This glossary explains the key concepts and terms for Optimizer. 

TermDefinition
ConstraintA limit on a value, such as its maximum, minimum, or that the value can’t be negative.
Feasibility

An alternative to optimality, this offers a possible solution to a problem with:

  • no distinct objective
  • a variable
  • one or more constraints
Input dataValues necessary for computing the solution, including any constraints.
Integer linear programA linear program where variables are constrained to integral values (whole numbers).
Linear functionA function with a polynomial of degree 0 or 1. Displays as a straight line on a graph.
Linear programThe pursuit of a solution in the form of a real number, where the Objective Function and the constraints are linear.
MIP GapFor mixed integer linear problems, the Optimizer compares the current best solution with the theoretically best possible solution. The optimization is successful when the gap between those values is within a specific tolerance (the MIP Gap).
Mixed integer linear programA linear program where only some of the variables are constrained to integral values. Other variables can be real values (decimal numbers).
ObjectiveAn expression that guides the optimization engine while it determines which assignments best support the business goal or solution, such as maximum income or minimal expense.
Optimality

An optimal solution to a problem with:

  • an objective (such as lowest cost or highest profit)
  • a decision variable
  • one or more constraints
Time-outThe number of seconds until an Optimizer action that is processing stops and abandons progress. This value must be set when creating an Optimizer action to prevent Optimizer from running indefinitely if a problem is unsolvable.
Upper bound, Lower boundThe maximum or minimum value for a variable.
VariableThe value that represents the solution to the problem (sometimes called the decision variable).

Variable data type

(Variables must have a numeric data type)

Integer (whole number)

Real (decimal, or floating point)

Binary (zero or one, can be called Boolean)