Step 3.4: Test that xRFI doesn't screw things up

[steven-murray]

3.4: Test that xRFI doesn't screw things up

This test was initially proposed by @AaronParsons. The idea would be to run xRFI over simulated visibilities that have no xRFI in them, to ensure that it doesn't inadvertantly flag things that will screw up the PS.

• Simulation Component: EoR, FG, noise, cross-talk(?), reflections (?)
• Simulators: rimez, hera_sim
• Pipeline Components: xRFI, maybe some calibration, if required to run xRFI.
• Depends on:

Why this test is required

Other RFI tests are tricky because we have to try to simulate RFI robustly. This test is much simpler and still tests something useful.

Summary

A brief step-by-step description of the proposed test follows:

• Produce true visibilities with GSM from rimez and a simple power-law P(k).
• Add systematics (not xRFI)
• Run xRFI. If no flags: done. If some flags:
• Run in-painting
• Run pspec.
• Compare output to known input EoR at relevant k modes.

Simulation Details

Will fill these out later...

• Freq. range:
• Channel width:
• Baseline/antenna configuration:
• Total integration time:
• Number of realisations:

Criteria for Success

• Either no flags or estimated P(k) within 1% of true P(k) for k > 0.5 or so.