Learning objectives
- solve a scalar gain model, quantify phase decorrelation, and combine independent plus common-mode uncertainty.
- Propagate a statistical or calibration uncertainty and report a limitation.
- Preserve a reproducible record of inputs, transformations, and outputs.
Preparation, materials, and setup
Use Python 3 with the standard library (or a spreadsheet capable of regression and plotting), the supplied ../data/calibration_records.csv, and the relevant lecture notes. Record file name, row count, columns, units, and provenance level before editing. No facility observing time is assumed.
Dataset status: the supplied compact record is synthetic/instructor-provided (level 3), not an archived observation. Compare its analysis workflow with the real survey or observatory source cited in ../reference-log.md.
Computed check: sigma_S = 7.397 mJy; fractional uncertainty = 0.06164; sigma_S = 7.397 mJy.
Apparatus and procedure
- Estimate antenna amplitude and phase residuals from the calibrator rows; flag the injected outlier before averaging.
- Treat the target amplitude_ratio values as observed-to-model ratios for a 120-mJy teaching source. Estimate the scalar amplitude gain from the three good calibrator scans, exclude the 1.55 outlier using a documented rule, and correct each target estimate before comparing with 120 mJy.
- Calculate Gaussian phase coherence for each residual-phase rms and compare predicted versus observed amplitude ratio.
- Build a covariance-aware flux error budget with thermal, gain-amplitude, and common flux-scale terms.
- Repeat the target analysis after removing one scan at a time and report leave-one-scan-out stability.
Measurement record and uncertainty
| Record | Required entry |
|---|---|
| Input | File, release/source, units, masks or flags, and provenance level. |
| Model | Equation, parameter values, convention, and approximation. |
| Statistics | Fit covariance or counting error plus one calibration/selection/systematic term. |
| Validation | Independent recomputation, residual plot, and limiting-case check. |
Report gain estimates, outlier rule, residual statistics, corrected flux, uncertainty covariance, and one diagnostic plot.
Deliverables and criteria
- Reproducible code or spreadsheet with inputs and units documented.
- One publication-quality labeled plot and a short interpretation (400-700 words).
- Uncertainty budget that separates random, common-mode, and model terms where relevant.
- Credit requires correct method, units, interpretation, provenance, and a limitation; a copied numeric result without an audit trail earns at most half credit.
Interpretive boundary
The common flux-scale term does not average down across sources sharing the same calibrator solution. State explicitly which parts of your result are facts about this prepared teaching record and which are transferable inference methods.
References and provenance
Course teaching record: ../data/calibration_records.csv, level 3. Real facility/survey context: reference log. No external image is used.