Rename qubitstatevector to stateprep (#134)

* rename qubitstatevector to stateprep

* default.qubit as bench

* Trigger multi-GPU CI

* add changelog

* update pybind to 2.11.1 (test)

* Fix typo in changelog.

* Revert pybind11 version to match lightning. Fix couple imports. Add formatting tools to reqs.

* Deprecate python 3.8.

* Build pennylane-lightning@update/nested_cmake_paths

* Build against update/nested_cmake_paths

* Revert to building against PLL@master

* WIP

* Change template input of templated OpsData.

* replace StateVectorRawCPU.hpp StateVectorLQubitRaw.hpp

* Increase CUDA ver to 20.

* Fix cpp tests headers and linking.

* Update pennylane ver in docs.

* Update cuda ver 12.

* Install devtoolset-11

* Uninstall devtoolset-10

* Build against PLL bugfix/numbers.

* Build against pennylane-lightning.git@bugfix/numbers

* pull pennylane-lightning.git@bugfix/numbers

* Reintroduce QubitStateVector.

* Reformat & fix pytest.raise.

* Remove python 3.8 refs.

* Change Lightning build branch from bugfix/numbers to master.

* Update CHANGELOG.md

* Update devices.rst

* Fix custatevec ver in reqs.

* Parametrize couple parallel tests over stateprep.

---------

Co-authored-by: Vincent Michaud-Rioux <[email protected]>

.github/CHANGELOG.md

@@ -7,6 +7,15 @@
 
 ### Breaking changes
 
+ * Rename `QubitStateVector` to `StatePrep` in the `LightningGPU` class.
+ [#134] (https://github.com/PennyLaneAI/pennylane-lightning-gpu/pull/134)
+
+ * Deprecate Python 3.8.
+ [#134] (https://github.com/PennyLaneAI/pennylane-lightning-gpu/pull/134)
+
+ * Update PennyLane-Lightning imports following the [(#472)] (https://github.com/PennyLaneAI/pennylane-lightning/pull/472) refactoring.
+ [#134] (https://github.com/PennyLaneAI/pennylane-lightning-gpu/pull/134)
+
 ### Improvements
 
  * Optimizes the single qubit rotation gate by using a single cuStateVector API call instead of separate Pauli gate applications. 
@@ -23,7 +32,7 @@
 
 This release contains contributions from (in alphabetical order):
 
-David Clark (NVIDIA), Shuli Shu
+David Clark (NVIDIA), Vincent Michaud-Rioux, Shuli Shu
 
 ---
 

.github/workflows/build_wheel_manylinux2014.yml

@@ -1,7 +1,7 @@
 name: Wheel::Linux::x86_64
 
 # **What it does**: Builds python wheels for Linux (ubuntu-latest) architecture x86_64 and store it as artifacts.
-#                   Python versions: 3.8, 3.9, 3.10, 3.11.
+#                   Python versions: 3.9, 3.10, 3.11.
 # **Why we have it**: To build wheels for pennylane-lightning-gpu installation.
 # **Who does it impact**: Wheels to be uploaded to PyPI.
 
@@ -27,7 +27,7 @@ jobs:
         os: [ubuntu-latest]
         arch: [x86_64]
         cibw_build: ${{ fromJson(needs.set_wheel_build_matrix.outputs.python_version) }}
-    name: ${{ matrix.os }} (Python ${{ fromJson('{"cp38-*":"3.8","cp39-*":"3.9","cp310-*":"3.10","cp311-*":"3.11" }')[matrix.cibw_build] }})
+    name: ${{ matrix.os }} (Python ${{ fromJson('{ "cp39-*":"3.9","cp310-*":"3.10","cp311-*":"3.11" }')[matrix.cibw_build] }})
     runs-on: ${{ matrix.os }}
 
     steps:
@@ -42,7 +42,7 @@ jobs:
       - name: Setup Python
         uses: actions/setup-python@v4
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       - name: Install cibuildwheel
         run: python -m pip install cibuildwheel~=2.11.0

