Sync DNLP fork with CVXPY upstream master

Absorbs 12 upstream commits: type hints modernization, SvecPSD cone,
power refactoring, mul_bounds NaN fix, and various bugfixes. All 8 merge
conflicts resolved in favor of DNLP (parameter support, chain rule
canonicalizers, negative power exponents). Lint fixes applied.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: Transurgeon

.git-blame-ignore-revs

@@ -0,0 +1,8 @@
+# Removes white space from blank lines (#3264)
+bebc9e0b17decfcdeebb37e6d72a6333bf42f7bb
+
+# Modernize type hints and add pyright baseline (#3173)
+99b21e3ea481d5a47c125bfde23c032462ce82d7
+
+# Add trailing whitespace and missing newline ruff rules (#3177)
+8f839a4be777a569dd49d94d92cb975fdc0b9198

.github/workflows/gcsopt.yml

@@ -8,7 +8,7 @@ on:
         tags:
           - '*'
 jobs:
-  test_backends:
+  test_gcsopt:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/setup-python@v6

.github/workflows/pyright.yml

@@ -0,0 +1,25 @@
+name: pyright
+
+on:
+  pull_request:
+  push:
+    branches:
+      - master
+
+jobs:
+  pyright:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v5
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v7
+        with:
+          python-version: "3.11"
+          enable-cache: true
+
+      - name: Install dependencies
+        run: uv sync --no-install-project
+
+      - name: Run pyright
+        run: uvx pyright

.github/workflows/scorecards.yml

@@ -51,6 +51,6 @@ jobs:
       # Upload the results to GitHub's code scanning dashboard (optional).
       # Commenting out will disable upload of results to your repo's Code Scanning dashboard
       - name: "Upload to code-scanning"
-        uses: github/codeql-action/upload-sarif@b1bff81932f5cdfc8695c7752dcee935dcd061c8 # v2.16.4
+        uses: github/codeql-action/upload-sarif@c10b8064de6f491fea524254123dbe5e09572f13 # v2.16.4
         with:
           sarif_file: results.sarif

.pre-commit-config.yaml

@@ -24,3 +24,4 @@ repos:
     rev: v0.23
     hooks:
       - id: validate-pyproject
+

cvxpy/atoms/__init__.py

@@ -155,6 +155,7 @@
     sigma_max,
     tr_inv,
     quantum_rel_entr,
+    von_neumann_entr,
 ]
 
 NONPOS_ATOMS = [

cvxpy/atoms/affine/add_expr.py

@@ -14,8 +14,9 @@
 limitations under the License.
 """
 import operator as op
+from collections.abc import Iterable
 from functools import reduce
-from typing import Any, Iterable, List, Tuple
+from typing import Any
 
 import numpy as np
 
@@ -38,12 +39,12 @@ def __init__(self, arg_groups: Iterable[Expression]) -> None:
         for group in arg_groups:
             self.args += self.expand_args(group)
 
-    def shape_from_args(self) -> Tuple[int, ...]:
+    def shape_from_args(self) -> tuple[int, ...]:
         """Returns the (row, col) shape of the expression.
         """
         return u.shape.sum_shapes([arg.shape for arg in self.args])
 
-    def expand_args(self, expr: Expression) -> List[Expression]:
+    def expand_args(self, expr: Expression) -> list[Expression]:
         """Helper function to extract the arguments from an AddExpression.
         """
         if isinstance(expr, AddExpression):
@@ -56,13 +57,13 @@ def name(self) -> str:
         for i in range(1, len(self.args)):
             result += " + " + str(self.args[i])
         return result
-    
+
     def format_labeled(self):
         """Format addition with labels where available."""
         # Check for own label first
         if self._label is not None:
             return self._label
-        
+
         # Build from sub-expressions using their labels
         result = self.args[0].format_labeled()
         for i in range(1, len(self.args)):
@@ -82,7 +83,7 @@ def is_atom_log_log_concave(self) -> bool:
         """
         return False
 
-    def bounds_from_args(self) -> Tuple[np.ndarray, np.ndarray]:
+    def bounds_from_args(self) -> tuple[np.ndarray, np.ndarray]:
         """Returns bounds for addition based on argument bounds."""
         # Start with first argument's bounds
         lb, ub = self.args[0].get_bounds()
@@ -133,8 +134,8 @@ def copy(self, args=None, id_objects=None) -> 'AddExpression':
         return copy
 
     def graph_implementation(
-        self, arg_objs, shape: Tuple[int, ...], data=None
-    ) -> Tuple[lo.LinOp, List[Constraint]]:
+        self, arg_objs, shape: tuple[int, ...], data=None
+    ) -> tuple[lo.LinOp, list[Constraint]]:
         """Sum the linear expressions.
 
