Use the methods proposed by Peter Jackel (spectral sqaure root of symmetric matrix) rather than cholesky.

Not sure about the exact definition but it's being used in similar context so that may be right.

<t>If I generate a series of future realisations of a yield curve to reprice a fix/flt IRS do I calculate the NPV based on the new curve or the current market curve. What I'm asking are should my forecast curve be based on the simulated curves and the discount curve on the current curve or both from...