Gray vs multigroup #5
Conversation
|
@brryan - sorry to lose track of this - are there any significant further updates pending? |
|
@brryan - if you are still catching updates here, is this ready to merge? |
Hey @RyanWollaeger sorry for my delay, yes I think this is ready to merge (or at least I don't remember any outstanding issues). It's not really tested so it might be useful to add a test of actual working multigroup, like frequency-dependent bremss thermalization that shows the gas temperature undershoot the equilibrium at first. |
+ Cancel log and exponential where possible. + Add comment explaining nu formula is log-space group midpoint.
+ Fix the formatting using style/format.sh. + Use new AbsorptionCoefficient API for mean opacity. + Update some parts of the README.md file. Note: similar to MeanOpacity's Emissivity, AbsorptionCoefficient defaults to Rosseland-weighted grey opacity.
|
Looks like |
+ Remove division by frequency in source particle creation. + Update singularity-opac commit to correct mean opacity accesses. + Have regressions fail if mean errors are NaN.
|
@pdmullen - the most recent commit should get the existing tests to pass. Still need to add an actual multi-frequency test. |
Guess not - the passed for me locally, at least. |
|
The |
It breaks, as reasonably expected. |
+ Update parthenon submodule. Note: the spiner header is included through singularity-eos, not singularity-opac.
RyanWollaeger
force-pushed
the
brryan/gray_vs_mg
branch
from
d284d50 to
a11cc40
Compare
(Thanks Adam Dempsey.)
+ Use mean opacities in transport_ddmc. + Add PARTHENON_REQUIRE gray assertions for DDMC.
RyanWollaeger
force-pushed
the
brryan/gray_vs_mg
branch
from
2bc1eb6 to
9376c83
Compare
|
What counts as documentation (looking at the checklist in the description)? |
Let's have a dedicated PR to add documentation for |
Note: this is the same as the emission CDF sampling, which fixes particle frequencies at the log-center of the sampled group.
+ Use Bremsstrahlung multifrequency opacity option. + Use more physical initial temperature values in inf*.in inputs.
|
Added a 16 group infinite medium test that uses the Bremsstrahlung opacity, but haven't added it to the CI - it is pretty fast though, since the time step is small. If there are no objections (e.g. making the test better, which is a reasonable request), will add it to the CI test suite... |
Background
We want to have both gray and multigroup modes
One question is whether this should be a continuous energy or a multigroup representation.
Singularity-opacright now always accepts a continuous frequency argument; maybe this needs to be expanded upon with a multigroup integer index variant. Should multigroup support nonuniformly binned data?Description of Changes
jaybennepackage initialization into gray vs mg depending on which singularity variants are presentChecklist
New features are documented