Hans Moritz Günther

@DougBurke I'm not sure how this interacts with your code to include new XSPEC versions, but all the XSPEC models I touch are pretty well established and stable, so hopefully...

And... apparently one of my rebases has gone wrong and removed 6000 lines of XSPEC models... will fix ASAP.

I'm leaving the test failures for now - please review. The failing test is an example what kind of difference we will get with this PR: Because the of numerical...

You did not build XSPEC successfully. That likely means one of two things (or both): - You did not run the script successfully. When you tried "% ./scripts/use_ciao_config", did it...

No, that would be additional work. There is more that could be done with this scheme of Sherpa doing part of the calculation (red-shfited absorption models and the norm for...

I note that this is not applied randomly: We need to hand-select which model to apply this to. So it's not "some model somewhere", it's models with known physics. In...

Yes - the idea is to replicate the XSPEC model with a relative tolerance of 1e-4 or so - that will preserve all real wiggles.

@anetasie When I talk about "wiggles" in https://github.com/sherpa/sherpa/pull/2080#issuecomment-2246980233 I mean "wiggles in a a plot of transmission vs. N_H for a fixed energy". That particular plot should be a smooth...

Testing over a larger range is the right call though. This plots shorts the difference between evaluating directly and using my cache, x-axis is energy in keV. Note the y-scale...

I've updated this. I still want to do some sleuthing about why exactly I needed to loosen up some of the numerical comparisons in the tests, but first, let me...