DFT++

Official Resources

• Source Repository: https://github.com/ifilot/pypwdft
• License: Open Source (GPL-3.0)

Overview

pypwdft is a pure Python implementation of a Plane-Wave Density Functional Theory (PW-DFT) code. It provides a complete, self-contained educational platform for understanding how plane-wave basis sets, pseudopotentials, and reciprocal space operations work in standard codes like VASP or Quantum ESPRESSO, but in a readable, non-HPC Python environment.

Scientific domain: Educational theory, Plane-Wave DFT Target user community: Students, developers learning PW algorithms

Theoretical Methods

• Kohn-Sham DFT
• Plane-Wave basis set
• Periodic Boundary Conditions (PBC)
• Pseudopotentials (Local and Non-local parts)
• Ewald Summation for electrostatics
• Conjugate Gradient minimization for electrons
• LDA/GGA Functionals

Capabilities

• Periodic crystal calculations (e.g., Silicon, Aluminum)
• Electronic band structure generation
• Total energy minimization
• Charge density calculation
• Reciprocal space handling (k-points)

Key Strengths

Pedagogical Value:

• Demonstrates FFT grid operations explicitely
• Shows construction of the Hamiltonian in G-space
• Clear implementation of Ewald summation
• Handles pseudopotential integration grids

Code Readability:

• Python classes map clearly to physical concepts (Cell, Atom, Wavefunction)
• Avoids complex parallelization logic

Inputs & Outputs

• Inputs:
  • Lattice parameters
  • Atom positions
  • Plane-wave cutoff energy
  • Pseudopotential files
• Outputs:
  • Total energy
  • Forces (if implemented)
  • Stress (if implemented)
  • Band structure data

Interfaces & Ecosystem

• Python: Core implementation.
• NumPy/SciPy: Heavy usage for FFTs and linear algebra.

Advanced Features

• CG Optimization: Uses conjugate gradient for wavefunction optimization (common in large codes).
• Pseudopotential parsing: Simple readers for format handling.

Performance Characteristics

• Slow: Pure Python + NumPy FFTs are significantly slower than C/Fortran.
• Memory: Stores dense grids, limited to small unit cells.

Computational Cost

• Moderate: Can handle small unit cells (2-8 atoms) on a desktop, but scales poorly compared to compiled codes.

Limitations & Known Constraints

• Performance: Not a production code.
• Features: Lacks advanced features like hybrid functionals, spin-orbit coupling, or complex constraints.
• Pseudopotentials: Limited library support compared to Quantum ESPRESSO.

Comparison with Other Codes

• vs SimpleDFT/eminus: Similar scope, but pypwdft often focuses more explicitly on the "traditional" VASP-like algorithms (CG minimization).
• vs Quantum ESPRESSO: pypwdft is the "textbook model" of what QE does at high scale.
• Unique strength: Accessibility for understanding reciprocal space logic without Fortran.

Application Areas

• Teaching: Solid State Physics courses.
• Algorithm Development: Testing new minimization schemes or functional forms in Python.

Best Practices

• Small Cutoffs: Keep kinetic energy cutoffs low for testing.
• Simple Systems: Start with Silicon or Aluminum bulk.
• Profile: Use Python profilers to seeing where the time goes (usually FFTs).

Community and Support

• GitHub: Repository by active maintainer (Ivo Filot).

Verification & Sources

Primary sources:

1. GitHub Repository: https://github.com/ifilot/pypwdft

Verification status: ✅ VERIFIED

• Source code: OPEN (GPL-3.0)
• Functionality: Functional PW-DFT prototype.