[Ipopt] ipopt run message

Andreas Waechter awaechter.iems at gmail.com
Fri Sep 2 16:09:59 EDT 2011


Hi,

>
> Does anybody know what the following messages mean:
>
>
> In iteration 1766, 1 Slack too small, adjusting variable bound
> In iteration 1937, 1 Slack too small, adjusting variable bound

This message means that a slack variable (for a variable bound or an 
inequality constraint) has become very small, too small within machine 
precision, and to make sure that the interior point algorithm can 
continue with a non-zero slack (and the logarithm of this slack variable 
can still be computed), Ipopt internally makes a small modification of 
the problem statement, relaxing the corresponding bound a little bit.

> MA27BD returned iflag=-4 and requires more memory.
>  Increase liw from 363145 to 3631450 and la from 1139815 to 11411860 
> and factorize again.

This means that the linear system that is solved in every iteration is 
becoming ill-conditioned and the linear solver MA27 produces more 
fill-in, and therefore require more memory.  In your case, it is more 
memory than you have available on your machine.  You might be able to go 
past this barrier by choosing a smaller number for the 
ma27_meminc_factor option which determines by how much the memory 
allocation should be increased when MA27 runs out of memory (default is 
10, so try 2 or so).

Andreas

>
> _______________________________________________
> 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/20110902/a3585f2a/attachment.html 


More information about the Ipopt mailing list