         Parameters

cvxpy/atoms/affine/affine_atom.py

@@ -13,7 +13,7 @@
 See the License for the specific language governing permissions and
 limitations under the License.
 """
-from typing import Any, List, Tuple
+from typing import Any
 
 import scipy.sparse as sp
 
@@ -30,7 +30,7 @@ class AffAtom(Atom):
     """ Abstract base class for affine atoms. """
     _allow_complex = True
 
-    def sign_from_args(self) -> Tuple[bool, bool]:
+    def sign_from_args(self) -> tuple[bool, bool]:
         """By default, the sign is the most general of all the argument signs.
         """
         return u.sign.sum_signs([arg for arg in self.args])
@@ -112,7 +112,7 @@ def is_nsd(self) -> bool:
                 return False
         return True
 
-    def _grad(self, values) -> List[Any]:
+    def _grad(self, values) -> list[Any]:
         """Gives the (sub/super)gradient of the atom w.r.t. each argument.
 
         Matrix expressions are vectorized, so the gradient is a matrix.

cvxpy/atoms/affine/binary_operators.py

@@ -16,7 +16,6 @@
 
 import operator as op
 from functools import reduce
-from typing import List, Tuple
 
 import numpy as np
 import scipy.sparse as sp
@@ -64,13 +63,13 @@ def name(self):
             else:
                 pretty_args.append(a.name())
         return pretty_args[0] + ' ' + self.OP_NAME + ' ' + pretty_args[1]
-    
+
     def format_labeled(self):
         """Format binary operation with labels where available."""
         # Check for own label first
         if self._label is not None:
             return self._label
-        
+
         # Build from sub-expressions using their labels
         pretty_args = []
         for i, a in enumerate(self.args):
@@ -90,7 +89,7 @@ def numeric(self, values):
         """
         return reduce(self.OP_FUNC, values)
 
-    def sign_from_args(self) -> Tuple[bool, bool]:
+    def sign_from_args(self) -> tuple[bool, bool]:
         """Default to rules for times.
         """
         return u.sign.mul_sign(self.args[0], self.args[1])
@@ -107,7 +106,7 @@ def is_complex(self) -> bool:
         return (self.args[0].is_complex() or self.args[1].is_complex()) and \
             not (self.args[0].is_imag() and self.args[1].is_imag())
 
-    def bounds_from_args(self) -> Tuple[np.ndarray, np.ndarray]:
+    def bounds_from_args(self) -> tuple[np.ndarray, np.ndarray]:
         """Returns bounds for elementwise multiplication based on argument bounds."""
         lb1, ub1 = self.args[0].get_bounds()
         lb2, ub2 = self.args[1].get_bounds()
@@ -192,7 +191,7 @@ def numeric(self, values):
         else:
             return values[0] @ values[1]
 
-    def shape_from_args(self) -> Tuple[int, ...]:
+    def shape_from_args(self) -> tuple[int, ...]:
         """Returns the (row, col) shape of the expression.
         """
         return u.shape.mul_shapes(self.args[0].shape, self.args[1].shape)
@@ -235,7 +234,7 @@ def is_atom_log_log_concave(self) -> bool:
         """
         return False
 
-    def bounds_from_args(self) -> Tuple[np.ndarray, np.ndarray]:
+    def bounds_from_args(self) -> tuple[np.ndarray, np.ndarray]:
         """Returns bounds for matrix multiplication based on argument bounds."""
         lb1, ub1 = self.args[0].get_bounds()
         lb2, ub2 = self.args[1].get_bounds()
@@ -312,8 +311,8 @@ def _grad(self, values):
         return [DX, DY]
 
     def graph_implementation(
-        self, arg_objs, shape: Tuple[int, ...], data=None
-    ) -> Tuple[lo.LinOp, List[Constraint]]:
+        self, arg_objs, shape: tuple[int, ...], data=None
+    ) -> tuple[lo.LinOp, list[Constraint]]:
         """Multiply the linear expressions.
 
