Conceptually it should be possible to pre-compute the Earth illumination values in order to substantially speed up ACIS FP model evaluation. There is a question of gridding. Offhand perhaps it would be sufficient to have:
20000 points evenly distributed over the hemisphere where illumination > 0
1000 points for Earth distance, probably spaced somewhat logarithmically (as Earth gets farther away then changes make less of a difference). At some point Earth is effectively point-like and one can just use 1/r^2.
This would result in a data file around 80 Mb assuming 20 million float32 values.
Current performance is faster than 2 ms / point, so this can be computed in less than a day.
[EDIT: number of grid points substantially reduced from this in #63 without impacting accuracy substantially]
Conceptually it should be possible to pre-compute the Earth illumination values in order to substantially speed up ACIS FP model evaluation. There is a question of gridding. Offhand perhaps it would be sufficient to have:
This would result in a data file around 80 Mb assuming 20 million float32 values.
Current performance is faster than 2 ms / point, so this can be computed in less than a day.
[EDIT: number of grid points substantially reduced from this in #63 without impacting accuracy substantially]