QUICK

QUICK (QUantum Interaction Computational Kernel) is a GPU-enabled ab initio and density functional theory software package developed by the Götz and Merz labs. It leverages CUDA for GPU acceleration, providing significant speedups for el…

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

Overview

QUICK (QUantum Interaction Computational Kernel) is a GPU-enabled ab initio and density functional theory software package developed by the Götz and Merz labs. It leverages CUDA for GPU acceleration, providing significant speedups for electronic structure calculations on modern HPC hardware.

Reference Papers (1)

Full Documentation

Official Resources

  • Homepage: https://quick-docs.readthedocs.io/
  • Documentation: https://quick-docs.readthedocs.io/en/latest/
  • Source Repository: https://github.com/merzlab/QUICK
  • License: Mozilla Public License 2.0

Overview

QUICK (QUantum Interaction Computational Kernel) is a GPU-enabled ab initio and density functional theory software package developed by the Götz and Merz labs. It leverages CUDA for GPU acceleration, providing significant speedups for electronic structure calculations on modern HPC hardware.

Scientific domain: Molecular quantum chemistry, GPU-accelerated calculations
Target user community: Researchers needing fast HF/DFT calculations for medium-to-large molecules on GPU clusters

Theoretical Methods

  • Hartree-Fock (RHF, UHF)
  • Density Functional Theory (pure and hybrid)
  • LDA, GGA, meta-GGA, hybrid functionals
  • Dispersion corrections (DFT-D3)
  • COSMO implicit solvation
  • Geometry optimization
  • Molecular dynamics
  • QM/MM (Amber interface)

Capabilities (CRITICAL)

  • GPU-accelerated two-electron integrals
  • CUDA implementation of ERI evaluation
  • Hybrid CPU/GPU execution
  • Large-scale molecular calculations
  • Direct SCF algorithm
  • Cutoffs for linear scaling
  • Geometry optimization
  • Amber/QUICK QM/MM interface
  • MPI multi-GPU parallelism
  • Single and double precision

Key Strengths

GPU Acceleration:

  • CUDA-optimized ERI kernels
  • 10-100x speedups over CPU
  • Multi-GPU scaling
  • Modern NVIDIA GPU support
  • Mixed precision capabilities

HF/DFT Performance:

  • Efficient integral evaluation
  • Schwarz screening
  • Grid-based DFT
  • Hybrid functional support
  • Large molecule capability

Amber Integration:

  • Native QM/MM interface
  • Biochemical applications
  • Enzyme active sites
  • Drug-receptor interactions

Scalability:

  • Multi-GPU support
  • MPI parallelism
  • Linear-scaling techniques
  • HPC cluster deployment

Inputs & Outputs

  • Input formats:

    • QUICK input files
    • Coordinate specifications
    • Basis set files
    • Amber interface files
  • Output data types:

    • Energies and gradients
    • Molecular orbitals
    • Population analysis
    • Optimized structures
    • QM/MM energies

Interfaces & Ecosystem

  • MD integration: Amber molecular dynamics
  • Basis sets: Standard Gaussian basis sets
  • Libraries: CUDA, cuBLAS, MPI
  • Visualization: Standard formats

Advanced Features

GPU Optimization:

  • Custom CUDA kernels
  • Batched integral evaluation
  • Memory management
  • Thread coarsening
  • Shared memory usage

QM/MM Calculations:

  • Electrostatic embedding
  • Mechanical embedding
  • Link atom scheme
  • Amber force fields

Linear Scaling:

  • Schwarz inequality screening
  • Distance cutoffs
  • Sparse matrix techniques
  • Large system efficiency

Performance Characteristics

  • Speed: 10-100x GPU acceleration
  • Accuracy: Standard HF/DFT accuracy
  • System size: Hundreds to thousands of atoms
  • Memory: GPU memory dependent
  • Parallelization: Multi-GPU MPI scaling

Computational Cost

  • HF: Highly efficient on GPU
  • DFT: Fast with grid optimization
  • Hybrid DFT: GPU-accelerated exchange
  • Gradient: Efficient for optimization
  • Typical: Orders of magnitude faster than CPU

Limitations & Known Constraints

  • GPU requirement: NVIDIA CUDA GPUs needed
  • Method scope: HF/DFT focus (no post-HF)
  • Basis sets: Some limitations
  • Memory: Limited by GPU memory
  • Features: Fewer than general-purpose codes
  • Relativistic: Not supported

Comparison with Other Codes

  • vs TeraChem: Both GPU-accelerated; different focus
  • vs Gaussian: QUICK much faster on GPU, fewer methods
  • vs GAMESS: QUICK GPU-native, GAMESS more features
  • vs PySCF-GPU4: Different implementation approaches
  • Unique strength: CUDA-optimized ERI, Amber QM/MM integration

Application Areas

Biochemistry:

  • Enzyme mechanisms
  • Drug binding energies
  • Protein-ligand interactions
  • Large biomolecular QM/MM

Large Molecules:

  • Organic semiconductors
  • Polymers
  • Supramolecular systems
  • Nanoscale materials

High-Throughput:

  • Virtual screening
  • Property prediction
  • Database generation
  • ML training data

Best Practices

GPU Utilization:

  • Use modern NVIDIA GPUs
  • Appropriate batch sizes
  • Memory management
  • Multi-GPU for large systems

Calculation Setup:

  • Appropriate cutoffs
  • Grid density for DFT
  • Convergence criteria
  • Basis set selection

Community and Support

  • Open-source MPL 2.0
  • Active GitHub development
  • Merz and Götz research groups
  • Documentation and tutorials
  • Academic publications

Verification & Sources

Primary sources:

  1. GitHub repository: https://github.com/merzlab/QUICK
  2. Documentation: https://quick-docs.readthedocs.io/
  3. Manathunga et al., J. Chem. Theory Comput. (2020) - QUICK paper
  4. Active development at Michigan State and SDSC

Confidence: VERIFIED

  • Source code: OPEN (GitHub, MPL 2.0)
  • Documentation: Comprehensive
  • Active development: Yes
  • Academic citations: Growing

Related Tools in 1.4 Quantum Chemistry Suites