Answer

Enter/Click/Tap Show/Complete
Previous
Next
B Bookmark

Optimization Foundations

Optimization is the study of choosing the best possible decision from a set of allowed decisions.

In operational research, optimization is used to improve decision-making when resources are limited.

See

General Optimization Problem

An optimization problem has the form:

\[\min f(x)\]

or:

\[\max f(x)\]

where:

\[f:\mathbb{R}^n \to \mathbb{R}\]

is the objective function.

The vector:

\[x=(x_1,x_2,\ldots,x_n)\]

contains the decision variables.

Maximization and Minimization

We use maximization when the objective represents a benefit.

Examples:

  • profit
  • revenue
  • expected return
  • production output
  • utility

We use minimization when the objective represents a cost or loss.

Examples:

  • cost
  • distance
  • time
  • waste
  • risk

Constraints

A constraint is a condition that the solution must satisfy.

Constraints can be:

  • equality constraints
  • inequality constraints
  • integer constraints
  • sign constraints

Example:

\[x_1+x_2\le 100\]

could mean that the total use of a resource cannot exceed $100$.

Feasible Set

The feasible set is the set of all points satisfying all constraints.

If the feasible set is called $S$, then a constrained optimization problem is:

\[\min f(x)\]

subject to:

\[x\in S\]

Only points inside $S$ are allowed.

Optimal Solution

An optimal solution is a feasible point that gives the best objective value.

For minimization:

\[x^* \in S\]

is optimal if:

\[f(x^*)\le f(x)\]

for every feasible $x\in S$.

For maximization:

\[x^* \in S\]

is optimal if:

\[f(x^*)\ge f(x)\]

for every feasible $x\in S$.

Optimal Value

The optimal value is the value of the objective function at the optimal solution.

If $x^*$ is optimal, then:

\[z^*=f(x^*)\]

The point $x^*$ is the optimal solution.

The scalar $z^*$ is the optimal value.

Decision Process

A typical operational research workflow is:

  1. identify the problem
  2. identify the variables
  3. formulate a mathematical model
  4. solve the model using an algorithm
  5. validate the model
  6. analyze the result

Why Models Matter

A real-world problem must be translated into a mathematical model before it can be solved.

The model should specify:

  • what we choose
  • what we want
  • what limits us

In mathematical terms:

  • what we choose → decision variables
  • what we want → objective function
  • what limits us → constraints

Linear Programming as a Special Case

Linear programming is an optimization problem where:

  • the objective function is linear
  • the constraints are linear

A general linear programming problem has the form:

\[\min \text{ or } \max \quad c_1x_1+c_2x_2+\cdots+c_nx_n\]

subject to linear constraints.

In matrix form, the standard form is:

\[\min c^Tx\]

subject to:

\[Ax=b\] \[x\ge 0\]

OR Mental Model

Operational research is about making the best decision under limitations.

Examples:

Situation Decision Objective Constraint
production how much to produce maximize profit limited machines/materials
transport how much to ship minimize cost supply and demand
diet how much food to buy minimize cost nutrient requirements
investment how much to invest maximize return budget and risk limits

Checklist

You understand optimization foundations if you can:

  • identify decision variables
  • write an objective function
  • distinguish maximization from minimization
  • identify constraints
  • define the feasible set
  • distinguish optimal solution from optimal value
  • explain why OR problems are decision problems
  • recognize linear programming as a special optimization model

See Also

Exam checkpoint

In exam problems, translate the story into variables, objective, constraints, feasible set, and optimal value. Always state whether the problem is a maximization or minimization and what the units are.