Primal-dual Path-following Method Research Articles

Primal-Dual Path-Following Methods and the Trust-Region Updating Strategy for Linear Programming with Noisy Data

In this article, we consider the primal-dual path-following method and the trust-region updating strategy for the standard linear programming problem. For the rank-deficient problem with the small noisy data, we also give the preprocessing method based on the QR decomposition with column pivoting. Then, we prove the global convergence of the new method when the initial point is strictly primal-dual feasible. Finally, for some rank-deficient problems with or without the small noisy data from the NETLIB collection, we compare it with other two popular interior-point methods, i.e. the subroutine pathfollow.m and the built-in subroutine linprog.m of the MATLAB environment. Numerical results show that the new method is more robust than the other two methods for the rank-deficient problem with the small noise data.

A Shifted Primal-Dual Penalty-Barrier Method for Nonlinear Optimization

In nonlinearly constrained optimization, penalty methods provide an effective strategy for handling equality constraints, while barrier methods provide an effective approach for the treatment of inequality constraints. A new algorithm for nonlinear optimization is proposed based on minimizing a shifted primal-dual penalty-barrier function. Certain global convergence properties are established. In particular, it is shown that a limit point of the sequence of iterates may always be found that is either an infeasible stationary point or a complementary approximate Karush--Kuhn--Tucker point; i.e., it satisfies reasonable stopping criteria and is a Karush--Kuhn--Tucker point under a regularity condition that is the weakest constraint qualification associated with sequential optimality conditions. It is also shown that under suitable additional assumptions, the method is equivalent to a shifted variant of the primal-dual path-following method in the neighborhood of a solution. Numerical examples are provided that illustrate the performance of the method compared to a widely used conventional interior-point method.

Analysis of some interior point continuous trajectories for convex programming

In this paper, we analyse three interior point continuous trajectories for convex programming with general linear constraints. The three continuous trajectories are derived from the primal–dual path-following method, the primal–dual affine scaling method and the central path, respectively. Theoretical properties of the three interior point continuous trajectories are fully studied. The optimality and convergence of all three interior point continuous trajectories are obtained for any interior feasible point under some mild conditions. In particular, with proper choice of some parameters, the convergence for all three interior point continuous trajectories does not require the strict complementarity or the analyticity of the objective function. These results are new in the literature.

An iteration primal–dual path-following method, based on wide neighbourhood and large update, for second-order cone programming

In this article we propose a new primal-dual path-following interior point algorithm for second-order cone programming. Each iterate always follows the usual wide neighborhood , it does not necessarily stay within it, but must stay within the wider neighbourhood 𝒩(τ, β). We show that the algorithm has iteration complexity bound which is better than that of usual wide neighbourhood algorithm O(n log ϵ−1), where n is the dimension of the problem, with ϵ the required precision, α ∈ [0, 1] the given constant number and (x 0, s 0) the initial interior solution. It is the best result in regard to the iteration complexity bound in the context of path-following method for second-order cone programming.

Semidefinite and second-order cone optimization approach for the Toeplitz matrix approximation problem

The nearest positive semidefinite symmetric Toeplitz matrix to an arbitrary data covariance matrix is useful in many areas of engineering, including stochastic filtering and digital signal processing applications. In this paper, the interior point primal-dual path-following method will be used to solve our problem after reformulating it into different forms, first as a semidefinite programming problem, then into the form of a mixed semidefinite and second-order cone optimization problem. Numerical results, comparing the performance of these methods against the modified alternating projection method will be reported.

A new primal-dual path-following method for convex quadratic programming

In this paper, we describe a new primal-dual path-following method to solve a convex quadratic program (QP). The derived algorithm is based on new techniques for finding a new class of search directions similar to the ones developed in a recent paper by Darvay for linear programs. We prove that the short-update algorithm finds an e-solution of (QP) in a polynomial time.

Semidefinite and second-order cone optimization approach for the Toeplitz matrix approximation problem

The nearest positive semidefinite symmetric Toeplitz matrix to an arbitrary data covariance matrix is useful in many areas of engineering, including stochastic filtering and digital signal processing applications. In this paper, the interior point primal-dual path-following method will be used to solve our problem after reformulating it into different forms, first as a semidefinite programming problem, then into the form of a mixed semidefinite and second-order cone optimization problem. Numerical results, comparing the performance of these methods against the modified alternating projection method will be reported.

General primal-dual penalty/barrier path-following Newton methods for nonlinear programming

In the present article rather general penalty/barrier-methods are considered, that define a local continuously differentiable primal-dual path. The class of penalty/barrier terms includes most of the usual techniques like logarithmic barriers, SUMT, quadratic loss functions as well as exponential penalties, and the optimization problem which may contain inequality as well as equality constraints. The convergence of the corresponding general primal-dual path-following method is shown for local minima that satisfy strong second-order sufficiency conditions with linear independence constraint qualification (LICQ) and strict complementarity. A basic tool in the analysis of these methods is to estimate the radius of convergence of Newton's method depending on the penalty/barrier-parameter. Without using self-concordance properties convergence bounds are derived by direct estimations of the solutions of the Newton equations. Parameter selection rules are proposed which guarantee the local convergence of the considered penalty/barrier-techniques with only a finite number of Newton steps at each parameter level. Numerical examples illustrate the practical behavior of the proposed class of methods.

