High performance algorithms for lattice-based derivative pricing models

In recent years, there have been tremendous advances in mathematical modeling of derivative securities, many of which are computational intensive. On the other hand, there have also been tremendous advances in computing technologies with rapidly improved computing performance. However, little research has been done on the impact of high performance computers on the design of efficient algorithms for financial modeling. The authors show that careful design of computer algorithms will make full use of the potential computing power of modern high performance computers. In particular, most computers have memory hierarchies to improve the performance of memory accesses. They show that a good design of data structures for lattice based algorithms can significantly improve the performance of the algorithm