         Parameters
@@ -359,7 +358,7 @@ def is_atom_log_log_concave(self) -> bool:
         """Is the atom log-log concave?"""
         return True
 
-    def bounds_from_args(self) -> Tuple[np.ndarray, np.ndarray]:
+    def bounds_from_args(self) -> tuple[np.ndarray, np.ndarray]:
         """Returns bounds for elementwise multiplication based on argument bounds.
 
         Overrides MulExpression to use elementwise multiplication bounds
@@ -397,7 +396,7 @@ def validate_arguments(self):
             np.empty(self.args[1].shape, dtype=np.dtype([]))
         )
 
-    def shape_from_args(self) -> Tuple[int, ...]:
+    def shape_from_args(self) -> tuple[int, ...]:
         """Call np.broadcast on multiply arguments."""
         return np.broadcast_shapes(self.args[0].shape, self.args[1].shape)
 
@@ -448,8 +447,8 @@ def _grad(self, values):
         return [DX, DY]
 
     def graph_implementation(
-        self, arg_objs, shape: Tuple[int, ...], data=None
-    ) -> Tuple[lo.LinOp, List[Constraint]]:
+        self, arg_objs, shape: tuple[int, ...], data=None
+    ) -> tuple[lo.LinOp, list[Constraint]]:
         """Multiply the expressions elementwise.
 
         Parameters
@@ -509,7 +508,7 @@ def has_quadratic_term(self) -> bool:
     def is_qpwa(self) -> bool:
         return self.args[0].is_qpwa() and self.args[1].is_constant()
 
-    def shape_from_args(self) -> Tuple[int, ...]:
+    def shape_from_args(self) -> tuple[int, ...]:
         """Returns the (row, col) shape of the expression.
         """
         return self.args[0].shape
@@ -539,7 +538,7 @@ def is_atom_quasiconvex(self) -> bool:
     def is_atom_quasiconcave(self) -> bool:
         return self.is_atom_quasiconvex()
 
-    def bounds_from_args(self) -> Tuple[np.ndarray, np.ndarray]:
+    def bounds_from_args(self) -> tuple[np.ndarray, np.ndarray]:
         """Returns bounds for division based on argument bounds."""
         lb1, ub1 = self.args[0].get_bounds()
         lb2, ub2 = self.args[1].get_bounds()
@@ -562,8 +561,8 @@ def is_decr(self, idx) -> bool:
             return self.args[0].is_nonneg()
 
     def graph_implementation(
-        self, arg_objs, shape: Tuple[int, ...], data=None
-    ) -> Tuple[lo.LinOp, List[Constraint]]:
+        self, arg_objs, shape: tuple[int, ...], data=None
+    ) -> tuple[lo.LinOp, list[Constraint]]:
         """Multiply the linear expressions.
 
         Parameters
@@ -646,7 +645,7 @@ def outer(x, y):
     y = Expression.cast_to_const(y)
     if y.ndim > 1:
         raise ValueError("y must be a 1-d array.")
-    
+
     x = reshape(x, (x.size, 1), order='F')
     y = reshape(y, (1, y.size), order='F')
     return x @ y

cvxpy/atoms/affine/broadcast_to.py

@@ -14,7 +14,6 @@
 limitations under the License.
 """
 
-from typing import List, Optional, Tuple
 
 import numpy as np
 
@@ -44,7 +43,7 @@ def is_atom_log_log_concave(self) -> bool:
     def numeric(self, values):
         return np.broadcast_to(values[0], shape=self.broadcast_shape)
 
-    def get_data(self) -> List[Optional[int]]:
+    def get_data(self) -> list[int | None]:
         return [self.broadcast_shape]
 
     def validate_arguments(self) -> None:
@@ -53,15 +52,15 @@ def validate_arguments(self) -> None:
             shape=self.broadcast_shape
         )
 
-    def shape_from_args(self) -> Tuple[int, ...]:
+    def shape_from_args(self) -> tuple[int, ...]:
         return self.broadcast_shape
 
     def graph_implementation(
         self,
         arg_objs,
-        shape: Tuple[int, ...],
+        shape: tuple[int, ...],
         data=None,
-    ) -> Tuple[lo.LinOp, List[Constraint]]:
+    ) -> tuple[lo.LinOp, list[Constraint]]:
         """Broadcast an expression to a given shape.
 
         Parameters