The data may be left unweighted or be weighted by the variance array. In general it is best to weight the data if there are many counts per time bin (> 10 ?) but otherwise the data should be left unweighted. It is worth using both methods to test if features in the resultant periodogram are real.
The data is folded over a number of periods and the chi-squared fit to a constant determined for each period. The period giving the highest chi-squared is deemed to be the best and the data is folded at this period and written into an output file. A further file comprising chi-squared v period is also produced.
Each phase bin in the folded file has an associated variance. This is given by :
Unweighted: Var(bin) = SUM (( Val(i) - mean(bin) ) ** 2 / (N*(N-1)))Where Val(i) is the value of a given data sample, mean(bin) is the average value of data samples contributing to this phase bin and N is the number of data samples in the phase bin.
Weighted: Var(bin) = 1.0 / SUM ( 1.0 / Var(i) )Where Var(i) is the variance of a data sample contributing to this phase bin.
Note: In both cases the variance of a phase bin is inversely proportional to the number of data samples in the bin.