Official Resources
- Homepage: https://github.com/Mofahdi/mace_phonopy
- Source Repository: https://github.com/Mofahdi/mace_phonopy
- License: Open Source
Overview
mace_phonopy is a code to generate second-order interatomic force constants (IFCs) from MACE machine learning potentials for use with Phonopy. It bridges MACE ML potentials with standard phonon calculation workflows.
Scientific domain: Machine learning potentials, phonon calculations, force constants
Target user community: Researchers using MACE potentials for phonon calculations
Theoretical Methods
- MACE machine learning potentials
- Finite displacement method
- Force constant extraction
- Phonopy integration
- Stability analysis
Capabilities (CRITICAL)
- Generate FORCE_CONSTANTS from MACE
- Phonopy-compatible output
- Stability checking
- Phonon dispersion plotting
- band.conf generation
- Automated workflow
Key Strengths
MACE Integration:
- Direct MACE potential support
- Fast force evaluation
- Near-DFT accuracy
- GPU acceleration
Phonopy Compatibility:
- Standard FORCE_CONSTANTS format
- band.conf generation
- Seamless integration
- Automated workflow
Stability Analysis:
- Automatic stability check
- Imaginary frequency detection
- Stability criteria customization
Inputs & Outputs
-
Input formats:
- Crystal structure (ASE-readable)
- MACE potential file
- Supercell parameters
-
Output data types:
- FORCE_CONSTANTS file
- Stability status
- band.conf file
- Phonon dispersion plot
Interfaces & Ecosystem
- MACE: ML potential
- Phonopy: Phonon calculations
- ASE: Structure handling
- PyTorch: ML backend
Advanced Features
MACE Integration:
- Direct MACE potential loading
- GPU-accelerated force evaluation
- Batch displacement calculations
- Efficient force constant extraction
Workflow Automation:
- Automatic supercell generation
- Displacement pattern creation
- Force constant symmetrization
- band.conf generation for Phonopy
Stability Analysis:
- Automatic imaginary frequency detection
- Customizable stability threshold
- Stability report generation
- Quick screening capability
Quality Control:
- Force accuracy checking
- Symmetry validation
- Acoustic sum rule verification
- Comparison with DFT option
Performance Characteristics
- Speed: Fast with GPU (seconds to minutes)
- Accuracy: Near-DFT quality (depends on MACE training)
- Scalability: Large systems feasible (100s-1000s atoms)
- GPU requirement: CUDA-capable GPU recommended
Computational Cost
- Force evaluation: Fast with GPU (milliseconds per structure)
- Supercell forces: Minutes for typical systems
- Total workflow: Minutes to hours (depends on system size)
- Memory: Moderate (MACE model + structures)
Limitations & Known Constraints
- Requires trained MACE potential
- MACE-specific
- Potential quality dependent
- Requires PyTorch
Comparison with Other Codes
- vs Phonopy direct: mace_phonopy uses ML forces; Phonopy uses DFT
- vs autoplex: Different ML potential (MACE vs MTP/GAP)
- Unique strength: Direct MACE-to-Phonopy bridge
Best Practices
Potential Quality:
- Use well-trained MACE potential
- Validate against DFT phonons
- Check force accuracy
- Test on known systems
Calculations:
- Use appropriate supercell size
- Check convergence
- Validate stability results
Application Areas
- High-throughput phonon screening
- Large system phonons
- ML potential validation
- Rapid phonon estimation
- Materials discovery
- Stability screening
Community and Support
- License: Open source
- Development: Active GitHub repository
- Documentation: README with examples
- Support: GitHub issues
- User base: MACE and Phonopy users
- Integration: MACE-Phonopy bridge
Verification & Sources
Primary sources:
- GitHub: https://github.com/Mofahdi/mace_phonopy
Confidence: VERIFIED
Verification status: ✅ VERIFIED
- Source code: OPEN (GitHub)
- Active development