November update: part 2

This update is to highlight some of the work going on with the Feature Based Scheduler (@yoachim likes to call it SLAIR, but mostly we acronym it FBS).

You’ve most likely heard of this scheduler from various presentations over the last year or so. At the moment, we’re using it to create a limited set of runs, but primarily we’re finishing up a code refactor to make it easier to use and extend, as well as adding important things like unit tests and documentation strings. The next goal is to create a ‘library’ documenting the effects of varying the strength of various basis functions (aka things like “what happens if we increase the weight on the slew costs”, “what happens if we change the weight on the number of visits”, “what happens if we change the weighting on the image depth”, etc.) that will help us understand the basics of how things are working together, and help us be more efficient when creating the next sets of opsim runs based on the white paper requests.

As I mentioned though, there are a limited set of new FBS based simulations. These have not gone through change control - and these representations of the simulations are subject to change (i.e. don’t take this “baseline style simulation” as an actual baseline yet). However, we know there are many of you who are interested in how these runs would look and so here’s some examples; we would appreciate any comments or insights you have on these runs and we definitely would like to hear if you uncover something that looks like a `misfeature’ (aka bug).

A set of FBS runs is available at


So pontus_2579 and pontus_2580 are labeled as rolling and clearly aren’t.

Sadly it turns out this is correct, and they are not rolling – they have been removed as not additionally informative beyond the pairs in same vs different filters sims that remain.

The best rolling cadence options at present are still the feature_rolling_* databases that @yoachim has linked previously:
The databases can be downloaded from these links: