Phase dispersion minimization (PDM) is a time series analysis mathematical technique used astronomy for finding periodically repeating components of a signal, especially measurements made over time such as those of light curves. PDM is a particular method used in data folding, a means of dealing with time-series data in which each measurement has some source(s) of randomness that obscures any pattern (such as variations in the instrument's sensitivity).
The Fourier transform can also be used for finding repeating components but it is best suited to signals collected periodically rather than with varying time gaps.
Data folding consists of trying out a possible period by mapping the data from an absolute time to phases of the proposed period, then evaluating it for plausibility. PDM is a formula to assign a score to such a trial, so candidate periods can be compared. PDM also allows the such a candidate period to be compared with the presumption that the signal is random and has no period.