Unconstrained STCGbased RSNK¶

class
kona.algorithms.
UnconstrainedRSNK
(primal_factory, state_factory, eq_factory, ineq_factory, optns=None)[source]¶ Bases:
kona.algorithms.base_algorithm.OptimizationAlgorithm
A reducedspace NewtonKrylov optimization algorithm for PDEgoverned unconstrained problems.
This algorithm uses a 2nd order adjoint formulation to compute matrixvector products with the Reduced Hessian.
The product is then used in a Krylov solver to compute a Newton step.
The step can be globalized using either linesearch or trustregion methods. The Krylov solver changes based on the type of globalization selected by the user. Unglobalized problems are solved via FGMRES, while trustregion and linesearch methods use ConjugateGradient.
Note
Insert inexactHessian paper reference here.
Variables:  factor_matrices (bool) – Boolean flag for matrixbased PDE solvers.
 iter (int) – Optimization iteration counter.
 hessian (
ReducedHessian
) – Matrix object defining the Hessian matrixvector product.  precond (
BaseHessian
like) – Matrix object defining the approximation to the Hessian inverse.  krylov (
FGMRES
orSTCG
) – A krylov solver object used to solve the system defined by the Hessian.  globalization (string) – Flag to determine which type of globalization to use.
 max_radius (radius,) – Trust radius parameters.
 line_search (
BackTracking
) – Backtracking line search tool.  merit_func (
ObjectiveMerit
) – Simple objective as merit function.