Caculating SAB overlap of two Kohn-Sham determinants

  • Context: Graduate 
  • Thread starter Thread starter limorsj
  • Start date Start date
  • Tags Tags
    Determinants Overlap
Click For Summary
SUMMARY

The discussion focuses on calculating the overlap (SAB) of two Kohn-Sham determinants, represented by matrices psiA and psiB containing molecular orbital coefficients. The required value for the overlap of identical matrices (SAA and SBB) is 1, indicating perfect overlap. The most efficient algorithm for computing matrix elements between non-orthogonal determinants is referenced from Utsuno et al. (2012). The proposed method involves computing the overlap matrix of occupied orbitals, performing singular value decomposition (SVD), and obtaining the overlap as the product of singular values.

PREREQUISITES
  • Understanding of Kohn-Sham determinants and molecular orbital theory
  • Familiarity with matrix operations, specifically overlap matrices
  • Knowledge of singular value decomposition (SVD)
  • Experience with computational chemistry algorithms
NEXT STEPS
  • Study the algorithm by Utsuno, Shimizu, Otsuka, and Abe for efficient computation of Hamiltonian matrix elements
  • Learn about the evaluation of non-orthogonal matrix elements as described by Verbeek and van Lenthe
  • Implement the calculation of overlap matrices in a computational chemistry software environment
  • Explore advanced techniques in quantum chemistry for handling non-orthogonal determinants
USEFUL FOR

Researchers and practitioners in computational chemistry, particularly those involved in quantum mechanical calculations and the analysis of molecular orbitals.

limorsj
Messages
6
Reaction score
0
Hello,

I would like to implement SAB=<psiA|psiB>which is the overlap of two Kohn-Sham determinants (psiA and psiB are two matrices containing each the molecular orbitals coefficients). Can anybody help me with this calculation? For case of SAA and SBB it is required to get the value 1 (perfect overlap for the same matrix...which is actually the probability of finding the electron somewhere),

Thank you!
 
Last edited:
Physics news on Phys.org
As far as I am aware of, the most straightforward algorithm for computing matrix elements between non-orthogonal determinants was given in
Utsuno, Shimizu, Otsuka, Abe - Efficient computation of Hamiltonian matrix elements between non-orthogonal Slater determinants (http://dx.doi.org/10.1016/j.cpc.2012.09.002 ).
Some background on the algorithms is described in
Verbeek, van Lenthe - On the evaluation of non-orthogonal matrix elements (http://dx.doi.org/10.1016/0166-1280(91)90141-6 ).

If you are only interested in overlap matrix elements between the determinants (instead of more complicated matrix elements, e.g., of the Hamiltonian), then my guess is that the above algorithms would turn into something along the following lines:
1. Compute the overlap matrix of the occupied orbitals \mathbf{S}_\mathrm{occ} = \mathbf{C}_\mathrm{A}^T \mathbf{S}_\mathrm{AO} \mathbf{C}_B, where \mathbf{C}_\mathrm{A} is the coefficient matrix of A's occupied orbitals, \mathbf{S}_\mathrm{AO} is the atomic orbital basis overlap matrix (overlap between basis functions in terms of which the occupied orbitals are expanded) and \mathbf{C}_B is the coefficient matrix of B's occupied orbitals.
2. Compute a singular value decomposition (SVD) of the occupied orbital overlap \mathbf{S}_\mathrm{occ}
3. The overlap between the determinants A and B is the product of the singular values from step 2.
 
Last edited by a moderator:

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
13K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 13 ·
Replies
13
Views
4K
  • · Replies 4 ·
Replies
4
Views
4K
  • · Replies 10 ·
Replies
10
Views
5K