Two-particle density matrix from full CCSD(T) calculation

How can I calculate the two-particle reduced density matrix from the CCSD(T) calculation?

The two-particle density is computed as part of an analytic gradient calculation. However, there is no simple way to extract it for separate calculations. What are you trying to do?

It is helpful for multiple purposes in my work. Such as these two,

  1. Check for the sparseness of the matrix
  2. Check for 2-PDM->1PDM Muller and BBC1 approximation

It would be helpful to write it in the disk for successive calculations.

Feel free to join the development effort to add this feature. We always welcome new contributors to the code.

I am sad to hear that it has not yet been implemented, but I am excited that it could be implemented now.

BTW, Can I obtain the CCSD(T) amplitudes written in some fch or out file? If yes, can you share an example?

No, to my knowledge, these are not written to a text file or provided to the Python layer.

AJ wrote a PR to expose T amplitudes to Python back in 2019.

You can see a sample input here.

1 Like

Oh, I’d forgotten about that. Thanks, Jonathan!

Thank you @jmisiewicz @crawdad . I have been working on the sample input that you have provided. It is very useful. I wanted to know,

  1. if I can get the l1 and l2 amplitudes as well from ccsd(T) calculation, since I followed a PySCF code that can calculate the 2-particle RDM
def make_rdm2(mycc, t1, t2, l1, l2, eris=None):
    d1 = _gamma1_intermediates(mycc, t1, t2, l1, l2, eris)
    d2 = _gamma2_intermediates(mycc, t1, t2, l1, l2, eris)
    return ccsd_rdm._make_rdm2(mycc, d1, d2, True, True)

which requires the l1 and l2 amplitudes as well. And
2. Where can I get the functions from wavefunction class? I followed the link but there is no
get_amplitude() function.

Thank you very much.

  1. Solutions to the lambda equations are not exported to Python. I’d be happy with exporting them; I just don’t have the time to do that myself.
  2. No, you get the functions from the ccwavefunction subclass.
1 Like

Thank You @jmisiewicz . I think I have to wait for the solution of lambda equations. However, can I ask if I try to do that, can you give me some hints/routes to accomplish this?
Thanks again,

Sure. It’s probably going to be a straightforward mimic of the PR that I posted earlier.

Thank you @jmisiewicz . I will try to find out and share the result here if I succeed. Please share when you have found the answer to this.

Thank you very much,

I’m confused. I’m offering to help you if you want to try this yourself.

There’s no question in this forum topic to which I’m trying to find an answer.

I am sorry for the misunderstanding. I will try to find this and let you know.

Thank you,

I need to pause or quit this quest.