Step 2.3: Verify that the Abscal Amplitude Bias is Fixed

[jsdillon]

2.3: Verify that the Abscal Amplitude Bias is Fixed

In the process of the H1C IDR2.2 limit validation, we discovered that noise introduced a bias low in the abscal gains. This was accounted for with a correction factor in the limit paper. We'd like to do better.

• Simulation Component: Foregrounds and Noise
• Simulators: Not particularly picky, just needs to do foregrounds and noise appropriately
• Pipeline Components: Redcal and Abscal
• Depends on: Test 2.0

Why this test is required

In H1C IDR3, we used a pipeline that includes a fix for the bias (see HERA-Team/hera_cal#647). This should be formally validated before H1C IDR3 is published.

Summary

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

• Simulate uncalibrated data as in Test 2.0
• Add realistic thermal noise
• Perform abscal
• Check ratio of input gains to output gains

Simulation Details

• Freq. range: 100-200 MHz
• Channel width: realistic for H1C
• Baseline/antenna configuration: realistic for H1C
• Total integration time: 60 integrations (1 file) of 10.7 sec each should be plenty
• Number of realisations: 1

Criteria for Success

• Abscal gain amplitudes are consistent with input to better than 1%

[jsdillon]

This ultimately got addressed in Test 4.1.0 (#75) in Figures 6 and 7 of this notebook and similar notebooks that we haven't made yet: https://github.com/HERA-Team/h1c_idr3_validation_notebooks/blob/main/lst_bin_inspect/h1c_idr3_2_validation_inspect_epoch_0.ipynb

There are some deeper issues here, like figuring out when the linear abscal solver drives the amplitude to 0 and what can be done about that. It's a very small effect (sub 1%) after smooth_cal, so I'm declaring this issue resolved, but there's still a student project that could be done here if someone is interested.