Package org.hipparchus.optim.nonlinear.vector.constrained
package org.hipparchus.optim.nonlinear.vector.constrained
This package provides algorithms that minimize the residuals
between observations and model values.
The
Algorithms in this category need access to a problem (represented by a
The problem can be created progressively using a
least-squares optimizers minimize the distance (called
cost or χ2) between model and
observations.
Algorithms in this category need access to a problem (represented by a
LeastSquaresProblem).
Such a model predicts a set of values which the algorithm tries to match
with a set of given set of observed values.
The problem can be created progressively using a
builder or it can
be created at once using a factory.- Since:
- 3.1
-
ClassDescriptionDeprecated.Abstract class for Sequential Quadratic Programming solversConvergence Checker for ADMM QP Optimizer.Alternative Direction Method of Multipliers Solver.TBD.Alternating Direction Method of Multipliers Quadratic Programming Optimizer. \[ min \frac{1}{2} X^T Q X + G X a\\ A X = B_1\\ B X \ge B_2\\ l_b \le C X \le u_b \] Algorithm based on paper:"An Operator Splitting Solver for Quadratic Programs(Bartolomeo Stellato, Goran Banjac, Paul Goulart, Alberto Bemporad, Stephen Boyd,February 13 2020)"Container for
ADMMQPOptimizersettings.Internal Solution for ADMM QP Optimizer.BFGS Hessian updater with dynamic damping and robustness improvements.Constraint with lower and upper bounds: \(l \le f(x) \le u\).Generic constraint.Abstract Constraint Optimizer.Functional interface to allow lightweight debug logging without coupling.Equality Constraint.Enumerate for gradient mode.Inequality Constraint with lower bound only: \(l \le f(x)\).Inverse of Cholesky factorization.Karush–Kuhn–Tucker Solver.Container for Lagrange t-uple.A set of linear inequality constraints expressed as ub>Ax>lb.A set of linear equality constraints given as Ax = b.Set of linear inequality constraints expressed as \( A x \gt B\).Robust Line Search strategy.Tolerance for symmetric matrix decomposition.Implementation of the Augmented Lagrangian Merit Function.Read-only, zero-copy row-oriented view of all linear constraints of a QP.Implements the dual active-set method by Goldfarb and Idnani (1983) for solving strictly convex quadratic programs of the form:Options for dual active-set method.Marker OptimizationData indicating that the quadratic matrix "G" passed to the QP solver is not the Hessian H, but the lower Cholesky factor L such that H = L*L^T or the inverse of Cholesky factor L^-1 When present and set to true, QPDualActiveSolver will interpret function.getP() as L or L^-1 and will NOT factorize it.Quadratic programming Optimizater.Updates a QR factorization when adding or removing constraints in active-set methods.Given P, Q, d, implements \(\frac{1}{2}x^T P X + Q^T x + d\).Calculates a Cholesky decomposition with adaptive diagonal regularization.Custom convergence criterion for an SQP optimizer.Adapter exposing the equality constraint set of anSQPProblemthrough theEqualityConstraintabstraction used by SQP solvers.Adapter representing the inequality constraint set of anSQPProblemas anInequalityConstraintsuitable for use by SQP solvers.Utility class for formatting SQP iteration logs with dynamic precision and aligned columns.Adapter class that exposes anSQPProblemobjective function under theTwiceDifferentiableFunctionabstraction used by SQP optimizers.Deprecated.as of 4.1, will be removed in 5.0; useSQPOptimizerS2instead.Deprecated.as of 4.1, replaced bySQPOptimizerS2; will be removed in 5.0.Sequential Quadratic Programming Optimizer (extended version).Parameter for SQP Algorithm.Defines a nonlinear constrained optimization problem suitable for use with Sequential Quadratic Programming (SQP) solvers.A MultivariateFunction that also has a defined gradient and Hessian.A MultivariateFunction that also has a defined gradient and Hessian.
AbstractSQPOptimizer2; will be removed in 5.0