Simulation Suite

[brandonshensley]

As suggested by @zonca, I am starting an issue where we can discuss how the new models could be used as a baseline simulation suite, with the goal being to have models that span optimistic/baseline/pessimistic (or similar spirit). My initial suggestion is:

1. d9s4f1a1
2. d10s4f1a1
3. d12s3f1a2

where I have assumed s4 corresponds to the new synchrotron model with small scales in both amplitude and spectral index. Another point of discussion is whether some or all of these should include CO emission and polarization. Thoughts welcome.

[seclark]

Thanks, this is a great starting point. Why s3 for #3rather than s4? I think yes to CO lines: probably for all of them unless there's a good argument against that. If we were just going for a steady ramp in complexity we could consider incorporating polarized AME into model #2 but with a lower polarization fraction than a2. But in our current conception of #2 I think a1 is the right call.

[brandonshensley]

Thanks, this is a great starting point. Why s3 for #3rather than s4?

s3 is more spectrally complex (curvature parameter) while s4 has fluctuations on smaller scales, including in beta. s4 is the natural pairing with d10, so I put s3 with d12. But we could also use s4 for both and change s4 to s0 in the "optimistic" model (which would make sense since d9 is an alternate version of d0). So:

1. d9s0f1a1 + CO
2. d10s4f1a1 + CO
3. d12s4f1a2 + CO

From a theory point of view, I would deem any AME polarization "pessimistic."

[giuspugl]

Thanks for starting this! I agree with models @brandonshensley proposed for the next sim run. About CO, we have several caveats to consider before including it:

• CO templates comes from Planck released maps. We smoothed them at 1 deg to avoid noise contamination especially at high galactic latitudes.
• we can simply treat it as an unpolarized component, in the pysm we have included an argument not to forecast polarization from CO (see https://so-pysm-models.readthedocs.io/en/latest/models.html#colines for documentation) ,so it can be essentially treated as free-free and ame.

[brandonshensley]

@giuspugl, maybe we can turn on polarization for #3 only?

[zonca]

I think for synchrotron we could do s4 for uniform spectral index, s5 with varying spectral index, and we can have unpolarized co1 and polarized co2, so:

1. d9,s4,f1,a1,co1
2. d10,s5,f1,a1,co1
3. d12,s5,f1,a2,co2

Using # automatically links to issues, need to escape it with backtick e.g. #3

[brandonshensley]

Thanks @zonca. I might have used co3 for #2 and #3--I realize the high latitude component is constrained by data only statistically, but it feels "realistic" to include.

How difficult would it be to have an s6 that is identical to s5 but also includes the curvature parameter from s3? Does anyone have a strong opinion about whether having a model with spectral curvature is important for next-round simulations?

[giuspugl]

@giuspugl, maybe we can turn on polarization for #3 only?

Yes !

I realize the high latitude component is constrained by data only statistically, but it feels "realistic" to include.

the numbering proposed in #86 perfectly goes in the perspective of optimistic, realistic and pessimistic.

[zonca]

for now we have (using co3 for 2 and 3):

1. d9,s4,f1,a1,co1
2. d10,s5,f1,a1,co3
3. d12,s5,f1,a2,co3

next we plan to have a Synchrotron model with curvature for 3 and possibly 2, see #105

Documentation reference:

• d9 d10 GNILC based models and d12 MKD 3D layered dust model: https://pysm3.readthedocs.io/en/latest/models.html#dust
• Synchrotron models s4 and s5: https://pysm3.readthedocs.io/en/latest/models.html#synchrotron
• CO: https://pysm3.readthedocs.io/en/latest/models.html#co-line-emission
• All other models are the same of PySM 2: https://pysm3.readthedocs.io/en/latest/models.html

[zonca]

Synchrotron curvature PySM 3

[zonca]

@brandonshensley @seclark should we use the new s7 model (https://pysm3.readthedocs.io/en/latest/models.html#synchrotron) with curvature for the most complicated sky model?

[brandonshensley]

@brandonshensley @seclark should we use the new s7 model (https://pysm3.readthedocs.io/en/latest/models.html#synchrotron) with curvature for the most complicated sky model?

Yes, let's do that.

[zonca]

Switched to s7 for the more complex sky model:

1. d9,s4,f1,a1,co1
2. d10,s5,f1,a1,co3
3. d12,s7,f1,a2,co3

[seclark]

Yes agree! Looks good.

[zonca]

@brandonshensley this would make a good post for the Panexp blog!

[zonca]

see https://galsci.github.io/blog/2022/common-fiducial-sky/