Official Resources
- Source Repository: https://github.com/magnum-af/magnum.af
- Documentation: https://magnum-af.github.io/
- License: Open source
Overview
magnum.af is a finite-difference/finite-element micromagnetic simulation package that combines CPU and GPU solvers. It supports standard micromagnetic energy terms plus spin-transfer torque, spin-orbit torque, and true periodic boundary conditions for stray field calculation.
Scientific domain: Micromagnetic simulation, spintronics, domain dynamics
Target user community: Researchers simulating magnetization dynamics with advanced boundary conditions and spin-torque effects
Theoretical Methods
- Landau-Lifshitz-Gilbert (LLG) equation
- Finite-difference and finite-element methods
- FFT-based demagnetization
- Spin-transfer torque (Zhang-Li, Slonczewski)
- Spin-orbit torque
- True periodic boundary conditions
- Thermal fluctuations
Capabilities (CRITICAL)
- Micromagnetic simulation (LLG dynamics)
- Energy minimization
- Spin-transfer torque simulation
- Spin-orbit torque simulation
- True periodic boundary conditions
- GPU acceleration (CUDA)
- Finite-difference and finite-element solvers
- Domain wall dynamics
- Skyrmion simulation
Sources: GitHub repository, published in J. Magn. Magn. Mater.
Key Strengths
Advanced Boundary Conditions:
- True periodic BC for stray field
- Eliminates edge effects
- Accurate for infinite thin films
- Novel approach for periodic systems
Spin-Torque Effects:
- Spin-transfer torque (STT)
- Spin-orbit torque (SOT)
- Zhang-Li and Slonczewski models
- Current-driven dynamics
GPU Acceleration:
- CUDA implementation
- Fast demagnetization calculation
- Large system sizes feasible
- Efficient time integration
Inputs & Outputs
-
Input formats:
- JSON configuration files
- Mesh files (for FEM)
- Material parameters
-
Output data types:
- Magnetization fields
- Energy vs time
- Hysteresis loops
- VTK output for visualization
Interfaces & Ecosystem
- Python: Scripting interface
- C++: Core computation
- CUDA: GPU acceleration
- ParaView: VTK visualization
Performance Characteristics
- Speed: Fast with GPU
- Accuracy: High (validated)
- System size: Millions of cells (GPU)
- Parallelization: GPU (CUDA)
Computational Cost
- Small systems: Minutes
- Large systems: Hours (GPU)
- Typical: Moderate with GPU
Limitations & Known Constraints
- CUDA required: GPU acceleration needs NVIDIA
- Limited documentation: Could be more extensive
- Community: Smaller than OOMMF/Mumax3
- No DFT integration: Standalone micromagnetics
Comparison with Other Codes
- vs OOMMF: magnum.af has GPU and true PBC, OOMMF is NIST standard
- vs Mumax3: magnum.af has FEM + true PBC, Mumax3 is pure FD/GPU
- vs Spirit: magnum.af is micromagnetic, Spirit is atomistic
- Unique strength: True periodic boundary conditions for stray field, FEM+FD solvers, spin-torque support
Application Areas
Spintronics:
- STT-MRAM switching
- SOT switching
- Domain wall motion by current
- Skyrmion Hall effect
Thin Films:
- Periodic domain structures
- Stripe domains
- Skyrmion lattices
- Domain wall pinning
Standard Problems:
- µMAG benchmarks
- Method comparison
- PBC-specific problems
Best Practices
Mesh Selection:
- Use cell size ≤ exchange length
- Test PBC effects
- Validate against analytical solutions
- Compare with non-PBC results
GPU Usage:
- Ensure CUDA compatibility
- Monitor GPU memory usage
- Use appropriate block sizes
- Compare CPU/GPU results
Community and Support
- Open source on GitHub
- Developed at TU Wien / Danube University Krems
- Published methodology
- Active development
Verification & Sources
Primary sources:
- GitHub: https://github.com/magnum-af/magnum.af
- P. Heistracher et al., J. Magn. Magn. Mater. 548, 168875 (2022)
- F. Bruckner et al., Scientific Reports 11, 9202 (2021)
Confidence: VERIFIED
Verification status: ✅ VERIFIED
- Source code: ACCESSIBLE (GitHub)
- Documentation: ACCESSIBLE
- Published methodology: J. Magn. Magn. Mater.
- Active development: Ongoing
- Specialized strength: True periodic boundary conditions, FEM+FD solvers, spin-torque support