GQC: requirements for DR3 mocks

Durham, July 15th

accuracy: \(\lesssim 5\%\), on scales \(k < 0.3-0.4 h/\mathrm{Mpc}\)
O(1000) fast lightcone mocks, with (ideally):
- BGSxLRG, LRGxELG and ELGxQSO cross-correlations
- bispectrum
- bias evolution on the full \(z\)-range
with fiber assignment, photometric (and spectroscopic) weights
DR2 mocks produced by CAI met the requirements
DR3 mocks are already being produced by CAI
I think we'll have analytic covariance matrices for (\(P\), \(B\), post-recon) for DR3

For the theory mock challenge:
- Exotic HOD flavors (a focused GQC - GQP - CAI activity)
- Current 25 repeated \((2 \mathrm{Gpc}/h)^3\) Abacus boxes (\(\sigma_\mathrm{DR2} / \sqrt{25}\))
- Averaging over the line-of-sight \(\sigma(f) / \sqrt{3}\)
- We should apply control variate, gaining us \(\times 1.5\) Hadzhiyska +23
Also used for the pipeline validation (box \(\Rightarrow\) cutsky, fiber assignment), but it's probably fine to rely on faster mocks for this?
We probably want to share these "pipeline validation mocks" across WG

Initial conditions are always useful! (CV, reconstruction)
Having a shared fast pipeline from DM lightcone to catalogs would be nice for GQC to be able to add / plug realistic effects one-by-one:
- deviation to flat-sky approximation
- selection effects \(\propto \partial n(z) / dz \)
- evolving bias, growth factor, growth rate
We probably also want a coordinated effort in test of observational systematics (imaging)

Bunch of observables

Potentially very large data vector

Compression with "Greedy combination"

Also including small scales...
(not clear what the requirements are there)

Thanks to our sponsors and

69 Participating Institutions!