[Ipopt] New Ipopt release 3.9.1

[Jelmer Ypma]

Dear Ipopt users,

This e-mail is a follow-up to Andreas' new release notes with some
more information about the R interface to Ipopt.

R is a free software environment and programming/scripting language
for data manipulation, calculation and graphics, which is used mainly
for statistical computing. It can be downloaded from
http://www.r-project.org/

A tutorial on installing and using the R interface to Ipopt can be found at
https://projects.coin-or.org/Ipopt/browser/stable/3.9/Ipopt/contrib/RInterface/inst/doc/ipoptr.pdf?format=raw
This document contains information on
- how to define objective and constraint functions in R to be solved by Ipopt
- how to define and visualize sparse matrices for the Jacobian and the
(optional) Hessian
- how to set Ipopt options from R
- how to use additional data in the objective function and constraints

If you have a previous version of Ipopt installed and want to install
the R interface without upgrading to the latest version, instructions
on how to install and download the R interface are at
http://www.ucl.ac.uk/~uctpjyy/ipoptr.html
https://r-forge.r-project.org/projects/ipoptr/

If you run into problems, have questions or would like to request
additional features, etc, please let me know.

Best wishes,
Jelmer


On Sat, Nov 27, 2010 at 22:52, Andreas Waechter <andreasw at watson.ibm.com> wrote:
> Dear Ipopt users,
>
> I created a new Ipopt release (3.9.1).  The changes include the following:
>
> - Switch to the new COIN BuildTools system.  The location of the libraries
>   and header files in the install directories has slightly changed, and
>   we now support pkg-config for the COIN projects.  This new setup has
>   been tested for a while, but in case you run into problems, please email
>   the mailing list.
>
> - Improved quasi-Newton approximation for the restoration phase.  The
>   previous versions were not working well in case the Hessian matrix was
>   approximated and the algorithm switched to the restoration phase.  In
>   case you are using Ipopt with Hessian approximation, these recent
>   changes might improve the performance in your setting.
>
> - Interface to the R Project for Statistical Computing.  Jelmer Ypma
>   contributed interface code for the R language (see
>   Ipopt/contrib/RInterface).  Thanks, Jelmer!
>
> - Intermediate callback functionality now available from C and Fortran
>   interfaces.  The C and Fortran examples show you how to use it.
>
>
> Regards,
>
> Andreas
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>