Home › Scientific Tools DB › 1. GROUND-STATE DFT › 1.6 Specialized

cmpy

AMDKIIT (`ab initio` Molecular Dynamics at KIIT/IIT) is a specialized Plane-Wave DFT software package developed to perform efficient molecular dynamics simulations. Created under the Indian National Supercomputing Mission (NSM), it is de…

1. GROUND-STATE DFT 1.6 Specialized VERIFIED
Back to Mind Map Official Website

Overview

AMDKIIT (`ab initio` Molecular Dynamics at KIIT/IIT) is a specialized Plane-Wave DFT software package developed to perform efficient molecular dynamics simulations. Created under the Indian National Supercomputing Mission (NSM), it is designed to run efficiently on high-performance computing clusters, including those with GPU acceleration. It bridges the gap between general-purpose DFT codes and specialized MD engines, focusing on the high-throughput generation of AIMD trajectories.

Reference Papers

Reference papers are not yet linked for this code.

Full Documentation

Official Resources

  • Homepage: https://github.com/AMDKIIT/amdkiit
  • Source Repository: https://github.com/AMDKIIT/amdkiit
  • License: GNU General Public License v3.0

Overview

AMDKIIT (ab initio Molecular Dynamics at KIIT/IIT) is a specialized Plane-Wave DFT software package developed to perform efficient molecular dynamics simulations. Created under the Indian National Supercomputing Mission (NSM), it is designed to run efficiently on high-performance computing clusters, including those with GPU acceleration. It bridges the gap between general-purpose DFT codes and specialized MD engines, focusing on the high-throughput generation of AIMD trajectories.

Scientific domain: Plane-Wave DFT, Ab Initio Molecular Dynamics Target user community: HPC users, Researchers in materials chemistry and dynamical processes

Theoretical Methods

  • Kohn-Sham DFT: Standard formulations (LDA, PBE).
  • Plane-Wave Basis: Periodic boundary conditions.
  • Pseudopotentials: Norm-Conserving and Ultrasoft (UPF format support).
  • Car-Parrinello & Born-Oppenheimer MD: Propagation schemes.
  • Thermostats: Nose-Hoover, Berendsen for NVT ensembles.

Capabilities (CRITICAL)

  • Electronic Minimization: Self-consistent field using iterative diagonalization (Davidson/Conjugate Gradient).
  • Forces & Stress: Analytic calculation of Hellmann-Feynman forces and stress tensors.
  • Molecular Dynamics: Long-time scale NVE and NVT simulations.
  • GPU Acceleration: Offloading of heavy FFT and BLAS operations to GPUs (CUDA).
  • Parallelism: Hybrid MPI/OpenMP parallelization.

Key Strengths

GPU Optimization:

  • Built from the ground up to leverage modern heterogeneous architectures (CPU+GPU).
  • improved performance-per-watt for long MD runs.

Local Development:

  • Major indigenous code development project from India (IIT Kanpur).
  • Open architecture allowing for academic contributions.

Inputs & Outputs

  • Inputs:
    • input.in: Main control file (cutoffs, convergence, MD steps).
    • structure.xyz: Initial coordinates.
    • Pseudopotentials (.upf).
  • Outputs:
    • trajectory.xyz: MD Steps.
    • energy.dat: Thermodynamic logs.
    • forces.dat: Atomic forces.
    • restart.bin: Binary checkpoint capability.

Interfaces & Ecosystem

  • File Formats: Compatible with standard UPF pseudopotentials (Quantum ESPRESSO ecosystem).
  • Visualisation: Trajectories readable by VMD, Ovito.

Advanced Features

  • Berry Phase: (Developmental) Polarization calculations.
  • Metadynamics: (Planned) Enhanced sampling integration.

Performance Characteristics

  • Speed: Competitive with major codes for standard MD benchmarks on GPU nodes.
  • Scaling: Good strong scaling on cluster partitions.

Computational Cost

  • High Efficiency: Design goal is to reduce wall-time for 10-100 ps simulations.

Limitations & Known Constraints

  • Feature Set: Less feature-rich than VASP/QE (e.g., no hybrid functionals, no GW yet).
  • Documentation: Documentation is evolving; usage requires familiarity with standard PW-DFT inputs.
  • Maturity: Newer code compared to established giants; expect rapid changes.

Comparison with Other Codes

  • vs Quantum ESPRESSO: Both use UPF/Plane-Waves; AMDKIIT is simpler and optimized specifically for MD on specific hardware.
  • vs CPMD: CPMD is the ancestor of AIMD; AMDKIIT is a modern C++/CUDA implementation.
  • vs VASP: VASP is the industry standard; AMDKIIT offers an open-source, GPU-ready alternative for basic MD tasks.
  • Unique strength: GPU-native design philosophy for AIMD.

Application Areas

  • Liquids & Solvation: Structure of water, ions in solution.
  • Diffusion: Ion migration in battery materials.
  • Surface Dynamics: Adsorption and reconstruction processes.

Best Practices

  • GPUs: Running on CPU-only nodes misses the main optimization point.
  • Potentials: Use standard GBRV or SSSP pseudopotentials (verify compatibility).
  • Time Step: Use appropriate time steps (0.5 - 1.0 fs) for stability.

Community and Support

  • Source: Developed at IIT Kanpur.
  • GitHub: Issues tracking available.

Verification & Sources

Primary sources:

  1. Repository: https://github.com/AMDKIIT/amdkiit
  2. NSM Project documentation.

Verification status: ✅ VERIFIED

  • Source code: OPEN (GPLv3)
  • Origin: Verified academic project.

Related Tools in 1.6 Specialized

FlapwMBPT cmpy