[Ipopt] Rép : Unpredictable results with mac OS X Lion vecLib

[Sylvain Auliac]

We have written a patch to correct the IpBlas.cpp file, it seems to work now with Lion's vecLib. I can send it to whoever wants it.


Le 13 mars 2012 à 13:50, Ray Zimmerman a écrit :

> I mentioned that Ipopt does not work with vecLib on Lion at https://projects.coin-or.org/Ipopt/wiki/Ipopt_on_Mac_OS_X. My solution was to use Goto BLAS, which performed almost as well in my tests on Snow Leopard.
> 
> -- 
> Ray Zimmerman
> Senior Research Associate
> 419A Warren Hall, Cornell University, Ithaca, NY 14853
> phone: (607) 255-9645
> 
> 
> 
> 
> On Mar 12, 2012, at 9:01 AM, Sylvain Auliac wrote:
> 
>> Hi,
>> Professor Hecht and I encountered a strange (and bad) behaviour of ipopt while trying to give it an interface in the FEM solver freefem++ developed in our laboratory. These were mainly memory issues. Mr Hecht later found that it was due to some incompatibilities in the file : Ipopt-3.10.2/Ipopt/src/LinAlg/IpBlas.cpp. When used with mac OS X Lion's vecLib BLAS, some of the functions are passed values which lead to unpredictable results (as said in the following page :  http://www.mathkeisan.com/usersguide/man/dcopy.html).
>> I don't know if someone already reported this, but having found nothing about that while doing a quick search, I thought it would be a good thing to do it. 
>> Best regards.
>> 
>> 
>> Sylvain Auliac
>> Laboratoire Jacques-Louis Lions
>> Université Pierre et Marie Curie
>> mail: auliac at ann.jussieu.fr
>> tel : 01 44 27 71 70
>> 
>> 
>> 
>> 
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/ipopt
> 

Sylvain Auliac
Laboratoire Jacques-Louis Lions
Université Pierre et Marie Curie
mail: auliac at ann.jussieu.fr
tel : 01 44 27 71 70



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20120313/d89d2849/attachment-0001.html>