multiAssetOptions (version 0.1-2)

matrixFDM: Finite Difference Matrix Generator

Description

Generates a modified coefficient matrix (M-matrix) used in the finite difference method from the option inputs. See Tavella and Randall (2000) for more on the standard formulation of the M-matrix.

Usage

matrixFDM(S, rf, q, vol, rho)

Arguments

S

list containing the vectors of spatial grid points associated with each underlying. Vector sizes of underlying spatial grid points need not be equal.

rf

scalar; applicable risk-free rate (domestic risk-free rate).

q

vector; holding costs of the option's underlyings (dividends, foreign risk-free rates, etc.).

vol

vector; volatilities of the option's underlyings.

rho

matrix; correlation matrix of the option's underlyings.

Value

matrixFDM returns a CsparseMatrix-class matrix used for timestepping in the finite difference method.

Details

matrixFDM first constructs the non-zero diagonals of the M-matrix and stores them as columns. The bandSparse function from the Matrix package then constructs a sparse banded matrix from the columns of the previously contructed matrix. Spatial domain boundaries are calculated first-order inwards with second difference terms dropped, maintaining block tridiagonality.

References

Tavella, D., Randall, C., 2000. Pricing Financial Instruments: The Finite Difference Method. John Wiley & Sons, Inc., New York.

Examples

# NOT RUN {
# finite difference matrix for uniformly-spaced two-asset option
S1 <- list(seq(0, 5, by=1), seq(0, 5, by=1))
rf <- 0.1
q <- c(0.05, 0.04)
vol <- c(0.20, 0.25)
rho <- matrix(c(1,-0.5,-0.5,1), 2, 2)
matrixFDM(S1, rf, q, vol, rho)
# }