Optimal Sampling

One of the major advantages of a robotic telescope network is the ability to request individual observations from geographically distinct telescopes within the network, at particular times. This provides an observer with much greater flexibility with regard to the placement of observations, in both time and space. For example, the opportunity arises for sparse temporal placement of observations when period searching, something impossible in the classical 'monolithic observing' paradigm.

However, how best to place a limited number of observations to cover the dynamic range of frequencies required by an observer is a non-trivial problem. Eric Saunders has been working on this problem as part of his PhD thesis work, and has recently submitted a paper detailing the work to Astronomy & Astrophysics (Full abstract). Our main conclusions are:

• Observations geoemetrically spaced in time can substantially improve signal detection quality, minimising the aliasing effects that arise from sparse sampling.

• The choice of geometric base is critical to the success of this strategy - a poor choice of geometric base can significantly hamper the effectiveness of the strategy.

• An optimal sampling may be reordered, as long as the distribution of separations is preserved, with almost no loss of quality. This allows an observing schedule to retain significant flexibility in the face of scheduling constraints.

Work has now begun to code an autonomous agent that implements the optimal sampling technique described in Saunders et al. 2006. This agent calculates the optimal gap spacing for a particular variable star observing project, based on the astronomer-defined maximum and minimum periods of interest. The difficult part is to map this optimal spacing to the real-world environment of an operational telescope network such as Robonet-1.0. Complications arise because a telescope may or may not make an observation requested by the user agent, for any number of reasons, which are not generally knowable by the user.