Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Resolve PyNumero incompatibilities with NumPy2 #3408

Merged
merged 21 commits into from
Nov 7, 2024
Merged
Show file tree
Hide file tree
Changes from 17 commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions .github/workflows/test_pr_and_main.yml
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,15 @@ jobs:
TARGET: linux
PYENV: pip

- os: ubuntu-latest
python: 3.12
other: /numpy2
slim: 1
skip_doctest: 1
TARGET: linux
PYENV: pip
PACKAGES: "gurobipy dill numpy>2.0 scipy networkx"

- os: ubuntu-latest
python: 3.9
other: /pyutilib
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -15,10 +15,10 @@
from pyomo.core.expr.visitor import identify_variables
import pyomo.environ as pyo

from pyomo.common.dependencies import networkx_available as nx_available
from pyomo.contrib.pynumero.dependencies import (
numpy as np,
numpy_available,
scipy,
scipy_available,
)

Expand Down Expand Up @@ -151,6 +151,7 @@ def flow_out_eqn(m, t):


class TestExternalGreyBoxBlock(unittest.TestCase):
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_construct_scalar(self):
m = pyo.ConcreteModel()
m.ex_block = ExternalGreyBoxBlock(concrete=True)
Expand All @@ -171,6 +172,7 @@ def test_construct_scalar(self):
self.assertEqual(len(block.outputs), 0)
self.assertEqual(len(block._equality_constraint_names), 2)

@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_construct_indexed(self):
block = ExternalGreyBoxBlock([0, 1, 2], concrete=True)
self.assertIs(type(block), IndexedExternalGreyBoxBlock)
Expand All @@ -192,6 +194,7 @@ def test_construct_indexed(self):
self.assertEqual(len(b._equality_constraint_names), 2)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_solve_square(self):
m = pyo.ConcreteModel()
m.ex_block = ExternalGreyBoxBlock(concrete=True)
Expand Down Expand Up @@ -234,6 +237,7 @@ def test_solve_square(self):
self.assertAlmostEqual(m_ex.y.value, y.value, delta=1e-8)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_optimize(self):
m = pyo.ConcreteModel()
m.ex_block = ExternalGreyBoxBlock(concrete=True)
Expand Down Expand Up @@ -292,6 +296,7 @@ def test_optimize(self):
self.assertAlmostEqual(m_ex.y.value, y.value, delta=1e-8)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_optimize_with_cyipopt_for_inner_problem(self):
# Use CyIpopt, rather than the default SciPy solvers,
# for the inner problem
Expand Down Expand Up @@ -427,6 +432,7 @@ def test_optimize_no_decomposition(self):
self.assertAlmostEqual(m_ex.x.value, x.value, delta=1e-8)
self.assertAlmostEqual(m_ex.y.value, y.value, delta=1e-8)

@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_construct_dynamic(self):
m = make_dynamic_model()
time = m.time
Expand Down Expand Up @@ -504,6 +510,7 @@ def test_construct_dynamic(self):
)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_solve_square_dynamic(self):
# Create the "external model"
m = make_dynamic_model()
Expand Down Expand Up @@ -571,6 +578,7 @@ def linking_constraint_rule(m, i, t):
self.assertStructuredAlmostEqual(values, target_values, delta=1e-5)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_optimize_dynamic(self):
# Create the "external model"
m = make_dynamic_model()
Expand Down Expand Up @@ -653,6 +661,7 @@ def linking_constraint_rule(m, i, t):
self.assertStructuredAlmostEqual(values, target_values, delta=1e-5)

@unittest.skipUnless(cyipopt_available, "cyipopt is not available")
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
def test_optimize_dynamic_references(self):
"""
When when pre-existing variables are attached to the EGBB
Expand Down Expand Up @@ -717,7 +726,8 @@ def test_optimize_dynamic_references(self):
self.assertStructuredAlmostEqual(values, target_values, delta=1e-5)


class TestPyomoNLPWithGreyBoxBLocks(unittest.TestCase):
@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
class TestPyomoNLPWithGreyBoxBlocks(unittest.TestCase):
def test_set_and_evaluate(self):
m = pyo.ConcreteModel()
m.ex_block = ExternalGreyBoxBlock(concrete=True)
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
import pyomo.common.unittest as unittest
import pyomo.environ as pyo

from pyomo.common.dependencies import networkx_available as nx_available
from pyomo.contrib.pynumero.dependencies import (
numpy as np,
numpy_available,
Expand Down Expand Up @@ -513,6 +514,7 @@ def test_explicit_zeros(self):
np.testing.assert_allclose(hess.data, data, rtol=1e-8)


