LagrangianHessian

class kona.linalg.matrices.hessian.LagrangianHessian(vector_factories, optns=None)

Bases: kona.linalg.matrices.hessian.basic.BaseHessian

Matrix object for the Hessian block of the reduced KKT matrix.

Uses the same 2nd order adjoint formulation as ReducedKKTMatrix, but only for the diagonal Hessian block, \(\mathsf{W} = \nabla^2_x \mathcal{L}\).

If slack terms are present, it will also perform a product with the slack derivative of the Lagrangian.

approx

linearize(X, at_state, at_adjoint, obj_scale=1.0, cnstr_scale=1.0)

multiply_W(in_vec, out_vec)

multiply_slack(in_vec, out_vec)

precond(in_vec, out_vec)

product(in_vec, out_vec)

set_projector(proj_cg)

solve(rhs, solution, rel_tol=None)