walter-feng / hartree-fock-in-cpp Goto Github PK

I notice in the code that t_squared=u/(1+u), however, in (27) of the 1982 paper, it is t_squared=u/(rho + u). But the other parts, i.e. the recursion relations as I verified are aligned with the paper. Am I missing anything here?

While the implementation of overlap and nuclear_attration integrals are following the recursive relations in the original 1985 paper, the implementation of kinetic and electron_repulsion integrals seem not to be exactly aligned. i.e. instead of computing recursively, they are both expanded into summing over all the pairs.

Does this implementation lead to fewer FLOPS compared to the original RR? Is there any reference that derives this alternative form?

BTW, it is mentioned here that the implementation has some error, has this been fixed?