Analytical nuclear derivatives for the parametric two-electron reduced density matrix method

作者：Valentine Andrew J S; Mazziotti David A^*

来源：Chemical Physics Letters, 2017, 685: 300-304.

DOI：10.1016/j.cplett.2017.07.064

Efficient and accurate nuclear gradients are essential to performing molecular optimizations. Here for the first time we present analytical nuclear gradients for the parametric two-electron reduced-density-matrix method (p2-RDM), which uses the 2-RDM as the primary variable in calculations in lieu of the many-electron wavefunction. While numerical gradients require six energy evaluations for each atom, analytical gradients require only a single calculation for each geometry sampled. We present benchmark p2-RDM geometry optimizations that show analytical gradients reduce CPU times by as much as 80%, even for small molecules. We also use p2-RDM to evaluate the bond length alternation (BLA), or the difference in length between adjacent single and double bonds, of trans-polyacetylene (PA). We find that the BLA in the extrapolated limit to be 0.080 angstrom, in agreement with experiment and closely mirroring the prediction of the more expensive coupled-cluster with single and double excitations with perturbative triples (CCSD(T)).

• 出版日期2017-10-1