Official Resources
- Source Repository: https://github.com/mumax/plus
- Documentation: Included in repository
- License: Open source
Overview
mumax+ (mumax plus) is a versatile and extensible GPU-accelerated micromagnetic simulator written in C++ and CUDA with a Python interface. It is the successor to mumax3, offering more extensibility, Python scripting, and additional physics capabilities beyond standard micromagnetics.
Scientific domain: GPU-accelerated micromagnetic simulation
Target user community: Researchers needing fast, extensible GPU micromagnetic simulations with Python control
Theoretical Methods
- Landau-Lifshitz-Gilbert (LLG) equation
- Finite-difference method on GPU
- FFT-based demagnetization
- Exchange, anisotropy, Zeeman energies
- Dzyaloshinskii-Moriya interaction
- Spin-transfer torque
- Thermal fluctuations
- Reduced units system
Capabilities (CRITICAL)
- GPU-accelerated micromagnetic simulation
- Python scripting interface
- Extensible C++ architecture
- All standard micromagnetic energy terms
- DMI support
- Spin-transfer torque
- Thermal fluctuations
- Custom energy terms (extensible)
- Batch simulation mode
- Real-time visualization
Sources: GitHub repository
Key Strengths
GPU Performance:
- CUDA-accelerated
- Fast FFT demagnetization
- Large system sizes
- Efficient memory usage
Extensibility:
- C++ plugin architecture
- Custom energy terms
- Custom evolvers
- Python scripting
- More flexible than mumax3
Python Interface:
- Full Python control
- Jupyter notebook compatible
- Easy post-processing
- Automation and scripting
mumax3 Compatibility:
- Familiar input format
- Migration path from mumax3
- Same reduced units
- Enhanced capabilities
Inputs & Outputs
-
Input formats:
- mumax+ input files
- Python scripts
- Material parameters
-
Output data types:
- Magnetization fields (OVF)
- Energy vs time
- Average magnetization
- Table data
- PNG snapshots
Interfaces & Ecosystem
- Python: Scripting interface
- C++/CUDA: Core computation
- OVF format: Standard micromagnetic output
- ParaView/MumaxView: Visualization
Performance Characteristics
- Speed: Very fast (GPU)
- Accuracy: High (validated)
- System size: Millions of cells
- Memory: GPU memory limited
Computational Cost
- Small systems: Seconds
- Large systems: Minutes to hours
- Typical: Very efficient with GPU
Limitations & Known Constraints
- NVIDIA GPU only: Requires CUDA
- Finite differences: No FEM
- Regular grids: No adaptive meshing
- In development: Not as mature as mumax3
- Documentation: Limited
Comparison with Other Codes
- vs mumax3: mumax+ is extensible, has Python interface, still GPU
- vs OOMMF: mumax+ is GPU, OOMMF is CPU
- vs MicroMagnetic.jl: mumax+ is NVIDIA-only, MicroMagnetic.jl is multi-platform
- Unique strength: Extensible GPU micromagnetic simulator with Python interface, mumax3 successor
Application Areas
Fast Micromagnetics:
- Large-scale domain dynamics
- Parameter sweeps
- Hysteresis loops
- Real-time simulation
Spintronics:
- STT-MRAM
- Domain wall motion
- Skyrmion dynamics
- Spin-orbit torque
Custom Physics:
- Extended Hamiltonians
- Coupled systems
- Novel energy terms
- Research code development
Best Practices
GPU Setup:
- Use NVIDIA GPU with sufficient memory
- Monitor GPU utilization
- Choose appropriate cell size
- Use reduced units
Python Scripting:
- Use Python for automation
- Post-process in Python
- Generate input files programmatically
- Use Jupyter for interactive analysis
Community and Support
- Open source on GitHub
- Developed alongside mumax3
- Active development
- GitHub Discussions for support
Verification & Sources
Primary sources:
- GitHub: https://github.com/mumax/plus
- Related: A. Vansteenkiste et al., AIP Adv. 4, 107133 (2014) (mumax3 paper)
Confidence: VERIFIED
Verification status: ✅ VERIFIED
- Source code: ACCESSIBLE (GitHub)
- Documentation: Included in repository
- Active development: Ongoing
- Specialized strength: Extensible GPU micromagnetic simulator with Python interface, mumax3 successor