An O$(\sqrtn L)$ Iteration Primal-dual Path-following Method, Based on Wide Neighborhoods and Large Updates, for Monotone LCP

In this paper we propose a new class of primal-dual path-following interior point algorithms for solving monotone linear complementarity problems. At each iteration, the method would select a target on the central path with a large update from the current iterate, and then the Newton method is used to get the search directions, followed by adaptively choosing the step sizes, which are, e.g., the largest possible steps before leaving a neighborhood that is as wide as the given ${\cal N}^-_{\infty}$ neighborhood. The only deviation from the classical approach is that we treat the classical Newton direction as the sum of two other directions, corresponding to, respectively, the negative part and the positive part of the right-hand side. We show that if these two directions are equipped with different and appropriate step sizes, then the method enjoys the low iteration bound of $O(\sqrt{n}\log L)$, where $n$ is the dimension of the problem and $L=\frac{(x^0)^Ts^0}{\ep}$ with $\ep$ the required precision and $(x^0,s^0)$ the initial interior solution. For a predictor-corrector variant of the method, we further prove that, besides the predictor steps, each corrector step also reduces the duality gap by a rate of $1-1/O(\sqrt{n})$. Additionally, if the problem has a strict complementary solution, then the predictor steps converge Q-quadratically.

Parallel Birge and Qi Factorization of Three-stage Stochastic Linear Programs

Many practical problems, such as portfolio management, problems in manufacturing, transportation or power generation, can be modeled as stochastic programs. Stochastic programs provide an effective framework for sequential decision problems with uncertain data, when uncertainty can be modeled by a discrete set of scenarios. In this paper we present an algorithm for solving a three-stage stochastic linear program based on the Birge and Qi factorization of a constraint matrix product in the frame of the primal-dual path-following interior point method. Moreover, we discuss the parallelization of this method for distributed-memory machines using Fortran/MPI and the linear algebra package LAPACK. Performance experiments on a 64-processor Beowulf cluster show the effectiveness of the proposed parallelization strategy, which exploits two levels of parallelism.

Some New Search Directions for Primal-Dual Interior Point Methods in Semidefinite Programming

Search directions for primal-dual path-following methods for semidefinite programming (SDP) are proposed. These directions have the properties that (1) under certain nondegeneracy and strict complementarity assumptions, the Jacobian matrix of the associated symmetrized Newton equation has a bounded condition number along the central path in the limit as the barrier parameter $\mu$ tends to zero; and (2) the Schur complement matrix of the symmetrized Newton equation is symmetric and the cost for computing this matrix is 2mn3 + 0.5m2n2 flops, where n and m are the dimension of the matrix and vector variables of the semidefinite program, respectively. These two properties imply that a path-following method using the proposed directions can achieve the high accuracy typically attained by methods employing the direction proposed by Alizadeh, Haeberly, and Overton (currently the best search direction in terms of accuracy), but each iteration requires at most half the amount of flops (to leading order).

A lower bound on the number of iterations of long-step primal-dual linear programming algorithms

Recently, Todd has analyzed in detail the primal-dual affine-scaling method for linear programming, which is close to what is implemented in practice, and proved that it may take at leastn1/3 iterations to improve the initial duality gap by a constant factor. He also showed that this lower bound holds for some polynomial variants of primal-dual interior-point methods, which restrict all iterates to certain neighborhoods of the central path. In this paper, we further extend his result to long-step primal-dual variants that restrict the iterates to a wider neighborhood. This neigh-borhood seems the least restrictive one to guarantee polynomiality for primal-dual path-following methods, and the variants are also even closer to what is implemented in practice.

On the Relationship Between the Curvature Integral and the Complexity of Path-Following Methods in Linear Programming

In this paper we study the complexity of primal-dual path-following methods which are a particular kind of interior point method for solving linear programs. In particular we establish a relationship between the complexity and the integral of a weighted curvature along the central trajectory (path) defined in the interior of the feasible solution space of a linear program. An important property of the trajectory, viz., that its higher order derivatives are bounded by a geometric series, is presented. Applying this property we show that the complexity of such methods is bounded from below and from above by the curvature integral. This theorem reduces the complexity analysis to a relatively simple problem of estimating the curvature integral.

Estimating the complexity of a class of path-following methods for solving linear programs by curvature integrals

In this paper we study a particular class of primal-dual path-following methods which try to follow a trajectory of interior feasible solutions in primal-dual space toward an optimal solution to the primal and dual problem. The methods investigated are so-called first-order methods: each iteration consists of a “long” step along the tangent of the trajectory, followed by explicit recentering steps to get close to the trajectory again. It is shown that the complexity of these methods, which can be measured by the number of points close to the trajectory which have to be computed in order to achieve a desired gain in accuracy, is bounded by an integral along the trajectory. The integrand is a suitably weighted measure of the second derivative of the trajectory with respect to a distinguished path parameter, so the integral may be loosely called a curvature integral.