Glossary

F
`free parameter`	A parameter which varies according to the optimizer algorithm: These parameters cannot be driven by a relation (formula, check, rule,...). Other parameters are either frozen or computed from the free parameter values.
G
`gap`	Equal to the left hand side minus the right hand side (absolute value): \|left hand side - right hand side\|
`Conjugate Gradient`	A fundamental technique that is often incorporated into various iterative algorithms in many areas of scientific computing. The Conjugate Gradient method was originally designed to minimize convex quadratic functions but has been extended to the general case. A CG algorithm is performing but not very robust.
K
`knowledgeware`	The set of software components dedicated to the creation and manipulation of knowledge-based information. Knowledge-based information consists of rules and other types of relations whereby designers can save their corporate know-how and reuse it later on to drive their design processes.
M
`maximization`	An option of the Optimization process which consists in searching for the maximum value of an objective function.
`minimization`	An option of the Optimization process which consists in searching for the minimum value of an objective function.
O
`objective function`	A function associated with an optimization problem which determines how good a solution is. The objective function is a literal feature formula that can be created by using the formula editor.
`optimization`	A computational problem in which the aim is to find the best of all possible solutions.
`optimization constraint`	A relation based on the Knowledgeware language that expresses a constraint between several persistent parameters used as input to be integrated in the definition of optimization.
`Optimizer`	An analysis tool which helps users solve an optimization problem.
S
`Simulated Annealing`	A technique used to approximate the solution of very large combinatorial optimization problems. The technique originates from the theory of statistical mechanics and is based upon the analogy between the annealing of solids and solving optimization problems. A Simulated Annealing algorithm is slow but more robust than a Conjugate Gradient algorithm.