libint2::OneBodyEngine d function error about libint

[evaleev]

There is no bug.

The difference with Q-Chem is due to the difference in normalization. Libint adheres to the CCA standard for integrals (see doi 10.1002/jcc.20815), whereby only cartesian functions with all quanta along the same axis are unit-normalized (e.g. d_xx, f_yyy, etc.), whereas others are not (e.g. d_xy, f_xyz, etc.). This is convenient.

The energy "convergence" is a non-issue ... the HF energy expression of this admittedly toy code uses the fock matrix with density from iteration i contracted with the density from iteration i+1. This is to make the code structure as simple as possible (the hartree-fock code as well as the libint C++11 interface was developed first to teach an NSF summer school ... hence the simplicity).