Jean-Pierre Daudey

Convergence of an improved CIPSI algorithm

Abstract The configuration interaction CIPSI algorithm defines three classes of determinants of decreasing importance; the most important ones (∼ 100) are generators, the mean class (∼ 10 3 ) is treated variationally or to the fourth order, while the less important ones (∼ 10 5 ) are treated to the second order only. The accuracy of the result is studied as a function of the borders between the classes in the case of H 2 O (double-zeta basis set), where the exact solution is known, and for the nearly degenerate CN + problem.

Variational calculation of small energy differences. The singlet-triplet gap in [Cu2Cl6]2−

Abstract Based on the theory of effective Hamiltonians, a variational procedure to calculate singlet—triplet energy differences in diradical systems is described, which includes all the differential second-order contributions. When applied to [Cu2Cl6]2−, for which the singlet—triplet gap ranges between −40 and 80 cm−1, depending on the structural parameters, a good agreement is reached. For large systems, a truncation procedure of the MO set based on singlet—triplet gap-dedicated MOs is described, which allows the dimension of the molecular integrals file to be reduced significantly without loss of efficiency.

Charge-transfer correction for improved time-dependent local density approximation excited-state potential energy curves: Analysis within the two-level model with illustration for H2 and LiH

Time-dependent density-functional theory (TDDFT) is an increasingly popular approach for calculating molecular excitation energies. However, the TDDFT lowest triplet excitation energy, ωT, of a closed-shell molecule often falls rapidly to zero and then becomes imaginary at large internuclear distances. We show that this unphysical behavior occurs because ωT2 must become negative wherever symmetry breaking lowers the energy of the ground state solution below that of the symmetry unbroken solution. We use the fact that the ΔSCF method gives a qualitatively correct first triplet excited state to derive a “charge-transfer correction” (CTC) for the time-dependent local density approximation (TDLDA) within the two-level model and the Tamm-Dancoff approximation (TDA). Although this correction would not be needed for the exact exchange–correlation functional, it is evidently important for a correct description of molecular excited state potential energy surfaces in the TDLDA. As a byproduct of our analysis, we sh...

Extended gaussian-type valence basis sets for calculations involving non-empirical core pseudopotentials

Revised values of core pseudopotential parameters are presented together with optimized PS-31 G valence basis sets for atoms from Li to Ca and from Ga to Kr. Studies with more than thirty test molecules indicate a very good agreement with 6–31 G results for atomization energies, geometries, dipole moments and force constants.

Size‐consistent self‐consistent truncated or selected configuration interaction

Based on the principle of intermediate effective Hamiltonians, a simple procedure is proposed in order to eliminate the unlinked contributions of any truncated or selected configuration interaction (CI). The corrections are diagonal energy shifts, easily calculated. A self‐consistent version is proposed, which insures separability if localized molecular orbitals (MO) are used. In the special case of double CI, the present method is an improved version of the coupled electron pair approximation (CEPA), but it may be applied to any selected model space, involving configurations of various degrees of excitation. The efficiency of the proposed algorithms is illustrated on a series of test calculations performed on Be2, F2, N2, and NH3.

Inclusion of core‐valence correlation effects in pseudopotential calculations. I. Alkali atoms and diatoms

A simple procedure is developed to include the core‐valence polarization and correlation effects in small metallic clusters of alkali atoms. The fluctuation of the electric field on the polarizable core, as calculated from optimized limited valence basis sets, plays the dominant role. The atomic ionization potentials and the ground state spectroscopic characteristics of Na2, K2, Cs2, and of their positive ions are significantly improved. The method is applicable to polar compounds as well, as illustrated through a study of CsH. The relationship between the present procedure and Bardsley’s −αD/2 (r2+d2)2 potentials is discussed.

AB initio study of the regular polyhedral molecules N4, P4, As4, N8, P8 and As8

Ab initio pseudopotential SCF calculations were performed on tetrahedral X4 molecules using double-zeta basis sets with and without d functions. The inclusion of d orbitals shortens the bond lengths, stabilizes the X4 structures and intensifies the electron density inside the tetrahedron. The cubic X8 molecules, calculated without d AOs, are not predicted to be more stable than 2X4. Repulsions between parallel bonds in X8 may compensate the lack of ring strain.

Theoretical studies of the actinides: method calibration for the UO22+ and PuO22+ ions

Abstract As part of method assessment for the theoretical study of actinide systems, we have performed 1-component relativistic pseudopotential calculations for the uranyl and plutonyl ions. The calculated spectroscopic constants compare well with fully relativistic 4-component results at both the Hartree–Fock and correlated levels of theory. We deduce that second-order spin–orbit effects in these systems are minor and show that the 1-component method chosen (B3LYP in particular) gives reliable results at low computational cost. The ground state of the plutonyl ion has been determined as 3 H g .

Multireference self‐consistent size‐consistent singles and doubles configuration interaction for ground and excited states

One proposes a state‐specific self‐consistent dressing of the configuration interaction (CI) matrix built on a multireference space and all the singly and doubly substituted determinants. The dressing insures size consistency [and separability when localized molecular orbitals (MOs) are used]. In the here‐proposed solution, which generalizes a previous single reference method [(SC)2CI, J. Chem. Phys. 99, 1240 (1993)] valid only for the research of the ground state, all the reference determinants play an equal role and the method is applicable to excited states. The implementation will be simpler if the reference space is a complete active space, but this restriction is not compulsory.

Inclusion of core-valence correlation effects in pseudopotential calculations. II. K2 and KH lowest Σ+ potential curves from valence-correlated wavefunctions

For pt.I see J. Chem. Phys., vol.77, p.3571 (1982). The recently proposed perturbative treatment of core-valence correlation effects is extended to valence correlated wavefunctions. As expected, the valence correlation slightly diminishes the core-valence correlation for the ground state. This effect is illustrated for the atomic electron affinity of K. The core-valence correlation plays a very important role in the spectroscopy of K2 and KH, where it increases the lowest X1 Sigma + to A1 Sigma + transition energies by 2300 and 1100 cm-1, respectively.