Opacity sampling in radiative acceleration calculations

The accuracy requirements on atomic data for the calculation of stellar evo lution with atomic diffusion are determined. In particular, the density of frequency grids needed for precise radiative acceleration (g(rad)) calculat ions via the sampling method are presented. In order to minimize the number of frequency points needed for precise g(rad) calculations, frequency grid s that are more refined in the regions of the spectrum where the radiative flux is large are suggested. The following number of frequency points are n eeded for opacity table calculations to be used in stellar evolutionary cod es including diffusion: 50,000 points for 4.20 less than or equal to log T less than or equal to 4.5, 30,000 points for 4.5 < log T less than or equal to 4.8, 10,000 points for 4.8 < log T less than or equal to 5.5, and 4000 points for log T > 5.5. These opacity tables would render possible the stud y atomic diffusion in the exterior regions of certain chemically peculiar s tars such as Ap or HgMn stars. In the sampling method, correction factors c an be applied after the basic integrations over sampled spectra to include such effects as ion velocity averaging, redistribution of momentum among io ns, and electron recoil during photoionization; these corrections are evalu ated and illustrated for a few typical stellar models. Silicon is used as a n example to show that the corrections are important mainly for T < 50,000 K. These corrections are used in stellar evolution calculations based on OP AL monochromatic opacity tables.