def model():
m = Model()
lower = xr.DataArray(np.zeros((10, 10)), coords=[range(10), range(10)])
upper = xr.DataArray(np.ones((10, 10)), coords=[range(10), range(10)])
m.add_variables(lower, upper, name="x")
m.add_variables(lower, upper, name="y")
return m
def test_constraint_repr():
m = Model()
x = m.add_variables()
c = m.add_constraints(x, ">=", 0)
c.__repr__()
c._repr_html_()
def test_variable_modification():
m = Model()
x = m.add_variables(0, 10)
x.upper = 20
assert x.upper.item() == 20
x.lower = 8
assert x.lower == 8
def test_str_arrays_chunked():
m = Model(chunk="auto")
x = m.add_variables(4, pd.Series([8, 10]))
y = m.add_variables(0, pd.DataFrame([[1, 2], [3, 4], [5, 6]]).T)
da = int_to_str(y.compute().values)
assert da.dtype == object
def test_str_arrays():
m = Model()
x = m.add_variables(4, pd.Series([8, 10]))
y = m.add_variables(0, pd.DataFrame([[1, 2], [3, 4], [5, 6]]).T)
da = int_to_str(x.values)
assert da.dtype == object
def test_variable_sanitize():
m = Model()
x = m.add_variables(coords=[range(10)])
# convert intentionally to float with nans
x = x.where([True] * 4 + [False] * 6, np.nan)
x = x.sanitize()
assert isinstance(x, linopy.variables.Variable)
assert x.loc[9].item() == -1
def test_variable_type_preservation():
m = Model()
x = m.add_variables(coords=[range(10)])
assert isinstance(x.bfill("dim_0"), linopy.variables.Variable)
assert isinstance(x.broadcast_like(x.to_array()),
linopy.variables.Variable)
assert isinstance(x.clip(max=20), linopy.variables.Variable)
assert isinstance(x.ffill("dim_0"), linopy.variables.Variable)
assert isinstance(x.fillna(-1), linopy.variables.Variable)
def test_str_arrays_with_nans():
m = Model()
x = m.add_variables(4, pd.Series([8, 10]), name="x")
# now expand the second dimension, expended values of x will be nan
y = m.add_variables(0, pd.DataFrame([[1, 2], [3, 4], [5, 6]]), name="y")
assert m["x"][-1].item() == -1
da = int_to_str(m["x"].values)
assert da.dtype == object
def create_model(N):
m = Model()
coords = [arange(N), arange(N)]
x = m.add_variables(coords=coords)
y = m.add_variables(coords=coords)
m.add_constraints(x - y >= arange(N))
m.add_constraints(x + y >= 0)
m.add_objective((2 * x).sum() + y.sum())
return m
def masked_variable_model():
m = Model()
lower = pd.Series(0, range(10))
x = m.add_variables(lower, name="x")
mask = pd.Series([True] * 8 + [False, False])
y = m.add_variables(lower, name="y", mask=mask)
m.add_constraints(x + y, ">=", 10)
m.add_constraints(y, ">=", 0)
m.add_objective(2 * x + y)
return m
def model_chunked():
m = Model(chunk="auto")
x = m.add_variables(name="x")
y = m.add_variables(name="y")
m.add_constraints(2 * x + 6 * y, ">=", 10)
m.add_constraints(4 * x + 2 * y, ">=", 3)
m.add_objective(2 * y + x)
return m
def model_anonymous_constraint():
m = Model(chunk=None)
x = m.add_variables(name="x")
y = m.add_variables(name="y")
m.add_constraints(2 * x + 6 * y >= 10)
m.add_constraints(4 * x + 2 * y >= 3)
m.add_objective(2 * y + x)
return m
def test_variable_modification_M():
m = Model()
lower = pd.Series(0, index=range(10))
upper = pd.Series(range(10, 20), index=range(10))
x = m.add_variables(lower, upper)
new_upper = pd.Series(range(25, 35), index=range(10))
