14
$\begingroup$

I am doing some research and trying to make sure the numbers I am reporting are accurate.

I found that the paper by Sylvetsky et. al. says:[1]

We were, however, able to complete a CCSDT(Q)/cc-pVTZ calculation on benzene, which entailed 2.2 trillion (Q) contributions and 3.1 billion iterative CCSDT amplitudes

But there is no indication that this was the largest number of amplitudes (3.1 billion), or largest number of perturbative contributions (2.2 trillion). Furthermore, the paper was done in 2016 so surely larger might have been possible since then.

Does anyone know what the largest couples cluster calculation done as of March 2019 was? By "largest" I mean containing the largest number of cluster amplitudes (of any excitation level).

  1. Nitai Sylvetsky; Kirk A. Peterson; Amir Karton; Jan M.L. Martin; Toward a W4-F12 approach: Can explicitly correlated and orbital-based ab initio CCSD(T) limits be reconciled? J. Chem. Phys. 2016, 144, 214101. DOI: 10.1063/1.4952410 arXiv:1605.03398 [physics.chem-ph]
Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ @CARNEGIE I've voted to reopen and I do think it is an interesting question (+1). My point was just to clarify what you wanted out of an answer. For example, I wanted to make sure it wouldn't make a difference if the largest was (nonreal examples to follow) ccsdtqp of water or ccsd of a protein. $\endgroup$
    – Tyberius ♦
    Commented Mar 19, 2019 at 23:50
  • 1
    $\begingroup$ @Tyberius, the CCSD for a protein (done by Frank Neese) was with DLPNO, which vastly reduces the size of the virtual space. In the end, does it still have more amplitudes than the benzene calculation which involved 3.1 billion amplitudes? If not, then the coupled cluster calculation was not bigger (although the size of the system they worked on, is certainly bigger). As for CCSDTQP on H2O, it seems only possible in a double-zeta basis set, which means it probably has fewer than a billion amplitudes involved! $\endgroup$
    – CARNEGIE
    Commented Mar 19, 2019 at 23:55
  • 3
    $\begingroup$ You might get more answers over at Materials Modeling, it’s new and active! $\endgroup$
    – kskinnerx16
    Commented May 24, 2020 at 21:03

0

Reset to default