All roads lead to Newton : Feasible second - order methods for equality - constrained optimization ∗
نویسندگان
چکیده
This paper considers the connection between the intrinsic Riemannian Newton method and other more classically inspired optimization algorithms for equality-constrained optimization problems. We consider the feasibly-projected sequential quadratic programming (FP-SQP) method and show that it yields the same update step as the Riemannian Newton, subject to a minor assumption on the choice of multiplier vector. We also consider Newton update steps computed in various ‘natural’ local coordinate systems on the constraint manifold and find simple conditions that guarantee that the update step is the Riemannian Newton update. In particular, we show that this is the case for projective local coordinates, one of the most natural choices that have been proposed in the literature. Finally we consider the case where the full constraints are approximated to simplify the computation of the update step. We show that if this approximation is good at least to second-order then the resulting update step is the Riemannian Newton update. The conclusion of this study is that the intrinsic Riemannian Newton algorithm is the archetypal feasible second order update for non-degenerate equality constrained optimisation problems.
منابع مشابه
Trust-region methods for rectangular systems of nonlinear equations
Here Θ : X → IR is a continuously differentiable mapping, X ⊆ IR is an open set containing the feasible region Ω and Ω is an n-dimensional box, Ω = {x ∈ IR : l ≤ x ≤ u}. These inequalities are meant component-wise and l ∈ (IR ∪−∞), u ∈ (IR ∪∞). Taking into account the variety of applications yielding the problem (1), we allow any relationship between m and n. The relevance of this problem is we...
متن کاملQuasi-Newton Methods for Nonconvex Constrained Multiobjective Optimization
Here, a quasi-Newton algorithm for constrained multiobjective optimization is proposed. Under suitable assumptions, global convergence of the algorithm is established.
متن کاملConstrained Multi-Objective Optimization Problems in Mechanical Engineering Design Using Bees Algorithm
Many real-world search and optimization problems involve inequality and/or equality constraints and are thus posed as constrained optimization problems. In trying to solve constrained optimization problems using classical optimization methods, this paper presents a Multi-Objective Bees Algorithm (MOBA) for solving the multi-objective optimal of mechanical engineering problems design. In the pre...
متن کاملQuasi-Newton acceleration for equality-constrained minimization
Optimality (or KKT) systems arise as primal-dual stationarity conditions for constrained optimization problems. Under suitable constraint qualifications, local minimizers satisfy KKT equations but, unfortunately, many other stationary points (including, perhaps, maximizers) may solve these nonlinear systems too. For this reason, nonlinear-programming solvers make strong use of the minimization ...
متن کاملStable barrier-projection and barrier-Newton methods in linear programming
The present paper is devoted to the application of the space transformation techniques for solving linear programming problems. By using a surjective mapping the original constrained optimization problem is transformed to a problem in a new space with only equality constraints. For the numerical solution of the latter problem the stable version of the gradient-projection and Newton’s methods ar...
متن کامل