[Coin-discuss] FLOPC++

[Brady Hunsaker]

I finally had some time to experiment with FLOPC++, Tim Hultberg's
modeling system based on C++ code.  I like it.  If you have not taken
a look, I recommend you do so.  See the message from Tim below.

I did have to modify the Makefile a bit (as usual) to get things
working on my system.  Also, there is not yet good documentation,
though Tim has made quite a few helpful examples.

Tim, as you indicate on the webpage, what appears to be the most
powerful feature is that although the syntax feels like a modeling
environment, you can run arbitrary C++ code to generate problems,
analyze the results, or run multiple problems.

I would say that documentation is the most pressing current need.
Writing it will probably bring up other possible improvements.  For
example, I had difficulty figuring out how to have a constraint
indexed over two elements of the same set.  What eventually seemed to
work was using an index for the second appearance of the set in the
declaration of the constraint as well as in its definition (I had
thought at first that the index would only be necessary in the
definition, but this alone didn't seem to work).  

I am certainly in favor of making this part of COIN-OR.  Based on
previous emails, it looks like JP Fasano might have been in touch with
you about the process, though perhaps the intent is to wait for
incorporation and the repository transfer to INFORMS?  

Brady

p.s. I obviously missed my two week goal for that OsiGlpk bug, but I
hope to look at it this weekend.

On Wed, Mar 19, 2003 at 06:28:35PM +0000, Tim Hultberg wrote:

> FLOPC++ is now available for downloading at its new built homepage: 
> 
>               http://www.mat.ua.pt/thh/flopc/
> 
> I still hope that FLOPC++ will become a part of COIN when INFORMS becomes the 
> host. In the meanwhile I would be happy to receive comments, bug reports and 
> other feed back.
> 
> Best regards,
>                           Tim Hultberg
>