The unrestricted natural orbital-restricted active space method: methodology and implementation

Citation
Pm. Kozlowski et P. Pulay, The unrestricted natural orbital-restricted active space method: methodology and implementation, THEOR CH AC, 100(1-4), 1998, pp. 12-20
Citations number
49
Categorie Soggetti
Physical Chemistry/Chemical Physics
Journal title
THEORETICAL CHEMISTRY ACCOUNTS
ISSN journal
1432881X → ACNP
Volume
100
Issue
1-4
Year of publication
1998
Pages
12 - 20
Database
ISI
SICI code
1432-881X(199812)100:1-4<12:TUNOAS>2.0.ZU;2-H
Abstract
The full configuration interaction method in the space of fractionally occu pied unrestricted natural orbitals (UNO-CAS method) is extended to excited states as well, as to strongly correlated and reactive systems with large a ctive spaces. This is accomplished by using restricted active space (RAS) w ave functions introduced by Olsen et al. [(1988) J Chem Phys 89: 2185] and using the UNOs without the expensive orbital optimization step. In RAS, the space of active orbitals is subdivided into three groups: a group with ess entially doubly occupied orbitals (RAS1) the usual CAS space (RAS2), and a space with weakly occupied active orbitals (RAS3). We select these spaces o n the basis of the occupation numbers of the UNOs. All possible electron di stributions are allowed in the usual CAS space, but the number of vacancies is limited in RAS1 and the number of electrons is limited in RAS3. We disc uss an efficient algorithm for generating a RAS wave function. This is base d on the Handy-Knowles determinantal expansion with an addressing scheme ad opted for the restricted expansion. Results for both ground and excited sta tes of azulene and free base porphyrin are presented.