PyFLOSIC - Python based Fermi-L\"owdin orbital self-interaction correction.

The Fermi-Lowdin orbital self-interaction correction (FLO-SIC) method addresses the self-interaction error within density functional theory (DFT). We present PyFLOSIC, an open-source python implementation of FLO-SIC that is based on the highly modular and modern PySCF electronic structure code. This implementation offers the possibility to use FLO-SIC with any kind of basis set (e.g from the EMSL data base), different numerical meshes (e.g. Lebedev-Laikov grids) and any exchange-correlation functional available in the libxc library. The FLO-SIC method relies on so-called Fermi-orbital descriptors (FODs). We provide a generator for starting values of these FODs which is based on centers of mass of localized orbitals (PyCOM). For the optimization of the FODs, a calculator and an optimizer class for the atomic simulation environment (ase) are included. We present several proof-of-principles calculations and extensive benchmarks against the reference FLO-SIC implementation. The thermochemical performance and ionization potentials of FLO-SIC for the G2-1 benchmark set are analyzed using the local spin density approximation (LSDA), the Perdew-Burke-Ernzerhof (PBE) and the strongly-constrained and appropriately-normed (SCAN) functional.