QuantLib goals

[Ferdinando M. Ametrano-2]

Hi everybody,

welcome to all the new subscribers.

I would like to receive feedback about the goals selected for QuantLib 1.0
I want to strip the goals down to an acceptable minimum in order to have as
soon as possible a release that can be used by end-user.

Let's start with few selected issues.

1 Platforms to be supported (OS + compiler)
We already support Windows32 + Borland 5.5 and MacOS + CodeWarrior.
Bernd is working on GNU/Linux + gcc (anyone willing to help?)
Windows 32 + Visual Studio is supported, but the resulting lib doesn't work
correctly: the american_with_dividends.py test fails. Any help on this area
is welcome since Visual Studio is almost a de facto standard in Win32

2 Executable implementations
This is very important since to have many executable implementation will
enlarge the user base.
2.1 porting to the Microsoft application world (VB and Excel): to do (as
COM or Excel add-in)
2.2 Matlab extension: to do
2.3 Python module: done

3  Generic Tools
3.1 date/time module: half done. Luigi is working to a date schedule class.
This should go hand in hand with a payment schedule class
3.2 one-dimensional solver: done
3.3 one-dimensional optimizer: to do. This may use the interface of the
one-dimensional solver
3.4 multi-dimensional solver and optimizer: to do
3.5 PDE module: done
3.6 statistical module: done
3.7 Montecarlo module: Marco is working on that

4 Financial Tools
It is very important to receive as soon as possible feedback on the
Instrument interface (Include/instrument.h), since that is the base class.
We designed a tentative Stock financial instrument
(Include\Instruments\stock.h).
We will propose shortly deposit, FRA, futures, swaps financial instrument
interface

Any takers?
Comments?

ciao -- Nando

[Marco Marchioro-2]

Howdy, Pards,
        in the on-going effort to have--as soon as possible--a Montecarlo engine,
I have just submitted a simple random number generator (RNG), namely
QuantLib::Math::RandomGenerator, which is going to be followed by others.
Actually, everyone is welcome to give his, more-or-less sophisticated, RNG
with the same interface as the one given.
Furthermore, two template classes, QuantLib::Math::BoxMuller and
QuantLib::Math::CLGaussian,
are added to transform a uniform deviate into a Gaussian.
As the policy goes, a python interface and a simple test program (written,
of course, in python), have been provided.

        best of my regards,
                        MarMar