QuoteAnd who decides which subset we need?Users, I suppose. QuoteLapack on Windows is easy if you use Intel or AMD libsExactly! What about the VS community? In that case PPL It is part of VS2010.This is an attention point.MKL LicenseSo, when can we expect an A-Z Lapack sandbox environment?

(double)

QuoteExactly! What about the VS community? In that case PPL It is part of VS2010.AMD Core Math Library is free and includes LAPACK/BLAS and few other things.Does VS express license allow commercial distribution? MKL costs $399 and there are no royalties. It seems there is no extra charge for running MKL linked code on a cluster.Slightly off topic, but anyway... Why would anyone bother with NAG or similar libs? Does NAG offer anything that isn't found in MKL or GSL? For in-house development GPL isn't a problem.

QuoteWhy would anyone bother with NAG or similar libs? It is one of the oldest, tested and robust libraries? This could be a reason for using it.

I cannot as well publish any code of LMM but about the theory the Andersen and Piterbarg book is really helpfull I wish i had it four years ago when i did LMM. The LMM introduction is even available free:http://www.andersen-piterbarg-book.com/ ... mple.pdfAn interesting discussion before implementing the LMM would be for which products it is appropriate compared to alternative models and then set up the list of products to be handled.

QuoteOriginally posted by: CuchulainnQuoteWhy would anyone bother with NAG or similar libs? It is one of the oldest, tested and robust libraries? This could be a reason for using it.Afaik MKL doesn't really even have much optimization (just some nonlinear least squares last time I checked), while NAG has a vast array of robust algos, and are still expanding it (all the way to swarms etc.). And they should have practical parallel PRNGS by now (MKL is again limited). NAG also has sorting etc...(And: I can volunteer for a boost-conform Acklam; what else?) And for an important task you might find libs with O(n) where MKL is O(n log n).Btw what do we want to use for optimization? Interfacing to GAMS would be nice

QuoteOriginally posted by: quartzwhile NAG has a vast array of robust algoswhich they used to provide in source codes. theose were the days, my friends, i thought they'd never end...