💡 Best Practices

Q: How to organize large simulation projects?

Recommended directory structure:

my_fusion_project/
├── models/                 # Modelica models
│   ├── package.mo
│   └── ...
├── configs/                # Configuration files
│   ├── baseline.json
│   ├── sensitivity.json
│   └── ...
├── scripts/                # Helper scripts
│   ├── prepare_data.py
│   └── post_analysis.py
├── data/                   # Input data
│   └── external_data.csv
├── results/                # Simulation results (auto-generated)
└── reports/                # Final reports
Q: How to version control configuration files?

Use Git for version control:

# Initialize Git repository
git init

# Create .gitignore
echo "results/" >> .gitignore
echo "*.log" >> .gitignore
echo "*.pyc" >> .gitignore
echo "__pycache__/" >> .gitignore

# Commit configuration files
git add configs/*.json
git commit -m "feat: add baseline configuration"
Q: How to run distributed simulations on multiple machines?

Currently, TRICYS does not directly support distributed computing, but you can:

  1. Manually split tasks:
  2. Divide the parameter scan into multiple subsets
  3. Run different subsets on different machines

  4. Manually merge the results

  5. Use a cluster scheduler (like SLURM):

  6. Submit each parameter combination as a separate job
  7. Use a post-processing script to aggregate the results
Q: How to optimize model performance?

At the Modelica model level: 1. Simplify the model structure, avoid overly complex equations 2. Use appropriate numerical solver settings 3. Avoid algebraic loops and excessive events

At the TRICYS level: 1. Enable concurrent execution 2. Reduce the number of output variables 3. Use co-simulation to replace complex subsystems