@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
class TestExternalPyomoModel(unittest.TestCase):
def test_evaluate_SimpleModel1(self):
model = SimpleModel1()
Expand Down Expand Up @@ -838,6 +840,7 @@ def test_evaluate_hessian_lagrangian_SimpleModel2x2_1(self):
np.testing.assert_allclose(hess_lag, expected_hess_lag, rtol=1e-8)


@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
class TestUpdatedHessianCalculationMethods(unittest.TestCase):
"""
These tests exercise the methods for fast Hessian-of-Lagrangian
Expand Down Expand Up @@ -1021,6 +1024,7 @@ def test_evaluate_hessian_equality_constraints_order(self):
)


@unittest.skipUnless(nx_available, "SCCImplicitFunctionSolver requires networkx")
class TestScaling(unittest.TestCase):
def con_3_body(self, x, y, u, v):
return 1e5 * x**2 + 1e4 * y**2 + 1e1 * u**2 + 1e0 * v**2
Expand Down
127 changes: 127 additions & 0 deletions pyomo/contrib/pynumero/sparse/base_block.py
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,8 @@

# These classes are for checking types consistently and raising errors

from ..dependencies import numpy as np


class BaseBlockVector(object):
"""Base class for block vectors"""
Expand Down Expand Up @@ -177,3 +179,128 @@ def transpose(self, *axes):
def tostring(self, order='C'):
msg = "tostring not implemented for {}".format(self.__class__.__name__)
raise NotImplementedError(msg)


#: NumPy ufuncs that take one vector and are compatible with pyNumero vectors
vec_unary_ufuncs = {
## MATH ufuncs
np.negative,
np.positive,
np.absolute,
np.fabs,
np.rint,
np.sign,
np.conj,
np.conjugate,
np.exp,
np.exp2,
np.log,
np.log2,
np.log10,
np.expm1,
np.log1p,
np.sqrt,
np.square,
np.cbrt,
np.reciprocal,
## TRIG ufuncs
np.sin,
np.cos,
np.tan,
np.arcsin,
np.arccos,
np.arctan,
np.sinh,
np.cosh,
np.tanh,
np.arcsinh,
np.arccosh,
np.arctanh,
np.degrees,
np.radians,
np.deg2rad,
np.rad2deg,
## COMPARISON ufuncs
np.logical_not,
## BIT-TWIDDLING ufuncs
np.invert,
## FLOATING ufuncs
np.isfinite,
np.isinf,
np.isnan,
# np.isnat, # only defined for datetime
np.fabs, # numpy docs list here and in MATH
np.signbit,
np.spacing,
# np.modf, # disabled because shape is not preserved
# np.frexp, # disabled because shape is not preserved
np.floor,
np.ceil,
np.trunc,
# OTHER (not listed in ufuncs docs)
np.abs,
}

#: NumPy ufuncs that take two vectors and are compatible with pyNumero vectors
vec_binary_ufuncs = {
## MATH ufuncs
np.add,
np.subtract,
np.multiply,
# np.matmult, # disabled because shape is not preserved
np.divide,
np.logaddexp,
np.logaddexp2,
np.true_divide,
np.floor_divide,
np.power,
np.float_power,
np.remainder,
np.mod,
np.fmod,
# np.divmod, # dieabled because shape is not preserved
mrmundt marked this conversation as resolved.
Show resolved Hide resolved
np.heaviside,
np.gcd,
np.lcm,
## TRIG ufuncs
np.arctan2,
np.hypot,
## BIT-TWIDDLING ufuncs
np.bitwise_and,
np.bitwise_or,
np.bitwise_xor,
np.left_shift,
np.right_shift,
## COMPARISON ufuncs
np.greater,
np.greater_equal,
np.less,
np.less_equal,
np.not_equal,
np.equal,
np.logical_and,
np.logical_or,
np.logical_xor,
np.maximum,
np.minimum,
np.fmax,
np.fmin,
## FLOATING ufincs
np.copysign,
np.nextafter,
np.ldexp,
np.fmod, # numpy docs list here and in MATH
}

#: NumPy ufuncs can be used as reductions for pyNumero vectors
vec_associative_reductions = {
np.add,
np.multiply,
np.bitwise_and,
np.bitwise_or,
np.bitwise_xor,
np.maximum,
np.minimum,
np.fmax,
np.fmin,
}
Loading
Loading