Maximizing Kepler science return per telemetered pixel: Detailed models of the focal plane in the two-wheel era1

14DFM is responsible for most of the data-driven modeling shown in this white paper. 15DL is responsible for most of the toy data simulations and the physical modeling shown in this white paper. – 2 – 1. Executive summary (1) The primary recommendation of this white paper is image modeling, that is, fitting to the pixel values downlinked from the Kepler satellite a quantitative description of pixel sensitivities (flat-field), point-spread function (PSF), and sky, bias, and dark signals, all as a function of focal-plane position. This kind of modeling has not been required for Kepler so far because its aperture-photometry precision has been maintained through point-ing stability. The degraded pointing precision in two-wheel operations will be a blessing as well as a curse: It reduces the precision of aperture photometry, but it provides diversity that permits inference of the detector sensitivity map and PSF. We propose developing a probabilistic generative model of the Kepler pixels. We argue that this may per-mit continuance of photometry at 10-ppm-level precision. We demonstrate some baby step