training Data
|
|
|
|
**Study Name**: {self.study_name}
|
|
**Generated**: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}
|
|
|
|
## Directory Structure
|
|
|
|
```
|
|
{self.export_dir.name}/
|
|
├── trial_0001/
|
|
│ ├── input/
|
|
│ │ └── model.bdf # NX Nastran input deck (BDF format)
|
|
│ ├── output/
|
|
│ │ └── model.op2 # NX Nastran binary results (OP2 format)
|
|
│ └── metadata.json # Design parameters, objectives, constraints
|
|
├── trial_0002/
|
|
│ └── ...
|
|
├── study_summary.json # Overall study metadata
|
|
└── README.md # This file
|
|
```
|
|
|
|
## Design Variables
|
|
|
|
{chr(10).join(f'- {name}' for name in self.design_variable_names)}
|
|
|
|
## Objectives
|
|
|
|
{chr(10).join(f'- {name}' for name in self.objective_names)}
|
|
|
|
## Constraints
|
|
|
|
{chr(10).join(f'- {name}' for name in self.constraint_names) if self.constraint_names else '- None'}
|
|
|
|
## Usage with AtomizerField
|
|
|
|
### 1. Parse Training Data
|
|
|
|
```bash
|
|
cd Atomizer-Field
|
|
python batch_parser.py --data-dir "{self.export_dir}"
|
|
```
|
|
|
|
This converts BDF/OP2 files to PyTorch Geometric format.
|
|
|
|
### 2. Validate Parsed Data
|
|
|
|
```bash
|
|
python validate_parsed_data.py
|
|
```
|
|
|
|
### 3. Train Neural Network
|
|
|
|
```bash
|
|
python train.py --data-dir "training_data/parsed/" --epochs 200
|
|
```
|
|
|
|
### 4. Use Trained Model in Atomizer
|
|
|
|
```bash
|
|
cd ../Atomizer
|
|
python run_optimization.py --config studies/{self.study_name}/workflow_config.json --use-neural
|
|
```
|
|
|
|
## File Formats
|
|
|
|
- **BDF (.bdf)**: Nastran Bulk Data File - contains mesh, materials, loads, BCs
|
|
- **OP2 (.op2)**: Nastran Output2 - binary results with displacements, stresses, etc.
|
|
- **metadata.json**: Human-readable trial metadata
|
|
|
|
## AtomizerField Documentation
|
|
|
|
See `Atomizer-Field/docs/` for complete documentation on:
|
|
- Neural network architecture
|
|
- Training procedures
|
|
- Integration with Atomizer
|
|
- Uncertainty quantification
|
|
|
|
---
|
|
|
|
*Generated by Atomizer Training Data Exporter*
|
|
"""
|
|
|
|
readme_path = self.export_dir / "README.md"
|
|
with open(readme_path, 'w', encoding='utf-8') as f:
|
|
f.write(readme_content)
|
|
|
|
logger.debug(f"Created README: {readme_path}")
|
|
|
|
|
|
def create_exporter_from_config(config: Dict[str, Any]) -> Optional[TrainingDataExporter]:
|
|
"""
|
|
Factory function to create TrainingDataExporter from workflow configuration.
|
|
|
|
Args:
|
|
config: Workflow configuration dictionary
|
|
|
|
Returns:
|
|
TrainingDataExporter instance if enabled, None otherwise
|
|
|
|
Example config YAML:
|
|
training_data_export:
|
|
enabled: true
|
|
export_dir: "atomizer_field_training_data/beam_study_001"
|
|
"""
|
|
if not config.get("training_data_export", {}).get("enabled", False):
|
|
logger.info("Training data export is disabled")
|
|
return None
|
|
|
|
export_config = config["training_data_export"]
|
|
|
|
# Get export directory
|
|
export_dir = export_config.get("export_dir")
|
|
if not export_dir:
|
|
logger.error("Training data export enabled but 'export_dir' not specified")
|
|
return None
|
|
|
|
# Get study name
|
|
study_name = config.get("study_name", "unnamed_study")
|
|
|
|
# Get design variable names
|
|
design_vars = config.get("design_variables", [])
|
|
design_var_names = [dv.get("name", dv.get("parameter", f"var_{i}"))
|
|
for i, dv in enumerate(design_vars)]
|
|
|
|
# Get objective names
|
|
objectives = config.get("objectives", [])
|
|
objective_names = [obj.get("name", f"obj_{i}")
|
|
for i, obj in enumerate(objectives)]
|
|
|
|
# Get constraint names
|
|
constraints = config.get("constraints", [])
|
|
constraint_names = [c.get("name", f"constraint_{i}")
|
|
for i, c in enumerate(constraints)]
|
|
|
|
# Additional metadata
|
|
metadata = {
|
|
"atomizer_version": config.get("version", "unknown"),
|
|
"optimization_algorithm": config.get("optimization", {}).get("algorithm", "unknown"),
|
|
"n_trials": config.get("optimization", {}).get("n_trials", 0)
|
|
}
|
|
|
|
try:
|
|
exporter = TrainingDataExporter(
|
|
export_dir=Path(export_dir),
|
|
study_name=study_name,
|
|
design_variable_names=design_var_names,
|
|
objective_names=objective_names,
|
|
constraint_names=constraint_names,
|
|
metadata=metadata
|
|
)
|
|
logger.info("Training data exporter created successfully")
|
|
return exporter
|
|
|
|
except Exception as e:
|
|
logger.error(f"Failed to create training data exporter: {e}", exc_info=True)
|
|
return None
|