Mixed-integer linear programming

Mixed-integer linear programming is an optimization method with linear constraints where some variables must be integers. In Intro to Industrial Engineering, it is used to model decisions like opening warehouses, choosing routes, or setting shipment counts.

What is mixed-integer linear programming?

Mixed-integer linear programming, usually shortened to MILP, is a way to build optimization models in Intro to Industrial Engineering when some decisions have to be whole numbers instead of fractions. The model still uses linear relationships, but at least one variable is restricted to integer values, often 0 or 1 for yes-no choices.

That mix is what makes MILP useful for logistics network optimization. A company might need to decide whether to open a warehouse, how much to ship from each facility, and which routes to use. The shipment amounts can often be continuous, but the warehouse decision is discrete, so a plain linear program is not enough.

A common setup is to use binary variables for facility decisions and continuous variables for flow decisions. For example, you might set one variable to 1 if a distribution center opens and 0 if it stays closed, then tie shipment constraints to that choice. This lets the model handle real-world rules like capacity limits, minimum service levels, and demand coverage.

The word linear matters because all the relationships in the objective function and constraints are written as sums of variables with constants. There are no products of decision variables, no powers, and no nonlinear curves. That structure keeps the model mathematically manageable, even though the integer part makes solving it harder than ordinary linear programming.

In practice, MILP models in industrial engineering are often built to compare trade-offs. You can test whether centralizing distribution lowers shipping cost enough to justify a larger warehouse, or whether a decentralized network improves delivery speed. The solver searches through possible integer choices and then finds the best feasible combination of opening decisions, routing, and flow amounts.

A big misconception is thinking integer just means numbers somewhere in the answer. In MILP, the integrality condition is part of the model itself, and it changes the solution process. If a variable represents a yes-no decision, a solution like 0.4 is not acceptable, even if it would look good in a regular linear program.

Why mixed-integer linear programming matters in Intro to Industrial Engineering

MILP shows up right where Intro to Industrial Engineering gets practical: designing systems that have both continuous quantities and discrete decisions. Logistics network problems are a perfect example because you are not just moving units around, you are also choosing which facilities exist, which lanes are active, and where constraints should bind.

This term connects the math of optimization to actual business choices. A network design model might minimize cost while still meeting customer demand, respecting warehouse capacities, and avoiding impossible fractional decisions like opening 2.6 distribution centers. MILP is the tool that makes those real choices fit into one model.

It also helps you see why some industrial engineering problems are harder than they first look. As soon as you add binary decisions, the model can no longer rely on simple formulas alone. That is why solvers, formulation quality, and constraint setup matter so much in class problems and case studies.

You will also see MILP when comparing centralized distribution with decentralized distribution, or when mixing transportation planning with inventory decisions. If your model has both continuous flows and yes-no choices, MILP is usually the structure you want.

How mixed-integer linear programming connects across the course

Linear Programming

MILP starts with a linear programming structure, so the objective function and constraints still behave like standard LP. The difference is that MILP adds integer restrictions, which makes the model closer to a real decision problem. If you can write the LP part clearly, you are most of the way to building a MILP.

Integer Programming

Integer programming is the broader category that covers models where decision variables must be whole numbers. MILP is a special case because only some variables are integer while others remain continuous. In industrial engineering, that mix is common when you need both quantity decisions and yes-no choices.

Network Flow Problem

Network flow models often become MILPs when the problem includes facility opening, route selection, or other discrete decisions. The flow part can stay linear and continuous, but the network design part needs integer variables. That is why MILP is so common in logistics network optimization.

Centralized Distribution

A centralized distribution design can be modeled with MILP when you need to decide whether one large hub or a smaller set of hubs should operate. The integer variables capture the facility choice, while flow variables handle how much product moves through the hub. This makes cost and service trade-offs visible in one model.

Is mixed-integer linear programming on the Intro to Industrial Engineering exam?

A problem set or quiz will usually ask you to identify which variables should be integer and which can stay continuous, then explain why the model needs both. You may be given a logistics case and asked to write the decision variables, objective function, and constraints for a MILP. If the question compares two network designs, your job is often to spot the discrete choices, like opening a warehouse, and the flow decisions, like shipment amounts.

When you solve a model by hand, watch for binary variables that enforce yes-no logic. If a proposed answer includes fractions for a facility-opening variable, that is a red flag. In written responses, the strongest answers name the integer constraint, explain the real-world decision it represents, and connect it to cost, capacity, or service level in the network.

Mixed-integer linear programming vs Linear Programming

Linear programming lets all decision variables be continuous, so you can get fractional answers. Mixed-integer linear programming keeps the same linear structure but forces some variables to be whole numbers, which is necessary for discrete choices like opening a warehouse or choosing a route.

Key things to remember about mixed-integer linear programming

• Mixed-integer linear programming is a linear optimization model with some variables restricted to integer values.

• In Intro to Industrial Engineering, MILP is especially useful for logistics network design, facility location, and routing decisions.

• Binary variables are common in MILP because they can represent yes-no choices such as opening a warehouse or using a route.

• The model stays linear, but the integer restrictions make it harder to solve than ordinary linear programming.

• If a decision cannot be fractional in real life, MILP is often the right way to model it.

Frequently asked questions about mixed-integer linear programming

What is mixed-integer linear programming in Intro to Industrial Engineering?

It is an optimization method that uses linear equations and constraints, but requires some variables to be integers. In Intro to Industrial Engineering, you use it for problems where part of the decision is continuous, like shipment amounts, and part is discrete, like whether to open a warehouse.

Why is mixed-integer linear programming used for logistics network optimization?

Logistics networks involve both flow decisions and location or route decisions. The flow can often be modeled with continuous variables, but facility openings and route choices need integer or binary variables. MILP lets you put both kinds of decisions into one model.

What is the difference between linear programming and mixed-integer linear programming?

Linear programming allows variables to take any real value within the constraints, so solutions can be fractional. MILP adds integer requirements for some variables, which makes the model better for real decisions that cannot be split into pieces, like opening 1.5 warehouses.

How do you spot a MILP problem in class?

Look for a mix of continuous amounts and yes-no or whole-number decisions. If the problem includes facility location, network design, route selection, or other discrete choices along with shipment quantities, it is probably a MILP model.