Re: [boost] Automatic differentiation (and complex numbers) with lazy component evaluation

Subject: Re: [boost] Automatic differentiation (and complex numbers) with lazy component evaluation
From: Kim Kuen Tang (kuentang_at_[hidden])
Date: 2010-04-26 14:13:44

Hi folks,

DE schrieb:
> on 26.04.2010 at 20:27
>  Hal Finkel wrote :
> 
> hi there
> i do like the idea of algorithmic differentiation very much and
> personally i would be very happy to have such a lib

As DE already stated it is a needed library in the real world 
programming. Contribution in this direction is welcome.

> 
> unfortunately i have no time to see your implementation or make any
> suggestions so this whole message is actually useless
> 
> but at least i can manifest my personal interest in such a library
> 
> and i think that implementing ET aside from proto is fine since it
> does not introduce additional dependencies (i.e. if i want to use a
> subset of boost collection)
> however more advanced boost developers might disagree with my point

Programming the AD is like designing a DSEL and this is what proto was 
designed for. There is a learning curve before using proto but since you 
are able to use Expression template this should not be a problem for 
you. The results using proto are also amazing. Take a look at spirit and 
the upcoming version of phoenix v3.

good luck.
i will be happy to see your implementation.

> 
> so good luck
> as far as i have time i'll try to join you
>