메뉴 건너뛰기




Volumn 6, Issue 7, 2010, Pages 1966-1970

A comparison of three variants of the generalized Davidson algorithm for the partial diagonalization of large non-Hermitian Matrices

Author keywords

[No Author keywords available]

Indexed keywords


EID: 79251634491     PISSN: 15499618     EISSN: 15499626     Source Type: Journal    
DOI: 10.1021/ct100111w     Document Type: Article
Times cited : (20)

References (12)
  • 6
    • 79251604469 scopus 로고    scopus 로고
    • Some elements of the A matrix are never explicitly calculated because the terms in the matrix-vector contractions can be properly organized in order to maintain O (N6) scaling, where N is the number of basis functions, and storage of at most four indexes quantities.1
    • Some elements of the A matrix are never explicitly calculated because the terms in the matrix-vector contractions can be properly organized in order to maintain O (N6) scaling, where N is the number of basis functions, and storage of at most four indexes quantities.1.
  • 7
    • 79251641151 scopus 로고    scopus 로고
    • Here we assume that the target roots of the similarity transformed Hamiltonian are real. Although a generic non- Hermitian matrix may have complex eigenvalues, the above assumption is justified in the context of the EOM-CCSD method because the eigenvalues are excitation energies, which are real quantities. If complex excitation energies are found among the target roots, this is an indication that there is a problem with the description of the wave function (for example the reference function may not be stable), and the rate of convergence of the diagonalization algorithm is, therefore, not relevant
    • Here we assume that the target roots of the similarity transformed Hamiltonian are real. Although a generic non- Hermitian matrix may have complex eigenvalues, the above assumption is justified in the context of the EOM-CCSD method because the eigenvalues are excitation energies, which are real quantities. If complex excitation energies are found among the target roots, this is an indication that there is a problem with the description of the wave function (for example the reference function may not be stable), and the rate of convergence of the diagonalization algorithm is, therefore, not relevant.


* 이 정보는 Elsevier사의 SCOPUS DB에서 KISTI가 분석하여 추출한 것입니다.