libNEGF

Official Resources

• Repository: https://github.com/libnegf/libnegf
• License: LGPL-3.0

Overview

libNEGF is a high-performance Fortran 2008 library capable of solving the Non-Equilibrium Green's Function (NEGF) equations for quantum transport. It is designed as a modular solver (middleware) that can be embedded into various electronic structure codes (like OMEN, CP2K, or custom drivers). libNEGF provides efficient algorithms for computing the Green's functions, self-energies, and transmission matrices for both orthogonal and non-orthogonal basis sets.

Scientific domain: Quantum Transport, Electronic Structure, High-Performance Computing Target user community: Developers of transport codes and researchers needing a robust NEGF kernel

Theoretical Methods

• NEGF Formalism: Stationary (steady-state) transport calculations.
• Recursive Green's Functions (RGF): specialized algorithms (including block-tridiagonal inversion) for wire-like systems to achieve linear scaling $O(N)$.
• Sancho-Rubio Method: Iterative calculation of surface Green's functions for semi-infinite leads.
• Scattering: Frameworks for including electron-phonon and electron-photon scattering self-energies (approximated).

Capabilities

• Observables:
  • Transmission Probability $T(E)$.
  • Density of States (DOS).
  • Current density.
• System Types:
  • Two-terminal devices (Source-Channel-Drain).
  • Nanowires, FinFETs, 2D materials (within a transport setup).
• Basis Agnostic: Works with block-sparse matrices provided by the caller (DFT or Tight-Binding).

Key Strengths

• Performance: Heavily optimized for modern supercomputers, utilizing MPI and OpenMP hybrid parallelism.
• Scalability: Can handle systems with >100,000 atoms when using RGF.
• Modularity: Clean API separates the physics of the Hamiltonian (handled by the caller) from the matrix algebra of transport (handled by libNEGF).

Inputs & Outputs

• Inputs:
  • Hamiltonians ($H$) and Overlap ($S$) matrices (CSR format).
  • Energy grid.
  • Contact self-energies (or descriptions to compute them).
• Outputs:
  • Green's function matrices ($G^R, G^<, G^>$).
  • Transmission scalars.
  • Parallel distributed data structures.

Interfaces & Ecosystem

• Integrated into:
  • OMEN: A state-of-the-art atomistic quantum transport solver.
  • CP2K: Used for transport calculations in the CP2K suite.
• Usage: Typically called from Fortran or C/C++ codes.

Performance Characteristics

• Speed: Optimized linear algebra using BLAS/LAPACK and ScaLAPACK.
• Parallelism: Multiple levels of parallelism (Energy points, Spatial domain decomposition, Linear algebra blocks).

Comparison with Other Codes

• vs. TB_Sim (TBT): Similar low-level solver role, but libNEGF is open-source and modern Fortran.
• vs. Kwant: Kwant is a user-facing Python package for models; libNEGF is a backend library for massive atomistic simulations.

Community and Support

• Development: Developed at ETH Zurich (Integrated Systems Laboratory) and partners.
• Source: GitHub.

Verification & Sources

• Repository: https://github.com/libnegf/libnegf
• Primary Publication: Check OMEN/libNEGF citations from ETH researchers (Mathieu Luisier et al.).
• Verification status: ✅ VERIFIED
  • Active and foundational library for HPC transport.