30.2.3 The fixed amplitude ansatz (FIX)

The wave function has the same form as for the diagonal ansatz, but the amplitudes are determined from the cusp conditions and fixed, i.e., $T^{ij,1}_{ij}=1/2$ for singlet pairs ($p=1$) and $T^{ij,-1}_{ij}=1/4$ for triplet pairs ($p=-1$). This restores the orbital invariance, and usually the results are almost as accurate as with the diagonal ansatz. This method is size consistent for any choice of orbitals. Since the (T) correction in CCSD(T)-F12 calculations requires the use of canonical orbitals, the FIX approximation is always used in CCSD-F12 and CCSD(T)-F12 calculations.