CuGBasis

CuGBasis is a free and open-source CUDA/Python library for efficient computation of density-based descriptors from electronic structure calculations. Using GPU acceleration, it achieves remarkable performance gains of up to 100x speedup…

1. GROUND-STATE DFT 1.4 Quantum Chemistry Suites VERIFIED
Back to Mind Map Official Website

Overview

CuGBasis is a free and open-source CUDA/Python library for efficient computation of density-based descriptors from electronic structure calculations. Using GPU acceleration, it achieves remarkable performance gains of up to 100x speedup compared to CPU implementations for evaluating electron densities, gradients, and related properties on 3D grids.

Reference Papers

Reference papers are not yet linked for this code.

Full Documentation

Official Resources

  • Homepage: https://github.com/theochem/cuGBasis
  • Documentation: In repository and J. Chem. Phys. 161, 082501 (2024)
  • Source Repository: https://github.com/theochem/cuGBasis
  • License: GNU General Public License v3.0

Overview

CuGBasis is a free and open-source CUDA/Python library for efficient computation of density-based descriptors from electronic structure calculations. Using GPU acceleration, it achieves remarkable performance gains of up to 100x speedup compared to CPU implementations for evaluating electron densities, gradients, and related properties on 3D grids.

Scientific domain: Post-processing, density-based analysis, GPU-accelerated wavefunction properties
Target user community: Researchers performing real-space wavefunction analysis on moderate to large molecular systems

Theoretical Methods (Analysis)

  • Electron density ρ(r) evaluation
  • Density gradient ∇ρ(r)
  • Laplacian of density ∇²ρ(r)
  • Kinetic energy density τ(r)
  • Electrostatic potential V(r)
  • Molecular orbital evaluation φ(r)
  • Reduced density gradient s(r)
  • Electron localization function (ELF)

Capabilities (CRITICAL)

  • CUDA GPU acceleration
  • Up to 100x speedup over CPU codes
  • Gaussian basis function evaluation on grids
  • Complete density-based descriptor suite
  • Real-space 3D grid calculations
  • Python interface for easy integration
  • Large molecular system support
  • Memory-efficient GPU algorithms
  • Batch processing capabilities
  • Standard wavefunction file input

Key Strengths

GPU Performance:

  • CUDA-optimized kernels
  • Massive parallelism on GPU
  • 100x typical speedups
  • Modern NVIDIA GPU support
  • Memory-efficient design

Descriptor Calculations:

  • Full density ρ(r)
  • Gradient and Laplacian
  • Kinetic energy density
  • ESP mapping on grids
  • All from single run

Integration:

  • Python front-end
  • wfn/fchk/wfx input
  • CUBE file output
  • Post-processing workflow
  • Easy scripting

Efficiency:

  • Large grid support
  • Batch processing
  • Memory management
  • Scalable algorithms

Inputs & Outputs

  • Input formats:

    • Gaussian fchk files
    • wfn format
    • wfx format
    • Molden files
  • Output data types:

    • Grid data arrays
    • CUBE files
    • NumPy arrays
    • Property visualizations

Interfaces & Ecosystem

  • TheoChem tools: Integration with ecosystem
  • NumPy/CuPy: Array backends
  • CUDA: GPU runtime
  • Visualization: VMD, ChimeraX compatible output

Advanced Features

Grid Specification:

  • Uniform and adaptive grids
  • User-defined resolution
  • Molecular boxes
  • Memory optimization

Basis Function Engine:

  • Contracted GTOs
  • Angular momentum handling
  • Normalization
  • Primitive batching

GPU Optimization:

  • Custom CUDA kernels
  • Shared memory usage
  • Coalesced memory access
  • Multiple GPU support potential

Performance Characteristics

  • Speed: 100x vs CPU implementations
  • Accuracy: Machine precision
  • System size: Large molecules feasible
  • Memory: GPU memory dependent
  • Grid size: Millions of points

Computational Cost

  • Small molecules: Milliseconds
  • Large molecules: Seconds
  • Large grids: Minutes
  • CPU comparison: Hours → Seconds
  • Typical: Real-time for moderate systems

Limitations & Known Constraints

  • GPU requirement: NVIDIA CUDA GPUs needed
  • Analysis only: No electronic structure calculation
  • Input quality: Depends on source calculation
  • GPU memory: Limits grid/molecule size
  • Platform: Linux primarily

Comparison with Other Codes

  • vs Multiwfn: CuGBasis GPU, Multiwfn more features
  • vs horton: CuGBasis GPU-accelerated
  • vs Critic2: Different focus (periodic)
  • vs CPU codes: Dramatic speedup
  • Unique strength: GPU acceleration for grid properties

Application Areas

Bonding Analysis:

  • Density topology
  • Bond characterization
  • Electron localization
  • Chemical reactivity

Visualization:

  • Density isosurfaces
  • ESP mapping
  • Orbital visualization
  • Publication figures

Reactivity Analysis:

  • Fukui functions
  • Electrophilicity
  • Nucleophilicity maps
  • Reaction mechanisms

Large Systems:

  • Proteins
  • Nanostructures
  • Materials surfaces
  • Previously infeasible analyses

Best Practices

GPU Setup:

  • Modern NVIDIA GPU
  • Sufficient VRAM
  • CUDA toolkit installed
  • Memory monitoring

Grid Selection:

  • Appropriate resolution
  • Balance accuracy vs memory
  • Benchmark convergence
  • Start coarse, refine

Community and Support

  • Open-source GPL v3
  • TheoChem group (McMaster University)
  • J. Chem. Phys. publication (2024)
  • GitHub issues for support
  • Academic citations

Verification & Sources

Primary sources:

  1. GitHub: https://github.com/theochem/cuGBasis
  2. J. Chem. Phys. 161, 082501 (2024) - Reference paper
  3. TheoChem group documentation
  4. CUDA/GPU computing resources

Confidence: VERIFIED

  • Source code: OPEN (GitHub, GPL v3)
  • Documentation: Paper and repository
  • Active development: Yes (2024 publication)
  • Benchmarked: 100x speedup demonstrated

Related Tools in 1.4 Quantum Chemistry Suites