[Quantlib-dev] Fwd: new MonteCarlo::PathGenerator2

[Ferdinando A. <na...@qu...>]

Hi all

I got no feedback to the message below. I'm going to assume that I can 
remove the antithetic calculation from all PathPricer in favour of the new 
MonteCarlo PathGenerator2.
Should I also assume that I can remove the old PathGenerator?

There will be a severe hit on performances, that will be solved in the 
medium term using the visitor pattern to make PathGenerator2 smarter.

ciao -- Nando

>Date: Thu, 06 Mar 2003 16:29:18 +0100
>To: QuantLib-dev <qua...@li...>
>From: Ferdinando Ametrano <na...@qu...>
>Subject: new MonteCarlo::PathGenerator2
>
>Hi all
>
>I've just committed a new path generator MonteCarlo::PathGenerator2. I 
>named it with the trailing 2 in the hope to allow it to live together with 
>the old MonteCarlo::PathGenerator.
>
>The new one is based on Sad's TimeGrid class to handle the time 
>discretization and Sad's DiffusionProcess class to actually generate 
>paths. Since the Black Scholes DiffusionProcess is now time/asset 
>dependant the new PathGenerator is also time/asset dependant.
>
>Unfortunately when the diffusion process is asset dependant we cannot 
>calculate the antithetic path in the path pricer anymore, since the drift 
>component is not common between a path and its antithetic one anymore. The 
>antithetic path must be calculated by the PathGenerator itself.
>
>This means that all the current PathPricers are wrong when the drift 
>process is asset dependant, and I'm going to remove the antithetic 
>calculation from all PathPricer in favour of the new MonteCarlo Pricing Engine.
>
>Then of course, I would also like to remove the old PathGenerator.
>
>I look forward to your feedback
>
>ciao -- Nando
>