.github/workflows/format.yml

@@ -17,7 +17,7 @@ jobs:
       - uses: actions/setup-python@v4
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       - name: Install dependencies
         run:

.github/workflows/tests_linux_x86.yml

@@ -69,7 +69,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -215,7 +215,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -259,7 +259,7 @@ jobs:
           python -m pip install pip~=22.0
           python -m pip install ninja cmake custatevec-cu11 pytest pytest-mock flaky pytest-cov
           # Sync with latest master branches
-          python -m pip install --index-url https://test.pypi.org/simple/ pennylane-lightning --pre --force-reinstall --no-deps
+          python -m pip install git+https://github.com/PennyLaneAI/pennylane-lightning.git@master --force-reinstall --no-deps
 
       - name: Build and install package
         env: 

.github/workflows/tests_linux_x86_mpich.yml

@@ -55,7 +55,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -225,7 +225,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -275,7 +275,7 @@ jobs:
           python -m pip install pip~=22.0
           python -m pip install ninja cmake custatevec-cu11 pytest pytest-mock flaky pytest-cov mpi4py
           # Sync with latest master branches
-          python -m pip install --index-url https://test.pypi.org/simple/ pennylane-lightning --pre --force-reinstall --no-deps
+          python -m pip install git+https://github.com/PennyLaneAI/pennylane-lightning.git@master --force-reinstall --no-deps
           
       - name: Build and install package (MPICH backend)
         env: 

.github/workflows/tests_linux_x86_openmpi.yml

@@ -38,7 +38,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -197,7 +197,7 @@ jobs:
         id: setup_python
         name: Install Python
         with:
-          python-version: '3.8'
+          python-version: '3.9'
 
       # Since the self-hosted runner can be re-used. It is best to set up all package
       # installations in a virtual environment that gets cleaned at the end of each workflow run
@@ -247,7 +247,7 @@ jobs:
           python -m pip install pip~=22.0
           python -m pip install ninja cmake custatevec-cu11 pytest pytest-mock flaky pytest-cov mpi4py
           # Sync with latest master branches
-          python -m pip install --index-url https://test.pypi.org/simple/ pennylane-lightning --pre --force-reinstall --no-deps
+          python -m pip install git+https://github.com/PennyLaneAI/pennylane-lightning.git@master --force-reinstall --no-deps
           
       - name: Build and install package (OpenMPI backend)
         env: 

CMakeLists.txt

@@ -109,7 +109,7 @@ set(ENABLE_KOKKOS OFF)
 FetchContent_Declare(
     pennylane_lightning
     GIT_REPOSITORY https://github.com/PennyLaneAI/pennylane-lightning.git
-    GIT_TAG       "${LIGHTNING_RELEASE_TAG}"
+    GIT_TAG        master
 )
 FetchContent_MakeAvailable(pennylane_lightning)
 
@@ -200,12 +200,12 @@ endif()
 
 # Create binding module
 if(PLLGPU_ENABLE_PYTHON)
-    pybind11_add_module(lightning_gpu_qubit_ops     "pennylane_lightning_gpu/src/bindings/Bindings.cpp")
-    target_link_libraries(lightning_gpu_qubit_ops PRIVATE pennylane_lightning_gpu)
-    set_target_properties(lightning_gpu_qubit_ops PROPERTIES CXX_VISIBILITY_PRESET hidden)
-    set_target_properties(lightning_gpu_qubit_ops PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/")
-    target_compile_options(lightning_gpu_qubit_ops PRIVATE "$<$<CONFIG:RELEASE>:-W>")
-    target_compile_definitions(lightning_gpu_qubit_ops PRIVATE VERSION_INFO=${VERSION_STRING})
+    pybind11_add_module(lightning_gpu_qubit_ops "pennylane_lightning_gpu/src/bindings/Bindings.cpp")
+    target_link_libraries(lightning_gpu_qubit_ops   PRIVATE lightning_algorithms lightning_qubit lightning_qubit_algorithms pennylane_lightning_gpu)
+    set_target_properties(lightning_gpu_qubit_ops   PROPERTIES CXX_VISIBILITY_PRESET hidden)
+    set_target_properties(lightning_gpu_qubit_ops   PROPERTIES INSTALL_RPATH "$ORIGIN/../cuquantum/lib:$ORIGIN/../cuquantum/lib64:$ORIGIN/")
+    target_compile_options(lightning_gpu_qubit_ops  PRIVATE "$<$<CONFIG:RELEASE>:-W>")
+    target_compile_definitions(lightning_gpu_qubit_ops  PRIVATE VERSION_INFO=${VERSION_STRING})
     if(PLLGPU_ENABLE_MPI)
         option(ENABLE_MPI "Enable MPI support" ON)
         target_compile_definitions(lightning_gpu_qubit_ops PRIVATE ENABLE_MPI)

