Dear Users and Developers,
I have a few queries,

1. I have performed localization of Canonical HF orbitals. Next, I want to run Localized CCSD(T). Can it be done with PySCF?
2. With these localized orbitals, I have run the CCSD(T) calculations as shown below example,

    mf = scf.RHF(mol)
    mf.kernel()
    hf_nrg = mf.e_tot  
    ccsd = cc.CCSD(mf)
    ccsd.mo_coeff = **matrix**
    ccsd.kernel()[0]
    ccsd_corr_nrg = ccsd.e_corr
    et = ccsd.ccsd_t()

Instead of using canonical orbitals, I feed the localized orbitals by this matrix
I am getting CCSD energies accurate to 0.03 kJ/mol, but the (T) correction energies deviate from 15-30 kJ/mol (increasing with larger systems). Would anyone be able to give me a suggestion?
3. If my way is correct, then the 1PDM at the CCSD(T) level should be sparser, which is different in my work. Does it mean I have to use a cut-off for not processing the distant orbitals in CCSD(T) calculations?

I have seen this issue, which is different from mine, #1782
Thanking You,
Prasanta