Currently, a user can specify a degree of oversampling of the PSF on the discrete grid (of the source map) so that the resulting PSF is more accurate for convolution. While this gives control to the user, it lacks clarity and protection to make sure the user is doing the right thing because there is no good default to pick. Any degree of oversampling would be overkill if someone supplied a source map with 0.1-arcsecond pixels.
An alternative is to automatically determine an appropriate level of oversampling. The resolution of the FOXSI optics isn't going to unexpectedly get dramatically better, so as long we subsample at the ~0.2-arcsecond level, the inaccuracy will be very low. We calculate and apply the degree of oversampling needed to reach that level of subsampling. In other words, for a source map with 3-arcsecond pixels, the automatic oversampling would be a factor of 15, while for a source map with 1-arcsecond pixels, the automatic oversampling would be a factor of 5.
The target of 0.2-arcsecond subsampling can be a user-specified parameter, so that the user can specify finer subsampling if accuracy is critical, but the default would be good for a broad range of source maps.
Currently, a user can specify a degree of oversampling of the PSF on the discrete grid (of the source map) so that the resulting PSF is more accurate for convolution. While this gives control to the user, it lacks clarity and protection to make sure the user is doing the right thing because there is no good default to pick. Any degree of oversampling would be overkill if someone supplied a source map with 0.1-arcsecond pixels.
An alternative is to automatically determine an appropriate level of oversampling. The resolution of the FOXSI optics isn't going to unexpectedly get dramatically better, so as long we subsample at the ~0.2-arcsecond level, the inaccuracy will be very low. We calculate and apply the degree of oversampling needed to reach that level of subsampling. In other words, for a source map with 3-arcsecond pixels, the automatic oversampling would be a factor of 15, while for a source map with 1-arcsecond pixels, the automatic oversampling would be a factor of 5.
The target of 0.2-arcsecond subsampling can be a user-specified parameter, so that the user can specify finer subsampling if accuracy is critical, but the default would be good for a broad range of source maps.