FICO Xpress Python notebook examples

The FICO® Xpress Python interface is a powerful tool for optimization modeling and solving. It allows users to leverage the capabilities of the FICO® Xpress Solver directly within Python. Here are some key features:

1. Modeling: You can create and manipulate optimization models using Python objects. This includes defining variables, constraints, and objective functions. It allows creating constraints and objectives using expressions with operator overloading.
2. Integration with NumPy and SciPy: The interface supports the use of NumPy and SciPy sparse arrays, making it easier and more efficient to handle large datasets and perform numerical operations.
3. Solving: It provides functions to solve various types of optimization problems, including linear programming (LP), mixed-integer programming (MIP), quadratic programming (QP) and (general) nonlinear programming (NLP), and access to the full set of solver features available with Xpress.
4. Callbacks: Users can implement custom callbacks to interact with the optimization process, such as monitoring progress or modifying the solver's behavior.
5. Examples and Documentation: The interface comes with extensive examples and documentation to help users get started and explore advanced features.

Contents

• XpressPythonAPI: A slide deck with descriptions, examples and links for the most relevant API methods.
• A Youtube playlist with several videos containing slides readouts and demos using Python notebooks is available. Links to individual videos are provided below.

Basic examples of API usage

Example name	Description	Notes	Video
callback_newnode	Using a callback function during the branch and bound search		Link
indicators	Using indicator constraints		Link
load_problem	Comparison of model building performance between low-level and high-level methods		Link
loadlp	Using the low-level API function loadLP()
logic_cons	Using general constraints for logical expressions		Link
modeling	Modeling a basic MIP problem		Link
multiobj_knapsack.ipynb	Multi-objective knapsack problem with blended and lexicographic approaches
numpy_arrays	Using the numerical library NumPy		Link
piecewise_linear	Using piecewise linear functions		Link
sos	Defining Special Ordered Set (SOS) constraints		Link
write_read	Writing and reading a problem file		Link

Modeling examples

Example name	Description	Notes	Video
assignment	One-to-one assignment of projects to persons to maximize the overall preference level
bin_packing	Pack items into bins to minimize the number of bins used	Showcases the use of the xp.Dot operator
campaign_polars	Campaign conversion optimization to maximize expected value subject to budget and channel capacity constraints	Showcases the use of Polars dataframes to generate aggregate expressions
circle_packing	Place $N$ disjoint circles in the unit square with Xpress Global
facility_location	A basic facility location problem
firestation_scipy	Using a SciPy sparse matrix formulation to model the fire station location problem
inscribed_square	Inscribed square problem solved with Xpress NonLinear or Xpress Global	To run Xpress NonLinear with Knitro, a Knitro license is required.	Link
markowitz_multiobj	Multi-objective formulation of Markowitz portfolio optimization problem		Link
max_flow	Finding the maximum number of vertex-disjoint paths between two nodes in a telecommunications network
n_queens	The problem of placing $n$ queens on a chessboard
pairwise_distance	Determine the positions of $N$ points in $D$-dimensional space	Interactive 3D visualization
portfolio_pandas	Showcases the use of Pandas operations to generate expressions and constraints for a portfolio selection problem	Requires Xpress version 9.8 or later
sudoku	Solving a Sudoku problem
tsp_callbacks	Solving a TSP problem using callbacks	This example requires a full license of the FICO® Xpress Optimizer.
unitcommitment_indicators	Unit commitment problem formulation with indicator constraints

Documentation and other examples

• Python Interface Reference Manual
• Xpress Python examples

Running the examples using GitHub codespaces

1. Open Codespaces and create a codespace:

   • Click on the "Code" (green) button on this repository page.
   • On the "Codespaces" tab, select "Create a Codespace on main". This will set up a cloud-based development environment for you.

2. Open Python Notebook:

   • Once the Codespace is created and the environment is ready (wait for the README preview to appear), you can open a Python notebook.
   • Navigate to a notebook file (with a .ipynb extension) within the Codespace.
   • Click on the notebook file to open it in the Jupyter interface.

3. Run the Notebook:

   • Run a code cell in the Python notebook as you would normally do in a local environment.
   • When running for the first time, select "Install/Enable suggested extensions: Python + Jupyter" as suggested in the pop-up window at the top of the screen.
   • After installation, select the installed Python environment and wait for the code cell to be executed. You are all set to run all the code cells.

Legal

See source code files for copyright notices.

License

The examples in this repository are licensed under the Apache License, Version 2.0. See LICENSE for the full license text. The examples use FICO® Xpress software. By running it, you agree to the Community License terms of the Xpress Shrinkwrap License Agreement with respect to the FICO® Xpress software. See the licensing options overview for additional details and information about obtaining a paid license.