Which data should we compare to as a benchmark?

[bnord]

For both our simulation results and our downstream inference analyses, we'll need to compare to existing codes and/or benchmark simulations to prove the fidelity of our simulation and the fidelity of our inference approaches.

We should consider how to check and verify the following elements for simulation

1. SZ cluster Temperature or Decrement profile
2. Submap profile when noise is included
3. being able to recreate a subset of a realistic simulation or a given observed data set.

We should consider how to check and verify inference on our sims compared to inference on more complicated sims/data.

1. how inferring on our simulated data compares to inferring on existing data sets: what tools (e.g., SBI vs. neural networks) work better.
2. how systematics in the forward modeling in simulations make themselves apparent through different inference techniques.

Candidate data sets for benchmark comparisons

1. Existing observed SZ cluster data sets
2. Existing simulated SZ cluster data sets, like through Seghal, 2010.