The easiest way is to use the final result of the solution over a tile as your first guess as to the solution on the next tile.
> A thought crossed my mind - at present solvable parameters are being solved for separately from > scratch in each tile. The reality, however, is that after we arrive at a solution for the first > tile, things are unlikely > to change much for succeeding tiles (unless a source has intrinsic > time variability). So I would think that one could interate through > all tiles more quickly if the final solution results for the just processed tile could be stored > and provided as initial guesses for the > tile that is just being loaded. is that possible?
It's not just possible, it's the default behavior. 
I had explicitly disabled it to make the displays more "educational" -- otherwise you'd see no meaningful iterations past the first tile. This is controlled by the use_previous keyword to Parm; I set it from the reuse_solutions global variable defined at the top of the script. Set reuse_solutions=True and watch the speed improvement.