adds changes to test_problem with parametrize

Author: Transurgeon

CLAUDE.md

@@ -65,7 +65,7 @@ prob.solve(nlp=True, solver=cp.IPOPT, best_of=5)
 
 | Solver | License | Installation |
 |--------|---------|--------------|
-| [IPOPT](https://github.com/coin-or/Ipopt) | EPL-2.0 | `conda install -c conda-forge cyipopt` |
+| [IPOPT](https://github.com/coin-or/Ipopt) | EPL-2.0 | `conda install -c conda-forge cyipopt` (conda only — pip install is unreliable) |
 | [Knitro](https://www.artelys.com/solvers/knitro/) | Commercial | `pip install knitro` (requires license) |
 | [COPT](https://www.copt.de/) | Commercial | Requires license |
 | [Uno](https://github.com/cuter-testing/uno) | Open source | See Uno documentation |
@@ -129,9 +129,9 @@ Key reduction classes in `cvxpy/reductions/`:
 - `Chain` composes multiple reductions
 - `SolvingChain` orchestrates the full solve process
 
-For DNLP: `CvxAttr2Constr` → `Dnlp2Smooth` → `NLPSolver`
+For DNLP: `FlipObjective` (if Maximize) → `CvxAttr2Constr` → `Dnlp2Smooth` → `NLPSolver`
 
-Note: The standard `SolvingChain` in `solving_chain.py` handles DCP/DGP/DQCP solver selection automatically. NLP solving is triggered explicitly via `prob.solve(nlp=True)` and bypasses the standard chain.
+Note: The standard `SolvingChain` in `solving_chain.py` handles DCP/DGP/DQCP solver selection automatically. NLP solving is triggered explicitly via `prob.solve(nlp=True)` and bypasses the standard chain — `problem.py` delegates to `solve_nlp()` in `cvxpy/reductions/solvers/nlp_solving_chain.py`, which builds the chain, handles initial point construction (via `_set_nlp_initial_point`), and orchestrates the `best_of` multi-start logic. Solver algorithm variants (e.g., `"knitro_sqp"`) are mapped via `NLP_SOLVER_VARIANTS` in `nlp_solving_chain.py`.
 
 ### Solver Categories
 
@@ -215,6 +215,10 @@ class TestMyFeature(BaseTest):
 
 NLP tests are in `cvxpy/tests/nlp_tests/` with Jacobian and Hessian verification tests.
 
+## Build System
+
+Uses a custom build backend (`setup/build_meta.py`) that re-exports `setuptools.build_meta`. The `setup/` directory handles C extensions (cvxcore, sparsecholesky) and version management. Solver registration is in `cvxpy/reductions/solvers/defines.py` (`SOLVER_MAP_CONIC`, `SOLVER_MAP_QP`, `SOLVER_MAP_NLP`).
+
 ## Benchmarks
 
 Benchmarks use [Airspeed Velocity](https://asv.readthedocs.io/) and live in the `benchmarks/` directory. To run locally:

cvxpy/tests/nlp_tests/test_problem.py

@@ -1,91 +1,31 @@
 
 import numpy as np
+import pytest
 
 import cvxpy as cp
 from cvxpy.reductions.solvers.nlp_solving_chain import _set_nlp_initial_point
 
 
 class TestProblem():
-    """
-    This class can be used to test internal function for Problem that have been added
-    in the DNLP extension.
-    """
-
-    def test_set_initial_point_both_bounds_infinity(self):
-        # when both bounds are infinity, the initial point should be zero vector
-
-        # test 1
-        x = cp.Variable((3, ))
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == np.zeros((3, ))).all()
-
-        # test 2
-        x = cp.Variable((3, ), bounds=[None, None])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == np.zeros((3, ))).all()
-
-        # test 3
-        x = cp.Variable((3, ), bounds=[-np.inf, np.inf])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == np.zeros((3, ))).all()
-
-        # test 4
-        x = cp.Variable((3, ), bounds=[None, np.inf])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == np.zeros((3, ))).all()
-
-        # test 5
-        x = cp.Variable((3, ), bounds=[-np.inf, None])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == np.zeros((3, ))).all()
-
-
-    def test_set_initial_point_lower_bound_infinity(self):
-        # when one bound is infinity, the initial point should be one unit
-        # away from the finite bound
-
-        # test 1
-        x = cp.Variable((3, ), bounds=[None, 3.5])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == 2.5 * np.ones((3, ))).all()
-
-        # test 2
-        x = cp.Variable((3, ), bounds=[-np.inf, 3.5])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == 2.5 * np.ones((3, ))).all()
-
-    def test_set_initial_point_upper_bound_infinity(self):
-        # when one bound is infinity, the initial point should be one unit
-        # away from the finite bound
-
-        # test 1
-        x = cp.Variable((3, ), bounds=[3.5, None])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == 4.5 * np.ones((3, ))).all()
-
-        # test 2
-        x = cp.Variable((3, ), bounds=[3.5, np.inf])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == 4.5 * np.ones((3, ))).all()
-
-    def test_set_initial_point_both_bounds_finite(self):
-        # when both bounds are finite, the initial point should be the midpoint
-        # between the two bounds
-
-        # test 1
-        x = cp.Variable((3, ), bounds=[3.5, 4.5])
-        prob = cp.Problem(cp.Minimize(cp.sum(x)))
-        _set_nlp_initial_point(prob)
-        assert (x.value == 4.0 * np.ones((3, ))).all()
+    """Tests for internal NLP functions in the DNLP extension."""
+
+    @pytest.mark.parametrize("bounds, expected", [
+        (None, 0.0),
+        ([None, None], 0.0),
+        ([-np.inf, np.inf], 0.0),
+        ([None, np.inf], 0.0),
+        ([-np.inf, None], 0.0),
+        ([None, 3.5], 2.5),
+        ([-np.inf, 3.5], 2.5),
+        ([3.5, None], 4.5),
+        ([3.5, np.inf], 4.5),
+        ([3.5, 4.5], 4.0),
+    ])
+    def test_set_initial_point_scalar_bounds(self, bounds, expected):
+        x = cp.Variable((3, ), bounds=bounds)
+        prob = cp.Problem(cp.Minimize(cp.sum(x)))
+        _set_nlp_initial_point(prob)
+        assert (x.value == expected * np.ones((3, ))).all()
 
     def test_set_initial_point_mixed_inf_and_finite(self):
         lb = np.array([-np.inf, 3.5, -np.inf, -1.5, 2, 2.5])