Quick introduction¶

AMPL (A Mathematical Programming Language) is a high-level language for expressing optimization models. It is widely used in academia and industry to model and solve linear, nonlinear, and integer optimization problems.

With AMPL you can solve your models with a large variety of solvers, such as HiGHS, Gurobi, CPLEX, or XPRESS. You can use AMPL through a command-line interface or via an API.

AMPL Syntax¶

AMPL syntax matches naturally the mathematical description of the traditional algebraic descriptions of models as you can see in the following example:

Traditional informal algebraic description of a model:

\[\begin{split}\def\entity#1{{\color{##7E78D3}{#1}}} \def\index#1{{\color{##307090}{#1}}} \def\comment#1{: {\color{##7A8596}{\text{#1}}}&} \def\statement#1{{\color{##7FA492}{\text{#1}}}} \begin{align} & \entity{R} \comment{a set of raw materials}\\ & \entity{P} \comment{a set of products}\\ & \\ & \entity{a}_\index{ij}, \index{i} \in \entity{R}, \index{j} \in \entity{P} \comment{input-output coefficients}\\ & \entity{b}_\index{i}, \index{i} \in \entity{R} \comment{units available}\\ & \entity{c}_\index{j}, \index{j} \in \entity{P} \comment{profit per unit}\\ & \entity{u}_\index{j}, \index{j} \in \entity{P} \comment{production limit}\\ & \\ & \entity{x}_\index{j}, \index{j} \in \entity{P}, 0 \leq \entity{x}_\index{j} \leq \entity{u}_\index{j} \comment{units of j produced}\\ & \\ & \statement{maximize } \sum_{\index{j} \in \entity{P}} \entity{c}_\index{j}\entity{x}_\index{j} \comment{total profit}\\ & \\ & \statement{subject to } \sum_{\index{j} \in \entity{P}} \entity{a}_\index{ij}\entity{x}_\index{j} \leq \entity{b}_\index{i}, \index{i} \in \entity{R} \comment{limited availability of material}\\ \end{align}\end{split}\]

Corresponding AMPL model:

set R; # a set of raw materials
set P; # a set of products

param a {R, P} >= 0; # input-output coefficients
param b {R} > 0; # units available
param c {P} > 0; # profit per unit
param u {P} > 0; # production limit

var x {j in P} >= 0, <= u[j]; # units of j produced

maximize total_profit:
    sum {j in P} c[j] * x[j]; # total profit
subject to supply {i in R}:
    sum {j in P} a[i,j] * x[j] <= b[i]; # limited availability of material

Single-letter names for illustration purposes only! With AMPL you can use meaningful names for better model maintainability!

To get started with AMPL, you will need to learn the syntax of the language. You can find a reference manual and examples on the AMPL website (http://www.ampl.com/) or in the AMPL book (https://ampl.com/learn/ampl-book/). There are also online resources and courses available to help you learn AMPL.

AMPL Entities¶

Here are some simple examples of the basic elements of an AMPL model:

Sets¶

Sets are used to define the sets of decision variables and constraints. For example, you might define a set of cities, a set of products, or a set of time periods.

# define a set of cities
set CITIES;

# define a set of products
set PRODUCTS;

# define a set of time periods
set PERIODS;

Parameters¶

Parameters are used to define the data for your model. For example, you might define the cost of each product, the demand for each product, or the capacity of each facility.

# define the cost of each product
param cost{PRODUCTS};

# define the demand for each product
param demand{PRODUCTS};

# define the capacity of each facility
param capacity{CITIES};

Decision variables¶

Decision variables represent the decisions you need to make in your model. For example, you might define a binary variable to represent whether a product is produced at a particular facility, or a continuous variable to represent the amount of a product that is produced.

# define a binary variable to represent whether a product is produced at a facility
var x{PRODUCTS, CITIES} binary;

# define a continuous variable to represent the amount of a product that is produced
var y{PRODUCTS, PERIODS} >= 0;

Objective function¶

The objective function defines the goal of your optimization model. For example, you might want to maximize profit, minimize cost, or maximize customer satisfaction.

# maximize profit
maximize total_profit:
  sum{p in PRODUCTS, c in CITIES} (x[p,c] * (demand[p] - cost[p]));

# minimize cost
minimize total_cost:
  sum{p in PRODUCTS, c in CITIES} (x[p,c] * cost[p]);

Constraints¶

Constraints define the limitations of your model. For example, you might have a capacity constraint that limits the amount of a product that can be produced at a facility, or a budget constraint that limits the amount of money you can spend.

# capacity constraint
subject to capacity_constraint{c in CITIES}:
  sum{p in PRODUCTS} x[p,c] <= capacity[c];

# budget constraint
param budget;
subject to budget_constraint:
  sum{p in PRODUCTS, c in CITIES} (x[p,c] * cost[p]) <= budget;

AMPL Scripting¶

Here are some examples of the most important AMPL scripting commands:

model model_file.mod;: specifies the location of the model file
data data_file.dat;: specifies the location of the data file
option option_name option_value;: specifies options for the solver
solve;: solves the optimization model
display expression;: displays the values of the decision variables, objective function, and constraints
let name := value;: assigns a value to a declared parameter or a decision variable

Here is an example of a simple AMPL script:

# specify the model and data files
model model_file.mod;
data data_file.dat;

# specify the solver options
option solver highs;

# solve the model
solve;

# display the values of the decision variables, objective function, and constraints
display x, y, total_profit, capacity_constraint, budget_constraint;

# assign a new value to a parameter
let budget := 100;

# re-solve the model
solve;

# display the new values of the decision variables and objective function
display x, y, total_profit;

AMPL Resoures

How to install AMPL