problem.validatekkt

Purpose

Validates the first order optimality conditions also known as the Karush-Kuhn-Tucker (KKT) conditions versus the currect solution

Synopsis

problem.validatekkt (calculationmode, respectbasisstatus, updatemultipliers, kktviolationtarget)

Arguments

calculationmode

The calculation mode can be:

0	recalculate the reduced costs at the current solution using the current dual solution.
1	minimize the sum of KKT violations by adjusting the dual solution.
2	perform both.

respectbasisstatus

The following ways are defined to assess if a constraint is active:

0	evaluate the recalculated slack activity versus xslp_ECFTOL_R.
1	use the basis status of the slack in the linearized problem if available.
2	use both.

updatemultipliers

The calculated values can be:

0	only used to calculate the xslp_validationindex_k measure.
1	used to update the current dual solution and reduced costs.

kktviolationtarget

When calculating the best KKT multipliers, it is possible to enforce an even distribution of reduced costs violations by enforcing a bound on them.

Further information

The bounds enforced by kktviolationtarget are automatically relaxed if the desired accuracy cannot be achieved.