Fast Solution of High Stochastic Dimensional EM Problems Using Proper Orthogonal Decomposition

Electromagnetic (EM) systems are prone to parameter variations and their impact help in predictive analysis. In this letter, proper orthogonal decomposition is proposed as a fast uncertainty quantification method, which can efficiently handle a large number of stochastic parameters, without any restriction on the levels of variations. The performance of the approach is evaluated in a frequency domain stochastic EM problem employing vector finite element method and having permittivity variations. The computational time increases by only 15% even with a fivefold increase in the stochastic variables.