The present calculations were carried out in LS coupling, since the relativistic effects in photoionizing the low ionization stages of iron should be small. Although they might be significant for some types of transitions in Fe IV, it would be impractical at this stage to carry through a large-scale relativistic calculation involving a number of channels several times larger than the already huge number in the LS coupling case. Also, the inclusion of fine structure has only a marginal effect on the calculation of the Rosseland or the Planck mean opacities (Seaton etal. 1994).
The second summation in the CC expansion (Eq. 1) represents short range correlation functions. These functions are very important in obtaining accurate (e + ion) wavefunctions, but may cause pseudoresonances, particularly if the two summations in Eq. (1) are inconsistent (Berrington etal. 1987; also discussed later). Owing to the magnitude of the calculations, the set of functions for Fe IV was optimized to include only the minimum number of correlation functions having important effects on the bound states energies and bound-bound oscillator strengths. Nonetheless, this component of the close coupling expansion is large (Table 2 (click here)).
The overall calculation was divided into two groups of total (e + ion) symmetries according to their multiplicity, i.e. (2S+1) = 4, and 6. For each multiplicity we consider total angular momenta L = 0 - 8, for both parities, and a total of 31 's corresponding to the target terms in Table 1 (click here). Given the present large expansion, it was not possible to include the doublet symmetries, which are not connected to the ground state , due to memory and disk space constraints (several tens of gigabytes). On the other hand, a calculation with a reduced number of terms would probably yield results similar to those of Sawey & Berrington (1992; SB) who included only the terms in their OP calculations.