Scalable positivity preserving model reduction using linear energy functions

Abstract

In this paper, we explore positivity preserving model reduction. The reduction is performed by truncating the states of the original system without balancing in the classical sense. This may result in conservatism, however, this way the physical meaning of the individual states is preserved. The reduced order models can be obtained using simple matrix operations or using distributed optimization methods. Therefore, the developed algorithms can be applied to sparse large-scale systems.