This wiki service has now been shut down and archived
NGSS Aspen Report
Notes from Aspen workshop "Wide Fast Deep Surveys"
Aspen Center For Physics, USA, June 15-July 3rd
This was a three week workshop held in Aspen, in the classic relaxed Aspen style. On Mondays, Wednesdays and Fridays there were typically two plenary talks in the morning, and one breakout discussion in the afternoon. Tuesdays and Thursdays were left completely free for informal discussions. The workshop featured mostly US scientists, but with a handful of UK and other astronomers. It was centred around LSST and its prospects, but also featured updates from other current and upcoming projects, especially PanSTARRS. Presentations had a mixture of scientific discussion, project updates, and technical discussions.
All the plenary talks are the meeting wiki site. This also has some notes on the breakout discussions.
There were project updates from PanSTARRS-1, DES, WISE, SWIFT, Palomar Transient Factory (PTF), RAPTOR, SkyMapper, Euclid, BOSS and BigBOSS. I kept brief notes on each of these and in due course will add short desscriptions and links for each of these to the NGSS wiki.
There were scientific topic breakouts on asteroids and other solar system bodies, on supernovae, and on other astronomical transients.
Several talks and discussions concerned the accuracy of photometric redshifts. Jeff Newman gave a rather worrying talk about how the outliers could kill us even if sigma-z is good, but also presented interesting new techniques for improving photo-z's. There were also discussions comparing the efficacy of spectroscopic redshift surveys vs photometric surveys. Nikhil Padmanabhan claimed that to get the same accuracy on cosmological parameters, photometric surveys needed five times the area of a spectroscopic survey to the same apparent magnitude. There were also talks and discussions about how we need to take a much more serious attitude to photometric calibration, and aim to achieve millimag accuracy at least for colours. A presentation by Chris Stubbs was extremely impressive in this regard.
There were some more technical presentations and discussions, including a talk by Josh Bloom on applying machine learning techniques to astronomical databases, and a discussion led by Ryan Scranton on whether future databases should use Map Reduce instead of standard technology, or alternatively the array-based approach favoured by the ~SciDB initiative. It looks like different kinds of scientific analysis may be best served by different technologies. The strength of the Map Reduce seems to be that it is easier to write scripts to solve problems. This whole issue, of database technology and getting science out of very large databases, could be appropriate for a specific NGSS workshop.
There was also a discussion session led by me, on what the user interface should look like in the future, and related issues about getting science out of big databases. This has some reasonably extensive notes, which are here.
Finally, there was also an interesting discussion on whether the sociology of astronomy is changing because of big surveys, and how we can create proper careers for people who are dedicated survey scientists.