The question
How do you build a measurement whose uncertainty you can defend, and how do you test a model against the data instead of against expectation?
What I work on
- Instrument response and calibration
- Noise, covariance, and error propagation
- Computational models with tests
- Experiments that can be repeated by another person
Current work
The TESS timing project is the first formal measurement: a detector model, an aperture, a noise budget, and a fit with a covariance.
The method work carries over to laboratory measurements: characterize the sensor, record the raw signal, and keep the calibration with the data.
Methods and standards
- SI units and explicit conversions
- Calibration records kept with the data
- Covariance and sensitivity tests
- Injection or recovery tests where they apply
Directions
- Detector and sensor characterization
- Low-noise readout and laboratory electronics
- Computational physics with reproducible test cases
- Small experiments designed around a stated uncertainty
All research areas. Notes and records are in the notes archive and the data and code index.