-
Notifications
You must be signed in to change notification settings - Fork 5
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
VELOCIraptor zeros all fields with metallicity of star-forming gas #75
Comments
Update: the masses of star-forming gas are also zero if VR is run with config 2. |
This sounds like a similar problem to that from #72, but @EvgeniiChaikin clearly tested using the latest fixes introduced to fix #72. That plus the fact that @MatthieuSchaller was mentioned in the original description point out that this is a different problem, although probably closely related. In #72 (comment) there is a reference to some values not being zero though, so again, I suppose this is slightly different. |
I tried to reproduce this in comsa6 but it actually worked for me. My build has both MPI and OpenMP and HYDRO all ON. What were your compilation flags?
|
I just tried again with And actually I just compared the two original output files pointed at by @EvgeniiChaikin and their
@EvgeniiChaikin maybe there's a problem with the python tools you are using to read these files? Or maybe I'm missing something obvious from the original description of the problem? |
Hi @rtobar, Thank you for helping me out! I complied VR with I saved the output from the compilation process into This bug is probably not related to the python tools I am using because I was also able to reproduce it using
I presume the reason why you were not able to reproduce the bug so far is that you have been using |
@EvgeniiChaikin ahhhhh what a silly mistake on my side! I lost the sf bit in the name of the dataset, I'll re-run things to double-check again and confirm. |
I was now able to reproduce this, sorry it took so long! This is a problem specific to MPI builds it seems. I tried to reproduce the problem with an MPI enabled build, then with an MPI disabled one, and only the former produced the problem:
This will hopefully help constraining the search for the underlying problem, plus it indicates there is a workaround for small runs that can happen on a single node. |
Thanks a lot @rtobar ! I mostly feed VR with small cosmological boxes so your fix is perfect for me! |
Could the recent fix to the negative densities over MPI have magically fixed this? |
@JBorrow is that not something we see with the EAGLE runs? |
Is that version then different from the one @EvgeniiChaikin uses? |
@EvgeniiChaikin (when you have time) the fix (#79) that is now in master may have helped here. |
I wanted to try this out with the original config/inputs reported in this ticket but they were gone. Hopefully the fix in #79 has some positive effect on this problem -- the fact that MPI builds work in some cases and not in others could be indicative that this was an uninitialised variable problem indeed. |
Agreed, that seems plausible. |
I re-created the same test as in the description. I ran it for two versions of the VR
In both cases, VR was compiled with MPI. As before, the older version would produce the metallicities of star-formation gas equal to zero. However, that is not the case for the newer version (May 26). To confirm this result, I used 5 different snapshots for this test and never found the star-forming gas to have zero metallicities in the catalogues produced by the newer version of VR. I thus conclude that the bug has been fixed and close this thread. Thanks everyone for the help! |
Great news @EvgeniiChaikin, thanks for retrying this! |
Describe the bug
In short , when I introduce a new gas internal property field to the VELOCIraptor config file, VELOCIraptor begins to zero all fields with metallicity of star-forming gas.
More preciesly, I have two config files of VR, where the
diff
between the two isi.e. compared to config 1, config 2 has an additional field
IronOverHydrogenMasses
.After I run VR
and look into the output fields, I find
where one can see that in the latter case (config 2), the field
apertures.zmet_gas_sf_100_kpc
has no positive values. And this is true not only for the 100-kpc apertures but for all types of apertures.To Reproduce
/cosma7/data/dp004/dc-chai1/VR_ICRAR_27_March/VELOCIraptor-STF
(version64de17bff6925f47f3ebe8f8195108801d661d95
)/cosma7/data/dp004/dc-chai1/test_VR/colibre_0023.hdf5
/cosma7/data/dp004/dc-chai1/test_VR/vrconfig_3dfof_subhalos_SO_hydro_?.cfg
, where?
is1
for config 1 and2
for config 2@MatthieuSchaller
The text was updated successfully, but these errors were encountered: