Three stage period analysis and complementary methods

We generalize the multistage period search and analysis method for the case when the data error estimates (i.e. statistical weights) are available. Ad justment of model parameters allows us to detect complex wave forms in nois y data. Complementary statistical tests are also presented to estimate the significance of the detected periodicities, to verify the modelling reliabi lity, and to separate real periodicities from spurious ones. The applicatio n of these methods is illustrated on different samples of real photometric data.