Re: [eigen] high precision arithmetic matrix/vector computation

[Pavel Holoborodko <pavel@xxxxxxxxxxxxxxx>]

Hi Kevin,

I am the author of the toolbox and actually I am involved in enabling Eigen with arbitrary precision computations.

Toolbox relies on Eigen for precision more than 34 digits.

Could you provide examples of bugs you have found? Platform? 

We have many test cases and current version completely passes them.

Your input would be very valuable, it will helps us to improve the toolbox.

I would appreciate this very much. 

Pavel. 

P.S.

You can use my private email for communication regarding toolbox if you want: pavel@xxxxxxxxxxxx   

On Mon, May 27, 2013 at 12:07 PM, bayer00 <bayer00@xxxxxxx> wrote:

hello there,

I am new in C++ and eigen. But it is really helpful and powerful.

Will eigen support higher precision arithmetic computation in the future?

How to implement such arithmetic computation via eigen on Windows platform?

I tried a multiple precision toolbox these days  http://www.advanpix.com/ which uses Eigen as core library for matrix computations, but there are many bugs found for longer than machine precision calculation: results cannot be verified via the big M's mathematica/Maple softwares.

Thanks 

Kevin