[Scip] why the dual solution is non-positive?

[Tobias Achterberg]

XiangyongLi wrote:
> Dear Dr. Tobias,
> In previous emails, we discussed the issue of dual solutions. As we
> discussed, the dual solution for constraints in my model should be
> non-positive. But now my SCIP code sometimes returns positive dual
> solutions for some constraints for which the dual solution theoretically
> should be non-positive. Do you know some possible reasons? Thanks.

I cannot say without knowing the details. What are the exact values (are they almost zero 
and only slightly positive or are they pretty big)? What are the constraints to which the 
dual values belong to? Are they inequalities? Of what sense?

Tobias