Efficient solution of a partial integro-differential equation in finance

被引:66
|
作者
Sachs, E. W. [1 ,2 ]
Strauss, A. K. [3 ]
机构
[1] Univ Trier, Fachbereich 4, Abt Math, D-54286 Trier, Germany
[2] Virginia Tech, Interdisciplinary Ctr Appl Math, Blacksburg, VA 24061 USA
[3] Univ Lancaster, Dept Management Sci, Lancaster LA1 4YX, England
关键词
Levy process; Partial integro-differential equations; Conjugate Gradient method; Toeplitz matrices;
D O I
10.1016/j.apnum.2007.11.002
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Jump-diffusion models for the pricing of derivatives lead under certain assumptions to partial integro-differential equations (PIDE). Such a PIDE typically involves a convection term and a non-local integral. We transform the PIDE to eliminate the convection term, discretize it implicitly, and use finite differences oil a uniform grid. The resulting dense linear system exhibits so much structure that it can be solved very efficiently by a circulant preconditioned conjugate gradient method. Therefore, this fully implicit scheme requires only oil the order of O(n log n) operations. Second order accuracy is obtained numerically on the whole computational domain for Merton's model. Published by Elsevier B.V. on behalf of IMACS.
引用
收藏
页码:1687 / 1703
页数:17
相关论文
共 50 条