Neil P Firth <
[hidden email]> wrote:
>does anyone have a suggestion for a straightforward enhancement that would
>take 10 hours coding, say.
not sure about the 10 hours constraint, but here's a few suggestions:
- make unit test to work under Borland command line compiler
- add Gauss-Kronrod unit test
- add single factor calibration test (see Bermudan example)
- add bermudan (single factor) pricing test (see Bermudan example)
- add lookback (and exotic in general) Monte Carlo path pricers
- fix cubic spline: we have the NR algo that would like to replace with
Nicolas' algo. Both of them are in the repository. What we need is a
general way to apply boundary conditions (fist or second derivative) in
Nicolas' algo, as we do for Finite Difference tridiagonal system.
- generalize asian (and path dependant in general) path pricer to handle
the case the option already has fixings/accruals/etc
- implement future convexity adjustment for yield curve bootstrapping
(search wilmott forum for guidelines)
- implement bivariate and multivariate, translating from FORTRAN reference
implementation (see Genz work
http://www.math.wsu.edu/faculty/genz/homepage, research and TVPACK software)
as for contributing code to QuantLib: please do it on the mailing list to
avoid Luigi hoarding contributions ;-)
------------
ciao -- Nando
Locked
