TBPLaS: A tight-binding package for large-scale simulation

TBPLaS is an open-source software package for the accurate simulation of physical systems with arbitrary geometry and dimensionality utilizing tight-binding (TB) theory. It has intuitive object-oriented Python application interface (API) Cython/Fortran extensions performance-critical parts, ensuring both flexibility efficiency. Under hood, numerical calculations are mainly performed by exact diagonalization propagation method (TBPM) without diagonalization. Especially, TBPM based on solution time-dependent Schrödinger equation, achieving linear scaling system size in memory CPU costs. Consequently, provides a numerically cheap approach to calculate electronic, optical, plasmon transport properties large models billions atomic orbitals. Current capabilities include band structure, density states, local quasi-eigenstates, optical conductivity, electrical Hall polarization function, dielectric dispersion, carrier mobility velocity, localization length free path, Z2 topological invariant, wave-packet propagation, etc. All can be obtained only few lines code. Other algorithms involving Hamiltonians implemented easily due extensible modular nature In this paper, we discuss theoretical framework, implementation details common workflow TBPLaS, give demonstrations its applications. Program Title: CPC Library link program files: https://doi.org/10.17632/5cjmzj9jwm.1 Code Ocean capsule: https://codeocean.com/capsule/8296734 Licensing provisions: BSD 3-clause Programming language: Python, Cython, Fortran Nature problem: The TB powerful tool investigating electronic structures many other condensed matter physics. However, costs time using standard diagonalization-based techniques scale as O(N2) O(N3) model N, severely limiting their applications complex systems. development methods solve problems any diagonalization, well these packages, urgently needed promote capability atoms. Solution method: We proposed it package. TBPM, calculated solving equation which ensures size. Models atoms (orbitals) handled, available from evolution correlation functions. features API Moreover, implements hybrid MPI+OpenMP parallelism, exploit modern hardware high-performance computers.