Add Python CUDA dependencies as Lightning-GPU and Lightning-Tensor wh…

…eel dependencies (#1025)

### Before submitting

Please complete the following checklist when submitting a PR:

- [ ] All new features must include a unit test.
If you've fixed a bug or added code that should be tested, add a test to
the
      [`tests`](../tests) directory!

- [ ] All new functions and code must be clearly commented and
documented.
If you do make documentation changes, make sure that the docs build and
      render correctly by running `make docs`.

- [ ] Ensure that the test suite passes, by running `make test`.

- [X] Add a new entry to the `.github/CHANGELOG.md` file, summarizing
the
      change, and including a link back to the PR.

- [ ] Ensure that code is properly formatted by running `make format`. 

When all the above are checked, delete everything above the dashed
line and fill in the pull request template.


------------------------------------------------------------------------------------------------------------

**Context:**

**Description of the Change:**
Added all [Python CUDA
dependencies](https://github.com/PennyLaneAI/pennylane-lightning/blob/954a6e7d12befc043597769c6357a1003e8d588b/bin/auditwheel)
to `pyproject.toml` for LGPU and LT. `INSTALL_RPATH` updated in CMake
files.

**Benefits:**

**Possible Drawbacks:**

**Related GitHub Issues:**


[sc-79647]

---------

Co-authored-by: ringo-but-quantum <[email protected]>
Co-authored-by: Shuli Shu <[email protected]>

.github/CHANGELOG.md

@@ -29,8 +29,8 @@
 
 ### Improvements
 
-* Optimize lightning.tensor by adding direct MPS sites data set with `qml.MPSPrep`.
-  [(#983)](https://github.com/PennyLaneAI/pennylane-lightning/pull/983)
+* Add CUDA dependencies to Lightning GPU and Lightning Tensor Python wheels.
+  [(#1025)](https://github.com/PennyLaneAI/pennylane-lightning/pull/1025/)
 
 * Replace the `dummy_tensor_update` method with the `cutensornetStateCaptureMPS`API to ensure that further gates apply is allowed after the `cutensornetStateCompute` call.
   [(#1028)](https://github.com/PennyLaneAI/pennylane-lightning/pull/1028/)
@@ -85,6 +85,9 @@
 * The TOML files for the devices are updated to use the new schema for declaring device capabilities.
   [(#988)](https://github.com/PennyLaneAI/pennylane-lightning/pull/988)
 
+* Optimize lightning.tensor by adding direct MPS sites data set with `qml.MPSPrep`.
+  [(#983)](https://github.com/PennyLaneAI/pennylane-lightning/pull/983)
+
 * Unify excitation gates memory layout to row-major for both LGPU and LT.
   [(#959)](https://github.com/PennyLaneAI/pennylane-lightning/pull/959)
 

.github/workflows/tests_lgpumpi_cpp.yml

@@ -51,7 +51,7 @@ jobs:
         mpilib: ["mpich", "openmpi"]
         cuda_version_maj: ["12"]
         cuda_version_min: ["2"]
-    timeout-minutes: 30
+    timeout-minutes: 35
 
     steps:
       - name: Checkout

pennylane_lightning/core/_version.py

@@ -16,4 +16,4 @@
    Version number (major.minor.patch[-label])
 """
 
-__version__ = "0.40.0-dev42"
+__version__ = "0.40.0-dev43"

pennylane_lightning/core/src/simulators/lightning_gpu/CMakeLists.txt

@@ -63,7 +63,7 @@ set_property(TARGET ${PL_BACKEND} PROPERTY POSITION_INDEPENDENT_CODE ON)
 
 # To avoid DSO errors on platforms preferring static linkage, uncomment the following line:
 # string(REPLACE "libcudart_static.a" "libcudart.so" CUDA_SHARED_RT "${CUDA_LIBRARIES}")
-set_target_properties(${PL_BACKEND} PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/")
+set_target_properties(${PL_BACKEND} PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/../nvidia/cublas/lib:$ORIGIN/../nvidia/cusparse/lib:$ORIGIN/../nvidia/nvjitlink/lib:$ORIGIN/../nvidia/cuda_runtime/lib:$ORIGIN/")
 
 if(PL_DISABLE_CUDA_SAFETY)
     target_compile_options(${PL_BACKEND} INTERFACE $<$<COMPILE_LANGUAGE:CXX>:-DCUDA_UNSAFE>)
@@ -87,4 +87,4 @@ endforeach()
 if (BUILD_TESTS)
     enable_testing()
     add_subdirectory("tests")
-endif()
+endif()

pennylane_lightning/core/src/simulators/lightning_tensor/tncuda/CMakeLists.txt

@@ -51,7 +51,7 @@ set_property(TARGET ${PL_BACKEND} PROPERTY POSITION_INDEPENDENT_CODE ON)
 
 # To avoid DSO errors on platforms preferring static linkage, uncomment the following line:
 # string(REPLACE "libcudart_static.a" "libcudart.so" CUDA_SHARED_RT "${CUDA_LIBRARIES}")
-set_target_properties(${PL_BACKEND} PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/")
+set_target_properties(${PL_BACKEND} PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/../nvidia/cublas/lib:$ORIGIN/../nvidia/cusparse/lib:$ORIGIN/../cutensor/lib:$ORIGIN/../nvidia/nvjitlink/lib:$ORIGIN/../nvidia/cusolver/lib:$ORIGIN/../nvidia/cuda_runtime/lib:$ORIGIN/")
 
 if(PL_DISABLE_CUDA_SAFETY)
     target_compile_options(${PL_BACKEND} INTERFACE $<$<COMPILE_LANGUAGE:CXX>:-DCUDA_UNSAFE>)

requirements-dev.txt

@@ -16,8 +16,8 @@ isort==5.13.2
 click==8.0.4
 cmake
 jax[cpu]==0.4.28
-custatevec-cu12
-cutensornet-cu12
+custatevec-cu12; sys_platform == "linux"
+cutensornet-cu12; sys_platform == "linux"
 pylint==2.7.4
 scipy-openblas32>=0.3.26
 git+https://github.com/PennyLaneAI/pennylane.git@master

scripts/configure_pyproject_toml.py

@@ -122,6 +122,20 @@ def parse_args():
         "scipy-openblas32>=0.3.26",
     ]
 
+    if backend == "lightning_gpu":
+        dependencies += ["custatevec-cu12"]
+
+    if backend == "lightning_tensor":
+        dependencies += ["cutensornet-cu12", "nvidia-cusolver-cu12"]
+
+    if backend in ("lightning_gpu", "lightning_tensor"):
+        dependencies += [
+            "nvidia-nvjitlink-cu12",
+            "nvidia-cusparse-cu12",
+            "nvidia-cublas-cu12",
+            "nvidia-cuda-runtime-cu12",
+        ]
+
     if backend != "lightning_qubit":
         dependencies += ["pennylane_lightning==" + version]