We proposed a doubly stochastic primal-dual coordinate optimization algorithm for regularized empirical risk minimization that can be formulated as a saddlepoint problem. Different from existing coordinate methods, the proposed method randomly samples both primal and dual coordinates to update solutions, which is a desirable property when applied to data with both a high dimension and a large s...

We present an iterative primal-dual solver for nonconvex equality-constrained quadratic optimization subproblems. The solver constructs the primal and dual trial steps from the subspace generated by the generalized Arnoldi procedure used in flexible GMRES (FGMRES). This permits the use of a wide range of preconditioners for the primal-dual system. In contrast with FGMRES, the proposed method se...

Our goal in this proposal is to explore the connection between the local ratio technique and linear programming. We believe that a better understanding of this connection will strengthen and extend the local ratio technique and will enable us to apply the technique to a wide variety of problems. Specifically, we intend to improve the best performance guarantee and/or running time of approximati...

This paper introduces dual and primal-dual RAMP algorithms for the solution of the capacitated minimum spanning tree problem (CMST). A surrogate constraint relaxation incorporating cutting planes is proposed to explore the dual solution space. In the dual RAMP approach, primal-feasible solutions are obtained by simple tabu searches that project dual solutions onto primal feasible space. A prima...

A critical disadvantage of primal-dual interior-point methods against dual interior-point methods for large scale SDPs (semidenite programs) has been that the primal positive semidenite variable matrix becomes fully dense in general even when all data matrices are sparse. Based on some fundamental results about positive semidenite matrix completion, this article proposes a general method of exp...

A critical disadvantage of primal-dual interior-point methods compared to dual interior-point methods for large scale semidefinite programs (SDPs) has been that the primal positive semidefinite matrix variable becomes fully dense in general even when all data matrices are sparse. Based on some fundamental results about positive semidefinite matrix completion, this article proposes a general met...

When solving nonlinear least-squares problems, it is often useful to regularize the problem using a quadratic term, a practice which is especially common in applications arising in inverse calculations. A solution method derived from a trust-region Gauss-Newton algorithm is analyzed for such applications, where, contrary to the standard algorithm, the least-squares subproblem solved at each ite...

Nonlinear rescaling (NR) methods alternate finding an unconstrained minimizer of the Lagrangian for the equivalent problem in the primal space (which is an infinite procedure) with Lagrange multipliers update. We introduce and study a proximal point nonlinear rescaling (PPNR) method that preserves convergence and retains a linear convergence rate of the original NR method and at the same time d...

Motivated by various applications to computer vision, we consider an integer convex optimization problem which is the dual of the convex cost network flow problem. In this paper, we first propose a new primal algorithm for computing an optimal solution of the problem. Our primal algorithm iteratively updates primal variables by solving associated minimum cut problems. The main contribution in t...