Polynomial programming, a class of non-linear programming where the objective and the constraints are multivariate polynomials, has attracted the attention of many researchers in the past decade. Polynomial programming is a powerful modeling tool that captures various optimization models. Due to the wide range of applications, a research topic of high interest is the development of computationally efficient algorithms for solving polynomial programs. Even though some solution methodologies are already available and have been studied in the literature, these approaches are often either problem specific or are inapplicable for large-scale polynomial programs. Most of the available methods are based on using hierarchies of convex relaxations to solve polynomial programs; these schemes grow exponentially in size becoming rapidly computationally expensive. The present work proposes methods and implementations that are capable of solving polynomial programs of large sizes. First we propose a general framework to construct conic relaxations for binary polynomial programs, this framework allows us to re-derive previous relaxation schemes and provide new ones. In particular, three new relaxations for binary quadratic polynomial programs are presented. The first two relaxations, based on second-order cone and semidefinite programming, represent a significant improvement over previous practical relaxations for several classes of non-convex binary quadratic polynomial problems. The third relaxation is based purely on second-order cone programming, it outperforms the semidefinite-based relaxations that are proposed in the literature in terms of computational efficiency while being comparable in terms of bounds. To strengthen the relaxations further, a dynamic inequality generation scheme to generate valid polynomial inequalities for general polynomial programs is presented. When used iteratively, this scheme improves the bounds without incurring an exponential growth in the size of the relaxation. The scheme can be used on any initial relaxation of the polynomial program whether it is second-order cone based or semidefinite based relaxations. The proposed scheme is specialized for binary polynomial programs and is in principle scalable to large general combinatorial optimization problems. In the case of binary polynomial programs, the proposed scheme converges to the global optimal solution under mild assumptions on the initial approximation of the binary polynomial program. Finally, for binary polynomial programs the proposed relaxations are integrated with the dynamic scheme in a branch-and-bound algorithm to find global optimal solutions.

Polynomial programming, a class of non-linear programming where the objective and the constraints are multivariate polynomials, has attracted the attention of many researchers in the past decade. Polynomial programming is a powerful modeling tool that captures various optimization models. Due to the wide range of applications, a research topic of high interest is the development of computationally efficient algorithms for solving polynomial programs. Even though some solution methodologies are already available and have been studied in the literature, these approaches are often either problem specific or are inapplicable for large-scale polynomial programs. Most of the available methods are based on using hierarchies of convex relaxations to solve polynomial programs; these schemes grow exponentially in size becoming rapidly computationally expensive. The present work proposes methods and implementations that are capable of solving polynomial programs of large sizes. First we propose a general framework to construct conic relaxations for binary polynomial programs, this framework allows us to re-derive previous relaxation schemes and provide new ones. In particular, three new relaxations for binary quadratic polynomial programs are presented. The first two relaxations, based on second-order cone and semidefinite programming, represent a significant improvement over previous practical relaxations for several classes of non-convex binary quadratic polynomial problems. The third relaxation is based purely on second-order cone programming, it outperforms the semidefinite-based relaxations that are proposed in the literature in terms of computational efficiency while being comparable in terms of bounds. To strengthen the relaxations further, a dynamic inequality generation scheme to generate valid polynomial inequalities for general polynomial programs is presented. When used iteratively, this scheme improves the bounds without incurring an exponential growth in the size of the relaxation. The scheme can be used on any initial relaxation of the polynomial program whether it is second-order cone based or semidefinite based relaxations. The proposed scheme is specialized for binary polynomial programs and is in principle scalable to large general combinatorial optimization problems. In the case of binary polynomial programs, the proposed scheme converges to the global optimal solution under mild assumptions on the initial approximation of the binary polynomial program. Finally, for binary polynomial programs the proposed relaxations are integrated with the dynamic scheme in a branch-and-bound algorithm to find global optimal solutions.

Monotonic optimization consists of minimizing or maximizing a monotonic objective function over a set of constraints defined by monotonic functions. Many optimization problems in economics and engineering often have monotonicity while lacking other useful properties, such as convexity. This thesis is concerned with the development and application of global optimization algorithms for monotonic optimization problems. First, we propose enhancements to an existing outer-approximation algorithm | called the Polyblock Algorithm | for monotonic optimization problems. The enhancements are shown to significantly improve the computational performance of the algorithm while retaining the convergence properties. Next, we develop a generic branch-and-bound algorithm for monotonic optimization problems. A computational study is carried out for comparing the performance of the Polyblock Algorithm and variants of the proposed branch-and-bound scheme on a family of separable polynomial programming problems. Finally, we study an important class of monotonic optimization problems | probabilistically constrained linear programs. We develop a branch-and-bound algorithm that searches for a global solution to the problem. The basic algorithm is enhanced by domain reduction and cutting plane strategies to reduce the size of the partitions and hence tighten bounds. The proposed branch-reduce-cut algorithm exploits the monotonicity properties inherent in the problem, and requires the solution of only linear programming subproblems. We provide convergence proofs for the algorithm. Some illustrative numerical results involving problems with discrete distributions are presented.