Add PyNumeroEvaluationError

pyomo/contrib/pynumero/asl.py

@@ -11,6 +11,7 @@
 
 from pyomo.common.fileutils import find_library
 from pyomo.common.dependencies import numpy as np
+from pyomo.contrib.pynumero.exceptions import PyNumeroEvaluationError
 import ctypes
 import logging
 import os
@@ -364,7 +365,8 @@ def eval_f(self, x):
         res = self.ASLib.EXTERNAL_AmplInterface_eval_f(
             self._obj, x, self._nx, ctypes.byref(sol)
         )
-        assert res, "Error in AMPL evaluation"
+        if not res:
+            raise PyNumeroEvaluationError("Error in AMPL evaluation")
         return sol.value
 
     def eval_deriv_f(self, x, df):
@@ -373,7 +375,8 @@ def eval_deriv_f(self, x, df):
             x.dtype == np.double
         ), "Error: array type. Function eval_deriv_f expects an array of type double"
         res = self.ASLib.EXTERNAL_AmplInterface_eval_deriv_f(self._obj, x, df, len(x))
-        assert res, "Error in AMPL evaluation"
+        if not res:
+            raise PyNumeroEvaluationError("Error in AMPL evaluation")
 
     def struct_jac_g(self, irow, jcol):
         irow_p = irow.astype(np.intc, casting='safe', copy=False)
@@ -409,7 +412,8 @@ def eval_jac_g(self, x, jac_g_values):
         res = self.ASLib.EXTERNAL_AmplInterface_eval_jac_g(
             self._obj, xeval, self._nx, jac_eval, self._nnz_jac_g
         )
-        assert res, "Error in AMPL evaluation"
+        if not res:
+            raise PyNumeroEvaluationError("Error in AMPL evaluation")
 
     def eval_g(self, x, g):
         assert x.size == self._nx, "Error: Dimension mismatch."
@@ -423,7 +427,8 @@ def eval_g(self, x, g):
         res = self.ASLib.EXTERNAL_AmplInterface_eval_g(
             self._obj, x, self._nx, g, self._ny
         )
-        assert res, "Error in AMPL evaluation"
+        if not res:
+            raise PyNumeroEvaluationError("Error in AMPL evaluation")
 
     def eval_hes_lag(self, x, lam, hes_lag, obj_factor=1.0):
         assert x.size == self._nx, "Error: Dimension mismatch."
@@ -453,7 +458,8 @@ def eval_hes_lag(self, x, lam, hes_lag, obj_factor=1.0):
             res = self.ASLib.EXTERNAL_AmplInterface_eval_hes_lag(
                 self._obj, x, self._nx, lam, self._ny, hes_lag, self._nnz_hess
             )
-        assert res, "Error in AMPL evaluation"
+        if not res:
+            raise PyNumeroEvaluationError("Error in AMPL evaluation")
 
     def finalize_solution(self, ampl_solve_status_num, msg, x, lam):
         b_msg = msg.encode('utf-8')

pyomo/contrib/pynumero/exceptions.py

@@ -0,0 +1,20 @@
+#  ___________________________________________________________________________
+#
+#  Pyomo: Python Optimization Modeling Objects
+#  Copyright (c) 2008-2022
+#  National Technology and Engineering Solutions of Sandia, LLC
+#  Under the terms of Contract DE-NA0003525 with National Technology and
+#  Engineering Solutions of Sandia, LLC, the U.S. Government retains certain
+#  rights in this software.
+#  This software is distributed under the 3-clause BSD License.
+#  ___________________________________________________________________________
+
+
+class PyNumeroEvaluationError(ArithmeticError):
+    """An exception to be raised by PyNumero evaluation backends in the event
+    of a failed function evaluation. This should be caught by solver interfaces
+    and translated to the solver-specific evaluation error API.
+
+    """
+
+    pass

pyomo/contrib/pynumero/interfaces/tests/test_nlp.py

@@ -22,6 +22,7 @@
     raise unittest.SkipTest("Pynumero needs scipy and numpy to run NLP tests")
 
 from pyomo.contrib.pynumero.asl import AmplInterface
+from pyomo.contrib.pynumero.exceptions import PyNumeroEvaluationError
 
 if not AmplInterface.available():
     raise unittest.SkipTest("Pynumero needs the ASL extension to run NLP tests")
@@ -816,6 +817,57 @@ def test_util_maps(self):
         self.assertTrue(np.array_equal(expected_full_primals_lb, full_primals_lb))
 
 
+class TestExceptions(unittest.TestCase):
+    def _make_bad_model(self):
+        m = pyo.ConcreteModel()
+        m.I = pyo.Set(initialize=[1, 2, 3])
+        m.x = pyo.Var(m.I, initialize=1)
+
+        m.obj = pyo.Objective(expr=m.x[1] + m.x[2] / m.x[3])
+        m.eq1 = pyo.Constraint(expr=m.x[1] == pyo.sqrt(m.x[2]))
+        return m
+
+    def test_eval_error_in_constraint(self):
+        m = self._make_bad_model()
+        m.x[2] = -1
+        nlp = PyomoNLP(m)
+        msg = "Error in AMPL evaluation"
+        with self.assertRaisesRegex(PyNumeroEvaluationError, msg):
+            residuals = nlp.evaluate_constraints()
+
+    def test_eval_error_in_constraint_jacobian(self):
+        m = self._make_bad_model()
+        m.x[2] = -1
+        nlp = PyomoNLP(m)
+        msg = "Error in AMPL evaluation"
+        with self.assertRaisesRegex(PyNumeroEvaluationError, msg):
+            jacobian = nlp.evaluate_jacobian()
+
+    def test_eval_error_in_objective(self):
+        m = self._make_bad_model()
+        m.x[3] = 0
+        nlp = PyomoNLP(m)
+        msg = "Error in AMPL evaluation"
+        with self.assertRaisesRegex(PyNumeroEvaluationError, msg):
+            objval = nlp.evaluate_objective()
+
+    def test_eval_error_in_objective_gradient(self):
+        m = self._make_bad_model()
+        m.x[3] = 0
+        nlp = PyomoNLP(m)
+        msg = "Error in AMPL evaluation"
+        with self.assertRaisesRegex(PyNumeroEvaluationError, msg):
+            gradient = nlp.evaluate_grad_objective()
+
+    def test_eval_error_in_lagrangian_hessian(self):
+        m = self._make_bad_model()
+        m.x[3] = 0
+        nlp = PyomoNLP(m)
+        msg = "Error in AMPL evaluation"
+        with self.assertRaisesRegex(PyNumeroEvaluationError, msg):
+            hessian = nlp.evaluate_hessian_lag()
+
+
 if __name__ == '__main__':
     TestAslNLP.setUpClass()
     t = TestAslNLP()