Package de.labathome.trustconstr.records

Class State

java.lang.Object
de.labathome.trustconstr.records.State
Direct Known Subclasses:
StateIP
public class State extends Object
Mutable state threaded through the equality-path SQP outer loop. Updated once per iteration by MinimizeTrustConstr.updateState(de.labathome.trustconstr.records.State, de.labathome.trustconstr.matrix.Matrix, boolean, de.labathome.trustconstr.ScalarFunction, de.labathome.trustconstr.records.PreparedConstraint[], long, double, double, de.labathome.trustconstr.records.CGInfo) and consulted by the user-supplied IterationCallback.

  • Field Details

    • nIter

      public int nIter
      Number of completed outer iterations.

    • numEval

      public int numEval
      Number of objective evaluations so far.

    • numGradientEval

      public int numGradientEval
      Number of gradient evaluations so far.

    • numHessianEval

      public int numHessianEval
      Number of Hessian evaluations so far.

    • numConstraintEval

      public int[] numConstraintEval
      Per-constraint count of constraint evaluations.

    • numConstraintJacobianEval

      public int[] numConstraintJacobianEval
      Per-constraint count of constraint Jacobian evaluations.

    • numConstraintHessianEval

      public int[] numConstraintHessianEval
      Per-constraint count of constraint Hessian evaluations.

    • executionTime

      public long executionTime
      Wall-clock time elapsed since the start of the run, in nanoseconds.

    • trustRadius

      public double trustRadius
      Current trust-region radius.

    • constraintPenalty

      public double constraintPenalty
      Current merit-function penalty (the rho in f(x) + rho ||c(x)||).

    • cgNIter

      public int cgNIter
      Number of CG iterations performed in the most recent inner solve.

    • cgStopCond

      public PCGStoppingCondition cgStopCond
      Why the most recent CG inner solve terminated.

    • x

      public Matrix x
      Current iterate (n x 1).

    • fun

      public double fun
      Objective value f(x) at the current iterate.

    • grad

      public Matrix grad
      Gradient gradf(x) at the current iterate.

    • v

      public Matrix[] v
      Lagrange multipliers, one entry per constraint source.

    • constr

      public Matrix[] constr
      Constraint residual vectors, one entry per constraint source.

    • jac

      public Matrix[] jac
      Constraint Jacobians, one entry per constraint source.

    • lagrangianGrad

      public Matrix lagrangianGrad
      Lagrangian gradient gradf(x) + Sum v_i gradc_i(x) at the current iterate.

    • optimality

      public double optimality
      KKT optimality measure (infinity-norm of the Lagrangian gradient).

    • constrViolation

      public double constrViolation
      Infinity-norm of the constraint residual at the current iterate.

  • Constructor Details

    • State

      public State()
      Default-construct a State with all fields at their zero/null defaults.