|
Hello Ian,
Examples can be found in "unsupported/doc/examples/". You may
build them via
make eigen_ad_examples
At the moment, the only supported AD tool is dco/c++. You can
download the latest version from
https://www.nag.co.uk/content/downloads-dco-c-dcl6i34ngl
Please note that a license key is required to run dco/c++; you
can request a 30 day software trial there as well.
Best wishes,
Patrick
On 19.12.19 17:52, Ian Bell wrote:
This is exciting stuff! Do you have some
demos/examples in your fork? I wasn't able to find any after a
very preliminary search.
Hello,
The Eigen-AD repository can now be accessed here:
https://gitlab.stce.rwth-aachen.de/stce/eigen-ad
Best regards,
Patrick Peltzer, Johannes Lotz, Uwe Naumann
Informatik 12 (STCE), RWTH Aachen University, Germany
On 03.12.19 13:56, Patrick Peltzer wrote:
> Dear all,
>
> We would like to draw your attention to a new fork of
Eigen with
> support for Algorithmic Differentiation (AD) in the
context of AD by
> overloading in C++. Important new features include highly
efficient
> first and higher derivatives of Eigen solvers as well as
compatibility
> with a wide range of AD software tools. See
>
> https://arxiv.org/abs/1911.12604
>
> for details including benchmark results. Please contact
> info@xxxxxxxxxxxxxxxxxxx
for access to the software.
>
> We look forward to further discussions with potential
users and other
> AD software tool developers.
>
>
> With kind regards
>
> Patrick Peltzer, Johannes Lotz, Uwe Naumann
> Informatik 12 (STCE), RWTH Aachen University, Germany
|