search

bsonghao / Thermal-Field

thermal field coupled cluster method: A new method for finite temperature electronic structure
1 stars 0 forks source link

Debug thermal field code #3

Open bsonghao opened 2 years ago

bsonghao commented 2 years ago

Map T_1 amplitude from 1-RDM of HF calculation and plug the mapped T_1 amplitude in to R_1 residue (T_2 = 0) and verify that R_1(T_1(HF)) = 0

bsonghao commented 2 years ago

a1b42ab4a1a5f1501ef3f2633e6a64efee3e8ee6 construct Fock matrix based on thermal field normal ordering and conduct the above check. The single residue is correct but double residue is non-zero and the correlation energy + constant term is not equal to HF energy. There is definitely something wrong the require further scrutiny.

bsonghao commented 2 years ago

Understand the source of the bug. It is in ERI, in some cases, they use spin adapted ERI (w), in other cases, uses ERI without spin adaptations (v)

bsonghao commented 2 years ago

348de56d3df27351696bb2fe5e1f016da5bad2c2 resolve the issue stated above but the HF mapped CC residue still not exactly zero and the CC energy does not exactly match HF energy. Require further scrutiny

bsonghao commented 2 years ago

48cd42fa0bb03eb2975b94ec9250c177e0718a71 Actually, the previous though are not correct, the CC residue equation from PySCF source code has been spin adapted and there's no need to form spin-adapted two electron integrals. Using HF mapped t_1 amplitude, the single residue is close to zero but the energy still does not match HF energy.

bsonghao commented 2 years ago

3d235d8c82ecdf14190e12cafedd6dece9fb5ac8 The CC energy matches HF energy after add the nuclear repulsion energy