Real-time computation of singular vectors (Q1126572)

An analog circuit approach for the singular vector decomposition in real-time is considered. An analog computational model is proposed as a specialized analog computer. The asynchronous parallel processing, continuous-time dynamics, and high-speed computational capabilities are the key features of the proposed model. It is shown analytically and by simulation that the proposed circuit can provide the desired singular vectors of a given matrix \(A\), with arbitrarily small error during elapsed time of only a few characteristics time constants of the circuits. In addition, the parameters of the circuit can be obtained from the given matrix without any computation. For the practical use, the case is generalized for the complex values of the matrix. Based on the obtained simulation results it is demonstrated that the proposed approach is more powerful than the known sequential techniques for the singular vector decomposition.