x.upper = new_upper
assert isinstance(x.upper, xr.DataArray)
assert (x.upper == new_upper).all()
new_lower = pd.Series(range(15, 25), index=range(10))
x.lower = new_lower
assert isinstance(x.lower, xr.DataArray)
assert (x.lower == new_lower).all()
def test_to_blocks(tmp_path):
m = Model()
lower = pd.Series(range(20))
upper = pd.Series(range(30, 50))
x = m.add_variables(lower, upper)
y = m.add_variables(lower, upper)
m.add_constraints(x + y, "<=", 10)
m.add_objective(2 * x + 3 * y)
m.blocks = xr.DataArray([1] * 10 + [2] * 10)
m.to_block_files(tmp_path)
def test_variable_repr():
m = Model()
m.variables.__repr__()
x = m.add_variables()
x.__repr__()
x._repr_html_()
m.variables.__repr__()
y = m.add_variables(coords=[pd.Index([1, 2, 3], name="time")], name="y")
y.__repr__()
y._repr_html_()
m.variables.__repr__()
def test_get_name_by_label():
m = Model()
x = m.add_variables(coords=[range(10)])
y = m.add_variables(coords=[range(10)])
m.add_constraints(x + y <= 10, name="first")
m.add_constraints(x - y >= 5, name="second")
assert m.constraints.get_name_by_label(4) == "first"
assert m.constraints.get_name_by_label(14) == "second"
with pytest.raises(ValueError):
m.constraints.get_name_by_label(30)
with pytest.raises(ValueError):
m.constraints.get_name_by_label("first")
def test_to_file(tmp_path):
import gurobipy
m = Model()
x = m.add_variables(4, pd.Series([8, 10]))
y = m.add_variables(0, pd.DataFrame([[1, 2], [3, 4], [5, 6]]))
m.add_constraints(x + y, "<=", 10)
m.add_objective(2 * x + 3 * y)
fn = tmp_path / "test.lp"
m.to_file(fn)
gurobipy.read(str(fn))
def model_with_inf():
m = Model()
lower = pd.Series(0, range(10))
x = m.add_variables(coords=[lower.index], name="x", binary=True)
y = m.add_variables(lower, name="y")
m.add_constraints(x + y, ">=", 10)
m.add_constraints(1 * x, "<=", np.inf)
m.objective = 2 * x + y
return m
def model(n, solver, integerlabels):
m = Model()
if integerlabels:
N, M = [arange(n), arange(n)]
else:
N, M = [arange(n).astype(float), arange(n).astype(str)]
x = m.add_variables(coords=[N, M])
y = m.add_variables(coords=[N, M])
m.add_constraints(x - y >= N)
m.add_constraints(x + y >= 0)
m.add_objective((2 * x).sum() + y.sum())
m.solve(solver)
return
def test_to_netcdf(tmp_path):
m = Model()
x = m.add_variables(4, pd.Series([8, 10]))
y = m.add_variables(0, pd.DataFrame([[1, 2], [3, 4], [5, 6]]))
m.add_constraints(x + y, "<=", 10)
m.add_objective(2 * x + 3 * y)
fn = tmp_path / "test.nc"
m.to_netcdf(fn)
p = read_netcdf(fn)
for k in m.scalar_attrs:
if k != "objective_value":
assert getattr(m, k) == getattr(p, k)
for k in m.dataset_attrs:
assert_equal(getattr(m, k), getattr(p, k))
def masked_constraint_model():
m = Model()
lower = pd.Series(0, range(10))
x = m.add_variables(lower, name="x")
y = m.add_variables(lower, name="y")
mask = pd.Series([True] * 8 + [False, False])
m.add_constraints(x + y, ">=", 10, mask=mask)
# for the last two entries only the following constraint will be active
m.add_constraints(x + y, ">=", 5)
m.add_objective(2 * x + y)
return m
def milp_model_r():
m = Model()
lower = pd.Series(0, range(10))
x = m.add_variables(coords=[lower.index], name="x", binary=True)
