I've experimented with three different DEM computations thus far (implemented in Python):
simple regularized (Plowman method)
sparse (Cheung method)
regularized (Hannah-Kontar method)
The first seems to give quite different results than the other two: am I running it wrong? Need to go back and double check the EMToolKit implementation.
In particular, I've found that it seems to put very little emission anywhere except the brightest ARs.
The first is the fastest. The other two are exceedingly slow for the FOV/resolution we need to run on: well over 30 minutes.
All are slow because of the need to loop over every single pixel (many 100K) and to do multiple iterations per pixel.
Need to investigate approaches for accelerating this computation.
Some possible ideas:
I've experimented with three different DEM computations thus far (implemented in Python):
The first seems to give quite different results than the other two: am I running it wrong? Need to go back and double check the EMToolKit implementation. In particular, I've found that it seems to put very little emission anywhere except the brightest ARs.
The first is the fastest. The other two are exceedingly slow for the FOV/resolution we need to run on: well over 30 minutes. All are slow because of the need to loop over every single pixel (many 100K) and to do multiple iterations per pixel. Need to investigate approaches for accelerating this computation. Some possible ideas: