[Ipopt] (no subject)

[Damien]

Most of the linear solvers aren't thread safe or rentrant.  You can only 
run one copy of the linear solver per memory space (i.e. per process).  
Intel Pardiso, MUMPS, MA27 all have this problem, they use SAVE 
statements internally and multiple calls in one process overwrites those 
memory areas and they crash or give weird results. It's not specifically 
an IPOPT problem.

Damien

On 2015-07-29 11:41 AM, Gabriel Hackebeil wrote:
>> I understand that IPOPT is currently not thread safe
>
> I was under the impression it was thread safe 
> (https://projects.coin-or.org/Ipopt/ticket/167). Is this no longer true?
>
> Gabe
>
>> On Jul 29, 2015, at 1:03 PM, jmogali at andrew.cmu.edu 
>> <mailto:jmogali at andrew.cmu.edu> wrote:
>>
>> Hi,
>>        I am trying create a distributed system of agents where each agent
>> solves a convex optimization problem in Java on a single machine.
>> I understand that IPOPT is currently not thread safe, is there a
>> way I can still create a distributed system on a single machine
>> using IPOPT ? Instead of java threads, if I use different
>> processes for each of my agents will IPOPT work correctly ,
>> assuming I do interprocess communication? Any other solution
>> suggestions is also welcome.
>>
>> Thanks,
>> Jayanth
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org <mailto:Ipopt at list.coin-or.org>
>> http://list.coin-or.org/mailman/listinfo/ipopt
>
>
>
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20150729/c8ad0058/attachment.html>