The Functional Models of Purposeful Natural Phenomena Sign Search in the Covariance Matrix Eigenvectors of Time Series

The eigenvector method applied to time series analysis allows to discover the non-correlated time series' components. The capacity of which components may differ in wide area. These components are not mutually depending and may be researched using the other approaches (i.e. spectral analysis).

Nowadays a personal computer in enough to compute a 1000x1000 covariance matrix for the time series which have about 106 samples in few hours. But the problem is to connect the eigenvectors you discover with the theoretical base of the area which is formulated in another language. In order to eliminate this breach we offer to use the simulation of the time series according to the theoretical model. The eigenvector set you get in result of such simulation is considered as a sign set which is the set of signs that the time series you observe fit to the theory.

The effective applied research according to this approach may be conducted by the interdisciplinary research groups which need the appropriate organizational and technical support. We offer to build the activity of such groups upon the approaches of the purposeful behavior which should be described using SADT functional modeling and using a program system based upon these models.