y = m.add_variables(lower, name="y")
m.add_constraints(x + y, ">=", 10)
m.add_objective(2 * x + y)
return m
def test_constraint_accessor():
m = Model()
x = m.add_variables()
y = m.add_variables()
c = m.add_constraints(x, ">=", 0)
assert c.rhs.item() == 0
assert c.vars.item() == 0
assert c.coeffs.item() == 1
assert c.sign.item() == ">="
c.rhs = 2
assert c.rhs.item() == 2
c.vars = y
assert c.vars.item() == 1
c.coeffs = 3
assert c.coeffs.item() == 3
c.sign = "=="
assert c.sign.item() == "=="
def test_constraint_matrix():
m = Model()
x = m.add_variables(coords=[range(10)])
y = m.add_variables()
m.add_constraints(x, "=", 0)
A = m.constraints.to_matrix()
assert A.shape == (10, 11)
def test_constraint_accessor_M():
m = Model()
lower = pd.Series(range(10), range(10))
upper = pd.Series(range(10, 20), range(10))
x = m.add_variables(lower, upper)
y = m.add_variables(lower, upper)
c = m.add_constraints(x, ">=", 0)
assert c.rhs.item() == 0
assert (c.rhs == 0).all()
assert c.vars.shape == (10, 1)
assert c.coeffs.shape == (10, 1)
assert c.sign.item() == ">="
c.rhs = 2
assert c.rhs.item() == 2
c.lhs = 3 * y
assert (c.vars.squeeze() == y.data).all()
assert (c.coeffs == 3).all()
assert isinstance(c.lhs, linopy.LinearExpression)
def test_nvars():
m = Model()
m.add_variables(coords=[range(10)])
assert m.variables.nvars == 10
mask = pd.Series([True] * 5 + [False] * 5)
m.add_variables(coords=[range(10)], mask=mask)
assert m.variables.nvars == 15
def test_constraint_matrix_masked_constraints_and_variables():
"""
Test constraint matrix with missing constraints.
"""
# now with missing variables
m = Model()
mask = pd.Series([False] * 5 + [True] * 5)
x = m.add_variables(coords=[range(10)], mask=mask)
m.add_variables()
m.add_constraints(x, "=", 0, mask=mask)
A = m.constraints.to_matrix(filter_missings=True)
assert A.shape == (5, 6)
assert A.shape == (m.ncons, m.nvars)
A = m.constraints.to_matrix(filter_missings=False)
assert A.shape == (m._cCounter, m._xCounter)
def test_get_name_by_label():
m = Model()
m.add_variables(coords=[range(10)], name="x")
m.add_variables(coords=[range(10)], name="asd")
assert m.variables.get_name_by_label(4) == "x"
assert m.variables.get_name_by_label(14) == "asd"
with pytest.raises(ValueError):
m.variables.get_name_by_label(30)
with pytest.raises(ValueError):
m.variables.get_name_by_label("asd")
def test_constraints_accessor():
m = Model()
lower = xr.DataArray(np.zeros((10, 10)), coords=[range(10), range(10)])
upper = xr.DataArray(np.ones((10, 10)), coords=[range(10), range(10)])
x = m.add_variables(lower, upper)
y = m.add_variables()
m.add_constraints(1 * x + 10 * y, "=", 0)
assert m.constraints["con0"].shape == (10, 10)
assert isinstance(m.constraints[["con0"]], linopy.constraints.Constraints)
assert isinstance(m.constraints.inequalities,
linopy.constraints.Constraints)
assert isinstance(m.constraints.equalities, linopy.constraints.Constraints)
def modified_model():
m = Model()
lower = pd.Series(0, range(10))
x = m.add_variables(coords=[lower.index], name="x", binary=True)
y = m.add_variables(lower, name="y")
c = m.add_constraints(x + y, ">=", 10)
y.lower = 9
c.lhs = 2 * x + y
m.objective = 2 * x + y
return m