Makefile

@@ -30,7 +30,7 @@ ifndef CUQUANTUM_SDK
 	@test $(CUQUANTUM_SDK)
 endif
 ifndef PYTHON3
-	@echo "To install PennyLane-Lightning-GPU you must have Python 3.8+ installed."
+	@echo "To install PennyLane-Lightning-GPU you must have Python 3.9+ installed."
 endif
 	$(PYTHON) setup.py build_ext --cuquantum=$(CUQUANTUM_SDK) --verbose
 	$(PYTHON) setup.py install
@@ -85,13 +85,13 @@ ifndef CUQUANTUM_SDK
 	@test $(CUQUANTUM_SDK)
 endif
 	rm -rf ./BuildTests
-	cmake . -BBuildTests -DBUILD_TESTS=1 -DPLLGPU_BUILD_TESTS=1 -DCUQUANTUM_SDK=$(CUQUANTUM_SDK)
+	cmake . -BBuildTests -G Ninja -DBUILD_TESTS=1 -DPLLGPU_BUILD_TESTS=1 -DCUQUANTUM_SDK=$(CUQUANTUM_SDK)
 	cmake --build ./BuildTests
 	./BuildTests/pennylane_lightning_gpu/src/tests/runner_gpu
 
 test-cpp-mpi:
 	rm -rf ./BuildTests
-	cmake . -BBuildTests -DBUILD_TESTS=1 -DPLLGPU_BUILD_TESTS=1 -DPLLGPU_ENABLE_MPI=On -DCUQUANTUM_SDK=$(CUQUANTUM_SDK)
+	cmake . -BBuildTests -G Ninja -DBUILD_TESTS=1 -DPLLGPU_BUILD_TESTS=1 -DPLLGPU_ENABLE_MPI=On -DCUQUANTUM_SDK=$(CUQUANTUM_SDK)
 	cmake --build ./BuildTests
 	$(MPILAUNCHER) -np $(NUMPROCS) ./BuildTests/pennylane_lightning_gpu/src/tests/mpi_runner
 
@@ -121,5 +121,5 @@ endif
 .PHONY: check-tidy
 check-tidy:
 	rm -rf ./Build
-	cmake . -BBuild -DENABLE_CLANG_TIDY=ON -DBUILD_TESTS=1
+	cmake . -BBuild -G Ninja -DENABLE_CLANG_TIDY=ON -DBUILD_TESTS=1
 	cmake --build ./Build

README.rst

@@ -37,7 +37,7 @@ Features
 Installation
 ============
 
-PennyLane-Lightning-GPU requires Python version 3.8 and above. It can be installed using ``pip``:
+PennyLane-Lightning-GPU requires Python version 3.9 and above. It can be installed using ``pip``:
 
 .. code-block:: console
 
@@ -82,7 +82,7 @@ To build using Docker, run the following from the project root directory:
 
     docker build . -f ./docker/Dockerfile -t "lightning-gpu-wheels"
 
-This will build a Python wheel for Python 3.8 up to 3.11 inclusive, and be manylinux2014 (glibc 2.17) compatible.
+This will build a Python wheel for Python 3.9 up to 3.11 inclusive, and be manylinux2014 (glibc 2.17) compatible.
 To acquire the built wheels, use:
 
