Cosmo vec
Implement cosmological terms in vectorised hydro interactions.
Fixes #406 (closed).
Merge request reports
Activity
I have noticed that you recompute
a2_Hubble
andfac_mu
every call to the interaction function. Can you not compute these terms once in DOPAIR2 and DOSELF2 and pass them to the interaction function? Instead of recomputing them every call to "runner_iact"? Or is that an issue when it comes to the interaction function interface?It's an interface issue.
Gadget needs these terms. But other schemes might need other terms. The only thing I can guaranty is that any term we will need is always a combination of
a
andH
. Now, in general, I do not think that this is an issue anyway. We are currently bandwidth and register limited, not computing limited. Plus, thefac_mu
term is exactly1
(and implemented as such in the scalar code) for the most common casegamma=5/3
.added 1 commit
- 0901bce8 - Caclulate cosmological terms inside interaction function.
I have tested this with
make check
but are there any other tests for cosmology yet?I did not respond to this, sorry.
No, no more tests at this stage besides running the EAGLE_xx with cosmology on and checking that you get the same asnswwer as scalar. (Which still does not tell us whether both of them are correct...)
@pdraper for reasons I don't understand, I can't accept this merge request. Has the repository been locked?
mentioned in commit 11f5d292