Benders Decomposition - Algorithm for NLP, MILP and MINLP that relies on decomposition in which variables are partitioned into complicating and non-complicating |
|
Bilinear Function - Function given by the sum of products of two variables |
|
Branch and Bound - Algorithm for MILP and MINLP that relies on enumeration and bounding of a tree search to find the optimum integer solution |
|
Constraint Programming - Logic-based optimization technique that is based on implicit enumeration and constraint propagation |
|
Constraints - Equations and/or inequalities that constrain the values of the variables in an optimization model |
|
Convex Function - Function coincides or underestimates all linear interpolations between any two arbitrary points (e.g. all linear functions are
convex) |
|
Convex Region - Region in which any linear combination obtained from two arbitrary points yields a new point belonging to that region |
|
Cutting Planes - Additional constraints that are added to MILP problems to improve their LP approximation when all variables are treated as continuous variables |
|
Disjunctive Programming - Optimization problem with an objective function and constraints expressed in logic form with disjunctions ( OR operators), and
propositional logic. |
|
Feasible Region - Region given by set of values of variables that satisfy constraints. |
|
Global Optimization - Methods that guarantee findingb the global optimum in nonlinear optimization problems |
|
Global Optimum - Optimum solution to model (1) such that any feasible variable values different to that produce a worsening in the objective function
value |
|
Karush-Kuhn-Tucker Conditions - Generalization of the zero derivative optimality condition to problems with constraints |
|
Local Optimum - Optimum solution to a model such that small perturbations around that point lead to a worsening of the objective function value |
|
Linear Programming - Optimization problems with linear objective function and constraints involving only continuous variables |
|
Mixed-integer Linear Programming (MILP) - Extension of linear programming that allows some of the variables to take on discrete values (mostly 0-1) |
|
Mixed-integer Nonlinear Programming (MINLP) - Extension of nonlinear programming that allows some of the variables to take on discrete values (mostly 0-1) |
|
Multiobjective Optimization - Optimization problems with more than one objective function |
|
Multiperiod Optimization - Optimization problems in which constraints are specified over several time periods or a set of scenarios |
|
Nonlinear Programming - Optimization problems with nonlinear objective function and constraints involving only continuous variables |
|
Nonconvexity - Condition of a function or region that does not satisfy convexity conditions |
|
NP-Completeness - Theoretical characterization of worst case for computational requirements that increase exponentially with problem size |
|
Optimum Solution - Variable values that correspond to the solution to a mathematical optimization model with a single objective function and constraints. |
|
Outer Approximation - Algorithm for MINLP that relies on accumulation of linearizations to bound the objective function and feasible region |
|
Penalty Function - Redefined objective function which involves the original objective plus a weighted violation of the constraints |
|
Stochastic Optimization - Optimization problems in which some of the input data are random or subject to fluctuations |
|