 .. code-block:: console

doc/devices.rst

@@ -47,6 +47,7 @@ Supported operations and observables
     ~pennylane.PhaseShift
     ~pennylane.ControlledPhaseShift
     ~pennylane.QubitStateVector
+    ~pennylane.StatePrep
     ~pennylane.Rot
     ~pennylane.RX
     ~pennylane.RY
@@ -241,4 +242,4 @@ To enable the memory-optimized adjoint method with MPI support, ``batch_obs`` sh
     
     dev = qml.device('lightning.gpu', wires= n_wires, mpi=True, batch_obs=True)
 
-For the adjoint method, each MPI process will provide the overall simulation results.
+For the adjoint method, each MPI process will provide the overall simulation results.

doc/requirements.txt

@@ -6,5 +6,5 @@ exhale
 pybind11
 cmake
 ninja
-pennylane
+git+https://github.com/PennyLaneAI/pennylane.git@master
 pennylane-sphinx-theme

mpitests/test_adjoint_jacobian.py

@@ -174,7 +174,8 @@ def test_unsupported_hermitian_expectation(self, isBatch_obs):
     @pytest.mark.parametrize("theta", np.linspace(-2 * np.pi, 2 * np.pi, 7))
     @pytest.mark.parametrize("G", [qml.RX, qml.RY, qml.RZ])
     @pytest.mark.parametrize("isBatch_obs", [False, True])
-    def test_pauli_rotation_gradient(self, G, theta, tol, isBatch_obs, request):
+    @pytest.mark.parametrize("stateprep", [qml.QubitStateVector, qml.StatePrep])
+    def test_pauli_rotation_gradient(self, stateprep, G, theta, tol, isBatch_obs, request):
         """Tests that the automatic gradients of Pauli rotations are correct."""
 
         num_wires = 3
@@ -188,7 +189,7 @@ def test_pauli_rotation_gradient(self, G, theta, tol, isBatch_obs, request):
         dev_cpu = qml.device("default.qubit", wires=3)
 
         with qml.tape.QuantumTape() as tape:
-            qml.QubitStateVector(np.array([1.0, -1.0]) / np.sqrt(2), wires=0)
+            stateprep(np.array([1.0, -1.0]) / np.sqrt(2), wires=0)
             G(theta, wires=[0])
             qml.expval(qml.PauliZ(0))
 
@@ -202,7 +203,8 @@ def test_pauli_rotation_gradient(self, G, theta, tol, isBatch_obs, request):
     @pytest.fixture(params=[np.complex64, np.complex128])
     @pytest.mark.parametrize("theta", np.linspace(-2 * np.pi, 2 * np.pi, 7))
     @pytest.mark.parametrize("isBatch_obs", [False, True])
-    def test_Rot_gradient(self, theta, tol, isBatch_obs, request):
+    @pytest.mark.parametrize("stateprep", [qml.QubitStateVector, qml.StatePrep])
+    def test_Rot_gradient(self, stateprep, theta, tol, isBatch_obs, request):
         """Tests that the device gradient of an arbitrary Euler-angle-parameterized gate is
         correct."""
         num_wires = 3
@@ -218,7 +220,7 @@ def test_Rot_gradient(self, theta, tol, isBatch_obs, request):
         params = np.array([theta, theta**3, np.sqrt(2) * theta])
 
         with qml.tape.QuantumTape() as tape:
-            qml.QubitStateVector(np.array([1.0, -1.0]) / np.sqrt(2), wires=0)
+            stateprep(np.array([1.0, -1.0]) / np.sqrt(2), wires=0)
             qml.Rot(*params, wires=[0])
             qml.expval(qml.PauliZ(0))
 
