FICO Xpress Optimizer Python Interface

9.8.1 · active · verified Thu Apr 16

The FICO Xpress Optimizer Python interface (xpress) enables users to formulate, manage, and solve a comprehensive range of mathematical optimization problems including Linear Programming (LP), Quadratic Programming (QP), Second-Order Conic Programming (SOCP), and their mixed-integer counterparts (MILP, MIQP, MIQCQP, MISOCP), along with general nonlinear and mixed-integer nonlinear problems. It offers seamless integration with numerical libraries like NumPy and supports advanced features such as callbacks for solver control. The library is currently at version 9.8.1, with frequent updates including patches and minor versions released multiple times a year.

Common errors

```
ModuleNotFoundError: No module named 'xpress'
```
cause The `xpress` Python package has not been installed in the current environment, or the environment is not correctly activated.

fix Run `pip install xpress` to install the package. Ensure your Python environment is correctly set up and activated.
```
xpress.InterfaceError: Xpress licensing error 89: Your license only supports platform(s) win32, win_x86_64. Please contact support@fico.com to upgrade it.
```
cause This error indicates a mismatch between the platform for which the Xpress license (`xpauth.xpr`) was generated and the current operating system (e.g., using a Windows license on Linux), or the license file is invalid/missing.

fix Verify that your license file is valid for your current platform and that the `XPRESS` environment variable is correctly set to the directory containing `xpauth.xpr`. Obtain a platform-appropriate license if necessary.
```
SystemError: <built-in method addVariable of xpress.problem object at 0x...> returned NULL without setting an error
```
cause This generic Python error typically masks a more specific underlying Xpress API error, such as attempting to add a variable with a duplicate name or providing invalid arguments. The true error message is often logged to the console.

fix Check the terminal or IDE console output immediately preceding this error for a more descriptive message (e.g., '1030 Error: Duplicate column names are not allowed'). Correct the modeling mistake based on the detailed console error.

Warnings

gotcha The `xpress` library, while installable via pip, requires an underlying FICO Xpress license to solve problems. A 'Community License' is included, but it imposes strict limits on problem size (e.g., sum of variables and linear constraints <= 5000, nonlinear tokens <= 1000).
Fix: For larger problems, a commercial FICO Xpress license is necessary. Ensure the `XPRESS` environment variable points to the directory containing your `xpauth.xpr` license file.
gotcha Generic `SystemError` or `RuntimeError` messages like `<built-in method [...] of xpress.problem object at [...]> returned NULL without setting an error` are frequently encountered in Jupyter notebooks or similar environments. These often obscure the actual, more descriptive error.
Fix: Always check the console output or logs alongside the Python exception. The console typically contains the precise error message (e.g., 'Duplicate column names are not allowed').
breaking Model properties on variable, constraint, and SOS objects (e.g., `var.lb`, `constraint.rhs`) are not accessible from within callbacks or while the problem is in a presolved state.
Fix: Access these properties outside of callbacks or the presolve phase. If values are needed within a callback, store them in auxiliary data structures before the solve begins.
deprecated Several older Xpress Python module versions (specifically 8.14.1, and 8.14.5 through 8.14.8) were removed from PyPI and Conda repositories on April 22, 2024.
Fix: Users are advised to upgrade to Xpress 8.14.9 or a newer major release (e.g., 9.x). Archived versions may be available in the FICO Client area.

Install

pip install xpress Install from PyPI

Imports

xpress
```
import xpress as xp
```
Aliasing as 'xp' is a widely adopted convention for brevity and readability.

Quickstart

This quickstart defines and solves a simple Linear Programming (LP) problem using the FICO Xpress Python interface. It initializes a problem, adds variables and constraints, sets an objective, and then optimizes the problem, printing the solution if optimal.

import xpress as xp

p = xp.problem(name='my_first_lp')

x1 = p.addVariable(name='x1', lb=0, ub=xp.infinity)
x2 = p.addVariable(name='x2', lb=0, ub=xp.infinity)

p.setObjective(3*x1 + 2*x2, sense=xp.minimize)

p.addConstraint(4*x1 + 2*x2 >= 10, name='c1')
p.addConstraint(x1 + x2 >= 3, name='c2')

p.optimize()

if p.getProbStatus() == xp.lp_optimal:
    print(f"Solution Status: Optimal")
    print(f"Objective Value: {p.getObjVal()}")
    print(f"x1 = {p.getSolution(x1)}")
    print(f"x2 = {p.getSolution(x2)}")
else:
    print(f"Solution Status: {p.getProbStatusString()}")

view raw JSON →