Merge branch 'public25.09.1' into 'main'

Release 25.09.1 : sync with internal repo1 (commit cd0ea6c42)

See merge request cuda-hpc-libraries/cuquantum-sdk/cuquantum-public!40

Author: daniellowell

python/README.md

@@ -52,9 +52,9 @@ Runtime dependencies of the cuQuantum Python package include:
 * An NVIDIA GPU with compute capability 7.5+
 * Driver: Linux (525.60.13+ for CUDA 12, 580.65.06+ for CUDA 13)
 * CUDA Toolkit 12.x or 13.x
-* cuStateVec 1.10.0+
-* cuTensorNet 2.9.0+
-* cuDensityMat >=0.3.0, <0.4.0
+* cuStateVec 1.10.1+
+* cuTensorNet 2.9.1+
+* cuDensityMat >=0.3.1, <0.4.0
 * Python >=3.11, <3.14
 * NumPy v1.21+
 * nvmath-python ==0.6.0

python/cuquantum/_version.py

@@ -5,4 +5,4 @@
 # Note: cuQuantum Python follows the cuQuantum SDK version, which is now
 # switched to YY.MM and is different from individual libraries' (semantic)
 # versioning scheme.
-__version__ = '25.09.0'
+__version__ = '25.09.1'

python/extensions/cuquantum/densitymat/jax/cppsrc/CMakeLists.txt

@@ -8,7 +8,7 @@ project(cudensitymat_jax LANGUAGES CXX)
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
-find_package(Python3 REQUIRED REQUIRED COMPONENTS Interpreter Development)
+find_package(Python3 REQUIRED COMPONENTS Interpreter Development)
 message(STATUS "Python executable: ${Python3_EXECUTABLE}")
 
 find_package(CUDAToolkit REQUIRED)
@@ -43,7 +43,7 @@ find_package(pybind11 REQUIRED)
 
 # FIXME: This should be made more robust and moved to setup.py.
 execute_process(
-    COMMAND ${Python3_EXECUTABLE} -c "import site; print(f'{site.getsitepackages()[0]}/cuquantum')"
+    COMMAND ${Python3_EXECUTABLE} -c "import os; import cuquantum; print(os.path.dirname(cuquantum.__file__))"
     OUTPUT_STRIP_TRAILING_WHITESPACE
     OUTPUT_VARIABLE CUQUANTUM_PYTHON_ROOT
 )
@@ -85,6 +85,12 @@ else()
     message(STATUS "cuDensityMat library: ${CUDENSITYMAT_LIBRARY}")
 endif()
 
+set_target_properties(
+    ${PROJECT_NAME}
+    PROPERTIES
+        BUILD_RPATH "$ORIGIN"
+)
+
 target_link_libraries(
     ${PROJECT_NAME}
     PRIVATE

python/extensions/pyproject.toml

@@ -1,3 +1,3 @@
 [build-system]
-requires = ["setuptools>=77.0.3", "wheel", "build", "jax[cuda12-local]>=0.5,<0.7", "pybind11"]
+requires = ["setuptools>=77.0.3", "wheel", "build", "jax[cuda12-local]>=0.5,<0.7", "pybind11", "cuquantum-python-cu12~=25.09"]
 build-backend = "setuptools.build_meta"

python/extensions/setup.py

@@ -75,7 +75,7 @@ def finalize_options(self):
 with open(os.path.join(os.path.dirname(__file__), "README.md"), encoding="utf-8") as f:
     long_description = f.read()
 
-__version__ = "0.0.1"
+__version__ = "0.0.2"
 
 setup(
     name="cuquantum-python-jax",

python/samples/tensornet/experimental/network_state/generic_states/example06_noisy_unitary_channels_mps.py

@@ -0,0 +1,56 @@
+# Copyright (c) 2025, NVIDIA CORPORATION & AFFILIATES
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+"""
+MPS simulation of a noisy quantum state with unitary tensor channels.
+
+The sampling is drawn with a trajectory based simulation.
+"""
+import numpy as np
+
+from cuquantum.tensornet.experimental import NetworkState, MPSConfig
+
+
+def random_unitary(n):
+    """
+    Create a random unitary tensor
+    """
+    mat = np.random.random((2**n, 2**n)) + 1.j * np.random.random((2**n, 2**n))
+    q, r = np.linalg.qr(mat)
+    unitary = q.reshape((2, 2) * n)
+    return unitary
+
+
+N = 20
+
+mps_config = MPSConfig(max_extent=256,
+                       rel_cutoff=1e-5,
+                       abs_cutoff=1e-5,
+                       algorithm='gesvdj')
+state_mode_extents = (2,) * N
+dtype = 'complex128'
+state = NetworkState(state_mode_extents, dtype=dtype, config=mps_config)
+
+np.random.seed(1)
+
+h_op = (1 / np.sqrt(2)) * np.array([[1, 1], [1, -1]]).astype("complex128")
+
+for i in range(N):
+    modes_one_body = (i,)
+    tensor_id = state.apply_tensor_operator(modes_one_body,
+                                            h_op,
+                                            unitary=True,
+                                            immutable=True)
+
+    for j in range(i + 1, N):
+        u = random_unitary(2)
+        modes_two_body = (i, j)
+        tensor_id = state.apply_tensor_operator(modes_two_body,
+                                                u,
+                                                unitary=True,
+                                                immutable=True)
+for num_trajectories in range(2):
+    trajectory_sample = state.compute_sampling(100)
+
+state.free() 

python/setup.py

@@ -36,9 +36,9 @@
     'numpy>=1.21, <3.0',  # ">=1.21,<3"
     'nvmath-python==0.6.0', # strict version before nvmath.internal module is stable
     # 'torch', # <-- PyTorch is optional; also, the PyPI version does not support GPU...
-    f'custatevec-cu{utils.cuda_major_ver}~=1.10',  # ">=1.10.0,<2"
-    f'cutensornet-cu{utils.cuda_major_ver}~=2.9',  # ">=2.9.0,<3"
-    f'cudensitymat-cu{utils.cuda_major_ver}~=0.3.0', # ">=0.3.0,<0.4.0"
+    f'custatevec-cu{utils.cuda_major_ver}>=1.10.1, <2',  # ">=1.10.1,<2"
+    f'cutensornet-cu{utils.cuda_major_ver}>=2.9.1, <3',  # ">=2.9.1,<3"
+    f'cudensitymat-cu{utils.cuda_major_ver}>=0.3.1, <0.4', # ">=0.3.1,<0.4.0"
 ]
 if utils.cuda_major_ver == '12':
     install_requires.append('cupy-cuda12x>=13.0')  # no ambiguity

python/tests/cuquantum_tests/tensornet/trajectories_noise/test_quantum_volume_mid_circuit.py

@@ -8,7 +8,7 @@
 This test uses TrajectorySim API to simulate quantum volumue circuits
 """
 
-from qiskit.circuit.library import QuantumVolume
+from qiskit.circuit.library import quantum_volume
 from qiskit import transpile
 
 import numpy as np
@@ -38,7 +38,7 @@
 
 
 def get_qvolume_circuit(n_qubits, depth, seed=SEED):
-    circuit = QuantumVolume(n_qubits, depth, seed=seed)
+    circuit = quantum_volume(n_qubits, depth, seed=seed)
     circuit.measure_all()
     circuit = transpile(circuit, basis_gates=["u3", "cx"], optimization_level=0)
     return circuit