Difference between extract and live or die for models where Compton Scattering Dominates

[kslong]

In the live or die mode, we sum up all of the photons that emerge within a a range of degrees centered on the desired extraction angle, both above and below the xy plane.

In situations where photons are reflected or scattered from the disk or central object, essentially all photons are collected in this mode. This can result in differences between this approach and extract, if a lot of photons are scattered back onto either surface. This is because extraction takes place in a specific direction and photons that try to go though a surface are not counted.

An example of a case where this occurs, is for a spherical wind and a star. In this case, the photons that hit the "back" side of the star are not counted in extract mode, but do get counted in live or die.

The extract results are, in this case, more physical than live or die.

This does not happen if the star and disk are treated as absorbing.

(Effectively the same is true, if we want to calculate phase resolved spectra. While photons that hit the secondary are "absorbed", the spectra that are produced will be the same no matter what phase is chosen.)

I believe this is simply something that should be documented.

[kslong]

Here is an example of the problem.

And here is the same thing running the same model in live or die mode

[kslong]

This problem is almost certainly due to the fact that we do not re-weight photons in Compton scatters. This is needed to get the correct results because Compton scattering is not isotropic

[kslong]

I have made some progress on solving this issue at least in the Thompson limit.

Here is a comparison between the results before the fix and after:

[kslong]

I have made additional modifications to solve the live or die problem for true Compton scattering. For a model like the one above, but for monochromatic photons at 40 keV, one obtains, with commit 5b30a9ad2f2aeeb7769a95fcdf172bc5c64573ae the following: