[Ipopt] Lagrangian multipliers

[Stefan Vigerske]

Hi,

the sign of lambda_i tells you whether the left or the right hand side 
of the constraint is active. I just don't remember at the moment which 
sign corresponds to which side.

Stefan

On 07/05/2012 08:43 PM, Andrea Cassioli wrote:
> Hi,
> I am using ipopt in a column-generation framework. At every iteration
> I need the Lagrangian multipliers to solve the pricing problem. I have
> successfully  interface ipopt with my code and it seems to work well.
> I would like to be sure that the values I get are what I expect. My
> problem has two-sided constraints like
>
> l <= g_i(x) <= u
>
> so I suppose the multipliers lambda_i is either zero, when the
> constraints are not binding, or a non zero value for the binding
> constraint. Are the multipliers in ipopt suppposed to be positive?  I
> could not find explicit documentation on this.
>
> Best,
> Andrea Cassioli, PhD.
> _______________________________________________
> Ipopt mailing list
> Ipopt at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/ipopt
>