xband: windsave2fits, changes to radiation temperature and fixes to disk_diag file. #1104
Conversation
Only a place holder for a new mode has been added at this point.
Basically, at this point matrix_populations2 is identical to matrix_populations, so this is just a routine to be used to develop a new ionization mode
The normal modes are unchanged but various changes have been made in matrix_ion2 to slow down how rapidly the temperature can change. This still is not fundamentally different from the previous approach.
Previously setting make_tables to yes, just ran windsave totable on each cycle, now the disk_diag file is also saved. Cleaned up writing of the disk_diag file when no photons hit disk.
Variables have been added to the disk structure to allow for better tracking of where photons are emitted from the disk, and where they hit the disk.
|
Thanks for doing this Knox. This looks good overall, but before we merge we jsut need to a) make it pass the tests (something failing at the moment in make check) |
|
The problem is in the unit_test and does not show up on either may Mac or Linux. It’s associated with the test ionization mode I put in. I probably could remove it.
The radiation change is actually important. If you have fully covered the spectral range of the emission it does not matter, but for example if you set the low end of the ionization frequencies to 0.2 eV or something, and have low temperatures, one gets a very wrong answer to thae actual temperature.
Knox
|
|
I removed the mode I was planning to create for multicylcle determination of the temperature, since I will not have time to work on this before the transition to Sirocco. The tests got further but are now failing later in one of the tests for things that I have not touched. My guess is that something about the tests has changed, and that possibly one should to ahead. Or I can try merging dev in to xband first. But **I note there is no documentation for developers of the tests **, and the reason I was intially failing is the some of the Makefiles in tests must be updated if you add a file. Documenation is needed going forward. |
|
Unit test documentation is here: https://agnwinds.readthedocs.io/en/dev/developer/tests.html It looks to me like the reason the tests fail is because FRACTIONAL_ERROR was changed from 0.03 to 0.001 and FRACTIONAL_ERROR is used in a different test - since this would change our convergence criterion for ne, I'm assuming this is not intended? |
|
I did change the fractional error because we were concerned about some of the ionization results, but it does not make any difference. So I have restored the old value Note that this is also defined separately in saha.c. The tests now show no errors, but they also do not seem to finish. I apologize for missing the note about help for running the tests; neither currently works on either my Mac or Ubuntu. I will have to look into that after I return from fishing. Hopefully though this pull request can be merged, so we can restarte with Sirocco. |
jhmatthews
changed the title
|
Please note that, because I don't want to change the calculation of t_r just before release I've currently made the t_r calculation depend on a variable |
The main changes aside from documentation are: