At Northrop Grumman today, Justin Crepp (Notre Dame) talked about future extreme high precision radial-velocity spectrographs that could capitalize on adaptive optics to make them more precise, more stable, and smaller. He has a demo concept working and this could be the future! Rémi Soummer (STScI) talked about coronograph concepts for WFIRST and future missions. Soummer's talk had a huge impact on me: He showed that a coronograph actually brings the light to three successive foci, with a non-trivial mask or stop at each focus. This translates into three two-dimensional functions, each of which needs to be optimized, and possible adaptively (when there are incoming wavefronts that are not perfectly flat). This is a great set of problems, in optics and control. I fantasized about having a home-built code that does all the relevant electromagnetic propagation calculations. One simple and take-home rule of thumb from Soummer's talk: A one-picometer distortion to the wavefront at the telescope entrance aperture will make an Earth-brightness speckle for the mission concepts we care about!
Jon Arenberg (NG) pointed out that, from an engineering perspective, we should be thinking in terms not of a linear design process flowing from objectives to final data products, but an interactive feedback loop involving scientific objectives, engineering capabilities, both scientific and technological research, and flexible redesign, ideally as late into the mission process as possible, with in-mission servicing an additional possible bonus.
After all the talks were done, we talked about wish lists and important ideas, to start a longer conversation between the astronomical community and NG. I said that what I want (in the ideal world) is a platform in space where we can build, test, prototype, and fail fast. Right now (mirroring Arenberg's points), space-mission planning and execution is a very long, slow process, with no feedback. If we could work in space, we could consider all sorts of things in situ, and not have to build things at one gee and then deploy them at zero. Next decadal survey: Can we talk about some out-of-the-box ideas, that might be expensive, but might have many enthusiastic constituents?