BerkeleyGW

Official Resources

• Homepage: https://berkeleygw.org/
• Documentation: https://berkeleygw.org/documentation/
• Source Repository: Available to users (registration required)
• License: BSD-like license (free for academic use)

Overview

BerkeleyGW is a massively parallel code for computing the quasiparticle and optical properties of materials using many-body perturbation theory within the GW approximation and the Bethe-Salpeter equation (BSE). It is designed for large-scale calculations on leadership-class supercomputers and provides highly accurate band structures, band gaps, and optical spectra beyond DFT.

Scientific domain: Many-body perturbation theory, GW approximation, optical properties, excited states
Target user community: Researchers studying electronic excitations, band structures, and optical properties of materials

Theoretical Methods

• GW approximation (G₀W₀, eigenvalue self-consistent GW)
• Generalized Plasmon Pole (GPP) model
• Contour-deformation method (full frequency)
• Bethe-Salpeter equation (BSE) for optical spectra
• Time-dependent density functional theory (TDDFT)
• Electron-hole interaction (excitonic effects)
• Static and dynamic screening
• Coulomb-hole screened-exchange (COHSEX)
• Scissors operator corrections
• Contour-deformation method
• Bethe-Salpeter Equation (BSE)
• Static and dynamical screening
• Vertex corrections (selected cases)
• Hybrid functional starting points

Capabilities (CRITICAL)

• Quasiparticle band structures via GW
• Quasiparticle band gaps and corrections to DFT
• Optical absorption spectra including excitonic effects (BSE)
• Electron-hole interaction analysis
• Exciton wavefunctions and binding energies
• Static and frequency-dependent dielectric functions
• Interface to multiple DFT codes (mean-field starting point)
• Massively parallel execution (MPI + OpenMP)
• GPU acceleration (experimental)
• Real-space grids and subspace methods
• Static subspace approximation for reduced cost
• Spin-orbit coupling support
• Non-collinear magnetism support
• Interpolation schemes for band structures

Sources: Official BerkeleyGW website, documentation, cited in 7/7 source lists

Inputs & Outputs

• Input formats:

  • WFN files (wavefunctions from DFT codes)
  • RHO files (charge density)
  • VXC files (exchange-correlation potential)
  • Input files (epsilon.inp, sigma.inp, kernel.inp, absorption.inp)
  • k-point and q-point grids

• Output data types:

  • eps0mat, epsmat (dielectric matrices)
  • eqp.dat (quasiparticle energies)
  • sigma.log (self-energy calculations)
  • absorption_eh.dat (optical absorption)
  • eigenvectors (exciton wavefunctions)

Interfaces & Ecosystem

• DFT code interfaces (verified):

  • Quantum ESPRESSO - primary interface via pw2bgw.x
  • PARATEC - native support
  • PARSEC - supported
  • SIESTA - interface available
  • Octopus - interface available
  • ABINIT - can be interfaced

• Pre-processing tools:

  • kgrid.x - k-point grid generation
  • wfn2hdf.x - wavefunction format conversion
  • mf_convert_wrapper.sh - mean-field conversion scripts

• Post-processing tools:

  • plotxct.x - exciton wavefunction plotting
  • absorption.x - optical spectra
  • inteqp.x - band structure interpolation
  • offdiag.x - off-diagonal matrix elements

• Analysis utilities:

  • Python scripts for data analysis
  • Plotting utilities
  • Convergence checking tools

Limitations & Known Constraints

• Computational cost: Extremely expensive; GW scales as O(N⁴), BSE even worse
• Memory requirements: Very large; dielectric matrices can be tens to hundreds of GB
• Convergence: Many parameters to converge (cutoffs, bands, k-points, q-points)
• DFT dependency: Quality depends on mean-field starting point
• Registration required: Free but requires registration for download
• Learning curve: Steep; requires understanding of GW theory and convergence procedures
• System size: Practical limit ~100-200 atoms for GW; smaller for BSE
• Static approximation: Plasmon-pole models introduce approximations
• Parallelization: Requires careful setup for optimal performance
• GPU support: Experimental; not all features GPU-accelerated

Verification & Sources

Primary sources:

1. Official website: https://berkeleygw.org/
2. Documentation: https://berkeleygw.org/documentation/
3. J. Deslippe et al., Comput. Phys. Commun. 183, 1269 (2012) - BerkeleyGW code paper
4. M. S. Hybertsen and S. G. Louie, Phys. Rev. B 34, 5390 (1986) - GW method
5. M. Rohlfing and S. G. Louie, Phys. Rev. B 62, 4927 (2000) - BSE method

Secondary sources:

1. BerkeleyGW tutorials and workshops
2. Quantum ESPRESSO interface documentation
3. Published benchmarks and applications
4. Confirmed in 7/7 source lists (claude, g, gr, k, m, q, z)

Confidence: CONFIRMED - Appears in all 7 independent source lists

Verification status: ✅ VERIFIED

• Official homepage: ACCESSIBLE
• Documentation: COMPREHENSIVE and ACCESSIBLE
• Source code: OPEN (requires registration)
• Community support: Active (mailing list, workshops)
• Academic citations: >600 (main code paper)
• HPC optimization: Extensively benchmarked on supercomputers