Re: [Quantlib-users] Re: FdDividendAmericanOption

[Luigi B. <lui...@gm...>]

On 12/30/04 09:06:01, Joseph Wang wrote:
>=20
> The one thing that seems to pop out is the BSM operator which takes as
> input the log of the gridSpacing_.  Knowing nothing about the code, it =20
> would seem to me that the BSMOperator is expecting a linear grid, and =20
> putting in a logarithmic grid would appear to do bad things.

Joseph,
	the BSM operator discretizes the equation for x =3D log(S), therefore =20
a logarithmic grid in S is used which corresponds to a linear grid in x.

Also, I very much appreciate that you dived in to get data, as =20
unfortunately I don't have the time to dive in myself now---thanks a lot.

If I can suggest a line of investigation, I find interesting that, as you =20
said,

> If I run the code with zero dividend, I still get odd values.  I don't =20
> get any odd values if I just run FdAmericanOption.

You might get some insight by comparing the two calculations. Are the two =20
grids the same when the two classes are passed the same arguments? Are the =
=20
calculated values the same until the first dividend payment is reached? =20
What happens to the asset values when the code for applying dividends is =20
triggered, even with zero dividends?

Later,
	Luigi