How do we determine when the scheduler should change filters?

Please read the background information in the slides at

Elahe is investigating potential scheduling algorithms (note: this means, what kinds of algorithms are useful, from an operations research point of view, in developing scheduler ‘controllers’), and as part of that development, she is building prototype cost functions and evaluations of the resulting observations. Currently she is investigating ‘approximate optimal control’ solutions. The scheduling algorithms are focused on optimizing observations within a single night.

One of the questions Elahe has is how to decide whether a filter change is worthwhile or not?

The obvious cost of a filter change is the 120 second delay in observing, and perhaps this could be a weight that combines in some fashion with the slew time cost. There are some additional constraints or costs that may be desirable to include, such as preferring to change filter fewer times per night (wear and tear on the mechanisms) or fewer times per hour (the filter change time constraints from the camera team).

The benefits of filter changes would presumably be SNR depending on the phase of the moon (sky brightness in various filters) and seeing (tied into airmass, as the seeing depends on airmass and filter, as well as raw seeing at zenith) vs. the ‘best possible’ SNR for the field (minimum sky brightness and seeing/airmass), with some weighting due to amount of time the minimum vs. current SNR would be available for. I don’t know how much we have to define this in order to make it useful for Elahe, so perhaps she can weigh in with some clarifications.

@connolly has been charged with summarizing and making a final recommendation to Elahe.

I think nature provides some good spots for doing filter changes:

  • dome opens
  • twilight ends
  • moon rises/sets
  • twilight starts

A good start would be to let the scheduler change filter “for free” at those spots. Then we can see if we need more filter changes within those blocks and how to decide when to make those changes.

While we are optimizing the number of exposures and the total depth, we may want to do something like define the desired relative co-added depth between filters. So we want as deep as possible in r, and then g-r=0.5, u-r=1.5, r-i=-0.3, etc.