Template Function proxsuite::proxqp::sparse::detail::unscaled_primal_dual_residual

Function Documentation

template<typename T, typename I, typename P>
auto proxsuite::proxqp::sparse::detail::unscaled_primal_dual_residual(Workspace<T, I> &work, Results<T> &results, const Settings<T> &settings, VecMapMut<T> primal_residual_eq_scaled, VecMapMut<T> primal_residual_in_scaled_lo, VecMapMut<T> primal_residual_in_scaled_up, VecMapMut<T> dual_residual_scaled, T &primal_feasibility_eq_rhs_0, T &primal_feasibility_in_rhs_0, T &dual_feasibility_rhs_0, T &dual_feasibility_rhs_1, T &dual_feasibility_rhs_3, T &rhs_duality_gap, const P &precond, Model<T, I> const &data, const QpView<T, I> qp_scaled, VecMapMut<T> x_e, VecMapMut<T> y_e, VecMapMut<T> z_e, proxsuite::linalg::veg::dynstack::DynStackMut stack) -> proxsuite::linalg::veg::Tuple<T, T>

Derives the global primal and dual residual of the QP problem for determining whether the solution is reached (at the desired accuracy).

Parameters:

  • primal_residual_eq_scaled – vector storing the primal equality residual.

  • primal_residual_in_scaled_lo – vector storing the primal lower bound inequality residual.

  • primal_residual_in_scaled_up – vector storing the primal uppder bound inequality residual.

  • dual_residual_scaled – vector storing the dual residual.

  • primal_feasibility_eq_rhs_0 – scalar variable used when using a relative stopping criterion.

  • primal_feasibility_in_rhs_0 – scalar variable used when using a relative stopping criterion.

  • dual_feasibility_rhs_0 – scalar variable used when using a relative stopping criterion.

  • dual_feasibility_rhs_1 – scalar variable used when using a relative stopping criterion.

  • dual_feasibility_rhs_3 – scalar variable used when using a relative stopping criterion.

  • precond – preconditioner.

  • data – model of the problem.

  • qp_scaled – view on the scaled version of the qp problem.

  • x_e – current estimate of primal variable x.

  • y_e – current estimate of equality constrained lagrange multiplier.

  • z_e – current estimate of inequality constrained lagrange multiplier.

  • stak – stack.