Comparison of Sobol' sequences in financial applications

Shin Harase
DOI: https://doi.org/10.1515/mcma-2019-2029
2019-02-05
Abstract:Sobol' sequences are widely used for quasi-Monte Carlo methods that arise in financial applications. Sobol' sequences have parameter values called direction numbers, which are freely chosen by the user, so there are several implementations of Sobol' sequence generators. The aim of this paper is to provide a comparative study of (non-commercial) high-dimensional Sobol' sequences by calculating financial models. Additionally, we implement the Niederreiter sequence (in base 2) with a slight modification, that is, we reorder the rows of the generating matrices, and analyze and compare it with the Sobol' sequences.
Numerical Analysis,Computational Engineering, Finance, and Science,Computation
What problem does this paper attempt to address?