linopy

0.6.7 verified Fri May 01 auth: no python

Linear optimization with N-D labeled arrays in Python. Integrates with xarray and supports multiple solvers (HiGHS, Gurobi, CPLEX, etc.). Current version 0.6.7, released occasionally with bugfixes and performance improvements.

pip install linopy

Common errors

error NoSolverAvailable: No solver found. Install a solver like highspy, gurobipy, or cplex, or set the `solver` argument. ↓

cause No supported solver is installed or accessible in the environment.

fix

Install highspy: pip install highspy. Or install gurobipy if you have a license.

error ValueError: cannot add variable without a name ↓

cause You forgot to provide the name parameter in add_variables() or add_constraints().

fix

Always specify name='myvar' when adding variables or constraints.

error TypeError: unhashable type: 'DataArray' ↓

cause Using a DataArray as a dictionary key or index where hashable type is expected.

fix

Convert to tuple or string before use, or use .values.

Warnings

breaking In linopy v0.6.0, the auto_mask parameter was added to the Model class, changing its behavior for masked variables/constraints. Previously mask handling could be implicit; now the default is auto_mask=True which may mask out variables/constraints with zero coefficients. Set auto_mask=False to restore previous behavior. ↓

fix Set auto_mask=False when creating Model if you rely on unmasked behavior: Model(auto_mask=False)

deprecated The use of xarray.DataArray for coefficients in add_variables and add_constraints is encouraged over numpy arrays. Passing raw numpy arrays may lead to unexpected broadcasting or lack of named dimensions. ↓

fix Use xr.DataArray with proper dims/coords instead of np.ndarray.

gotcha Scalar getitem on LinearExpression or Variable (e.g., expr[0]) raised FutureWarning in older versions and may be removed in future. Use .values or .isel() instead. ↓

fix Replace var[0] with var.values[0] or var.isel(i=0) where i is the dimension name.

gotcha Linopy does not automatically install a solver. If no solver is found, solve() will raise an error. Common mistake: pip install linopy without solver, then call solve(). ↓

fix Install a solver like HiGHS (pip install highspy) or Gurobi, and ensure it is on PATH or properly configured.

Install

pip install linopy[gurobi]

conda install -c conda-forge linopy

Imports

Model
wrong
```
from linopy.model import Model
```
correct
```
from linopy import Model
```
Model is exported at package level; using submodule path may break if internal structure changes.

Variable

wrong

from linopy.variables import Variable

correct

from linopy import Variable

Variable is part of the public API at top-level.

LinearExpression

wrong

from linopy.expressions import LinearExpression

correct

from linopy import LinearExpression

LinearExpression is exported at package level.

Quickstart

Creates a simple linear optimization model with one variable, a constraint, and an objective. Solves with default solver (HiGHS if installed).

from linopy import Model
import xarray as xr

m = Model()
# Create a variable with coords
x = m.add_variables(name='x', lower=0, coords=[('i', [0,1,2])])
# Add constraint
m.add_constraints(x.sum('i') <= 10, name='total')
# Add objective
m.add_objective(x.sum('i'))
# Solve
m.solve()
# Retrieve solution
print(x.solution)