@@ -760,7 +762,7 @@ def circuit_2(params, wires):
 
 def circuit_ansatz(params, wires):
     """Circuit ansatz containing all the parametrized gates"""
-    qml.QubitStateVector(unitary_group.rvs(2**6, random_state=0)[0], wires=wires)
+    qml.StatePrep(unitary_group.rvs(2**6, random_state=0)[0], wires=wires)
     qml.RX(params[0], wires=wires[0])
     qml.RY(params[1], wires=wires[1])
     qml.adjoint(qml.RX(params[2], wires=wires[2]))

mpitests/test_apply.py

@@ -72,7 +72,7 @@ def apply_operation_gates_qnode_param(tol, operation, par, Wires):
     )
 
     def circuit(*params):
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(*params, wires=Wires)
         return qml.state()
 
@@ -105,7 +105,7 @@ def apply_operation_gates_apply_param(tol, operation, par, Wires):
 
     @qml.qnode(dev_cpu)
     def circuit(*params):
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(*params, wires=Wires)
         return qml.state()
 
@@ -153,7 +153,7 @@ def apply_operation_gates_qnode_nonparam(tol, operation, Wires):
     )
 
     def circuit():
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(wires=Wires)
         return qml.state()
 
@@ -186,7 +186,7 @@ def apply_operation_gates_apply_nonparam(tol, operation, Wires):
 
     @qml.qnode(dev_cpu)
     def circuit():
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(wires=Wires)
         return qml.state()
 
@@ -225,7 +225,7 @@ def expval_single_wire_no_param(tol, obs):
     dev_gpumpi = qml.device("lightning.gpu", wires=num_wires, mpi=True, c_dtype=np.complex128)
 
     def circuit():
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         return qml.expval(obs)
 
     cpu_qnode = qml.QNode(circuit, dev_cpu)
@@ -252,7 +252,7 @@ def apply_probs_nonparam(tol, operation, GateWires, Wires):
     dev_gpumpi = qml.device("lightning.gpu", wires=num_wires, mpi=True, c_dtype=np.complex128)
 
     def circuit():
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(wires=GateWires)
         return qml.probs(wires=Wires)
 
@@ -292,7 +292,7 @@ def apply_probs_param(tol, operation, par, GateWires, Wires):
     dev_gpumpi = qml.device("lightning.gpu", wires=num_wires, mpi=True, c_dtype=np.complex128)
 
     def circuit():
-        qml.QubitStateVector(state_vector, wires=range(num_wires))
+        qml.StatePrep(state_vector, wires=range(num_wires))
         operation(*par, wires=GateWires)
         return qml.probs(wires=Wires)
 
@@ -501,7 +501,7 @@ def test_qubit_state_prep(self, tol, par, Wires):
         dev_gpumpi = qml.device("lightning.gpu", wires=num_wires, mpi=True, c_dtype=np.complex128)
 
         def circuit():
-            qml.QubitStateVector(par, wires=Wires)
+            qml.StatePrep(par, wires=Wires)
             return qml.state()
 
         cpu_qnode = qml.QNode(circuit, dev_cpu)
@@ -1152,7 +1152,7 @@ def test_prob_four_wire_param(self, tol, operation, par, GateWires, Wires):
 
 def circuit_ansatz(params, wires):
     """Circuit ansatz containing all the parametrized gates"""
-    qml.QubitStateVector(unitary_group.rvs(2**numQubits, random_state=0)[0], wires=wires)
+    qml.StatePrep(unitary_group.rvs(2**numQubits, random_state=0)[0], wires=wires)
     qml.RX(params[0], wires=wires[0])
     qml.RY(params[1], wires=wires[1])
     qml.adjoint(qml.RX(params[2], wires=wires[2]))

pennylane_lightning_gpu/_serialize.py

@@ -24,7 +24,7 @@
     PauliY,
     PauliZ,
     Identity,
-    QubitStateVector,
+    StatePrep,
     Rot,
 )
 from pennylane.operation import Tensor
@@ -286,7 +286,7 @@ def _serialize_ops(
     sv_py = LightningGPU_C64 if use_csingle else LightningGPU_C128
 
     for o in tape.operations:
-        if isinstance(o, (BasisState, QubitStateVector)):
+        if isinstance(o, (BasisState, StatePrep)):
             uses_stateprep = True
             continue
         elif isinstance(o, Rot):