DC-0 Validation spectra: Dust

zonca commented 1 year ago

@giuspugl I am looking rough spectra of DC-0 inputs for validation purposes. These are unmasked full sky anafast spectra up to 2.5 * ellmax.

I noticed a drop around ell=2000 in dust d10 in all channels, worse at low frequencies:

What is strange is that this does not show up in d9, which has the same templates.

so it must be coming from dust temperature or spectral index?

I haven't transferred the maps to NERSC yet, will post here when I do.

zonca commented 1 year ago

Polarization looks fine:

jdborrill commented 1 year ago

Did this (and the synchrotron) issue not show up in the SO sims for the same sky models you ran a month ago, Andrea?

J

On Fri, Jun 2, 2023 at 2:34 PM Andrea Zonca @.***> wrote:

Polarization looks fine:

[image: image] https://user-images.githubusercontent.com/383090/242995320-ebc31748-3b86-4e77-b61d-bcba52612af2.png

— Reply to this email directly, view it on GitHub https://github.com/CMB-S4/s4mapbasedsims/issues/28#issuecomment-1574341974, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAC4LSV37QQXMWN4E7MQJCTXJJL6ZANCNFSM6AAAAAAYY2FCPY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

zonca commented 1 year ago

@jdborrill the synchrotron issue yes, see https://github.com/simonsobs/map_based_simulations/issues/44#issuecomment-1481586936. I asked if someone else could investigate the issue on the Penexp call and on Slack https://simonsobs.slack.com/archives/CE44LGAAK/p1679592345528889.

About dust, I also noticed it, see https://nbviewer.org/gist/zonca/7026e5f4fd9ef304a89f1c171e43f2ce, I didn't flag it because I assumed it was due to how we generate small scales. Also now I think the same, but if someone else also looks at it better. And anyway it is better to flag it.

giuspugl commented 1 year ago

it might be related to spectral indices indeed, @zonca the spectra param. maps should be produced at ellmax=2000 (this was because of testing purposes but we should have kept the same ellmax as the amplitude maps. therefore small scales behave differently at different multipoles and different freq., this is why we do see a clear glitch .

zonca commented 1 year ago

thanks @giuspugl, I will look into this.

zonca commented 1 year ago

confirm there is an issue both in beta and Td, templates are unaffected

Beta at 2048 and 4096

Td at 2048 and 4096

Templates 2048 and 4096

zonca commented 1 year ago

Synchrotron beta looks fine

zonca commented 1 year ago

it was a bug in the notebook that generates the spectrum of the small scales, it was using Lmax=2048 instead of 16384. now:

zonca commented 1 year ago

I am regenerating the beta and Td maps at all resolutions, after we figure out the synchrotron issue, I'll release 3.4.0b9 and rerun the updated models.

zonca commented 1 year ago