[FLASH-USERS] Flash 4.4 OPAL opacity

[Amit Kashi]

On Fri, December 2, 2016 6:52 pm, Klaus Weide wrote:
> On Fri, 2 Dec 2016, Amit Kashi wrote:
>
>> Thanks for helping me. The main indication was that the code
>> crashed during the first timestep with no specific error message.
>> Together
>> with the empty opacity_printout_tables.txt I thought it is related to an
>> issue with loading the opacity tables. But I don't have a definite
>> indication that this is the reason. The last line in the log file is
>> just
>> the step: n=1  line.
>
> Hi Amit.
>
> I have seen no indication in the output you have provided that anything
> went wrong in loading the tables. However, you should make sure you see
> those lines for each of the table files that you have specified. For the
> full set, you should be reading in 20 files, so make sure you see 20 of
> those sections in your output, and that the filenames are all different.
>
> 20 = op_opalNumHydrogenAbundances * 2; op_opalNumHydrogenAbundances is the
> number of abundance ranges and defaults to 10; the factor 2 is
> because there is a lowT and a HighT table for each abundance range.
>
>
> Anyway, you write that the code 'crashed with no specific error message'.
> What is the last output? Are there any 'non-specific' messages?
> Anything about segfaults or similar?
>
> I ask because it is not clear from what you have written that the abnormal
> end of your run happens in the Opacity code.
>
>

Hi Klaus,
There was no message at all. The code just stopped running.
The strange thing is that when I ran it not in debugging mode it didn't
crash, so there might be something in the debugging mode that causes it.


>> I am quite sure that the OPAL files I use are in the right format, they
>> even have the same names as in the README file.
>
> Not exactly the same names. You have 'gn93' where the names in the README
> (and the files that I have tested with) have 'gs98'.  I have not compared
> the different sets (and am not familiar with them), but can see already
> from your output that there are some differences in the sizes of some
> tables.
>
> Perhaps you could check whether the code behaves better if you use 'gs98'
> tables (even if those aren't what you want to use in the end).
>
>

I changed to gs98 but I don't think this was the issue.

>> Maybe the runtime parameters are not correctly set? I use only one
>> species
>> (h1).
>> Here are the relevant lines:
>>
>> ...
>>
>> And different values of the parameter op_opalnumhydrogenabundances
>> that accepts values from 0-10, but I don't understand what it exactly
>> does.
>
> I agree that the OPAL table feature is underdocumented...
>
> However, the following files should be useful to get some better
> understanding of runtime parameters, especially their meanings
> and their current values:
>
> * Source code:
>     physics/materialProperties/Opacity/OpacityMain/OPAL/README (You know
> this)
>     physics/materialProperties/Opacity/OpacityMain/OPAL/Config (parameters
> defined)
>
> * After setup:
>     object/setup_params : all runtime parameters configured in with doc
> strings
>     object/default.par  : all runtime parameters configured in
>
> * After running:
>     flash.log (or whatever.log): values of 'op_' runtime parameters used
>
> Please let me know if you need further explanation about the runtime
> parameters after you have consulted those files.
>
> Klaus
>

Thank you, this was helpful.
It now runs well.
The information in the files you listed helped me understand what the
undocumented parameters do.
One problem I encountered is that in the temperature read in Opacity.F90
was TELE that I don't use in my simulation rather then TEMP, so that lead
to NAN in the opacity values. After replacing it with TEMP it runs well.

Thanks again for your help.
Amit