An inverse evolution method for studying nonlinear dynamics of an object is suggested. Its main process is to deduce a probability differential (or partial differential) equation to the object using known solution from non-differential equation, therefore the solution domain or space would be extended, in other words, those lost solutions are restored. Meanwhile its application to Fabry-Perot (F-P) interferometry is discussed in detail (also suitable to Mechelson interferometry). A new parameter definition, opaqueness of F-P, is recommended, then one can deduce the dynamic property of F-P and get other methods for measuring the fineness and optic phase of F-P. It is worth to note that not only an extended solution space is contained in the differential equation but also one can further in phenomenon infer it according to the known typical nonlinear evolution equations or current experience and knowledge, hence the representation for object would be more and more close to its true state.