Looking for periodicity in sampled data requires that periods (lags) of different length are represented in the sampling plan. We here present a method to assist in planning of temporal studies with sparse resources, which optimizes the number of observed time lags for a fixed amount of samples within a fixed time window given a maximum time lag of interest. The method can also optimize the temporal sampling specifically for situations where samples are at risk of being rescheduled due to otherwise unpredictable events such as weather, faulty equipment, etc. The method is based on the framework of simulated annealing in which we have defined an energy function to be minimized. We compare the calculated sampling plan with a random plan and a cyclic design and demonstrate how our calculated plan provides the most information about temporal autocorrelation.
- Sparse sampling
- Simulated annealing