Complexity reduction for parametric high dimensional models in the analysis of financial risk

Abstract

This paper presents a parametric model order reduction (pMOR) approach based on the proper orthogonal decomposition approach for financial risk analysis. pMOR requires solving the high dimensional model for some training parameters to obtain the reduced basis. We propose an adaptive greedy sampling approach based on surrogate modeling for the selection of training parameters. The developed algorithms are tested on an industrial example of a puttable steepener. The results show that the reduced model works excellent and has potential applications in historical or Monte-Carlo value at risk calculations.