Rory Holmes (MPIA) showed up today for two weeks of coding and writing. We are asking—for Euclid and in general—how one should design an imaging strategy such that the data will be possible to self-calibrate or uber-calibrate. That is, how should you set the pointings so as to get good field coverage and good measurements (internally) of your system variations with space and time.
The big picture, for me, is this: Astronomers think in terms of taking science data and separately calibration data. The former is used for science and the latter is used to set instrument parameters like flat-field and bias and so-on. But typically you have far more photons in your science data; aren't they incredibly constraining on calibration themselves? Indeed, in the SDSS, we got much more calibration information out of the science data than the calibration data. But, of course, we had to adjust the SDSS imaging strategy to make it work: You need to have good redundancy in the data stream, and redundancy of a very specific kind. That's what we are setting out these two weeks.