Add GateImplementationsLM::applyGeneratorDoubleExcitation(Minus/Plus)… (

#512) * Add failing tests. * Add revWireParityN. * Fix applyMultiQubitOp implementation. * Auto update version * Update changelog * Refactor revWireParity overloads. * WIP fix Kokkos too. * Fix bug in multiQubitOpFunctor * Refactor unitary/hermitian tests and fix L-Kokkos multiQubit kernels. * Fix L-Qubit BitUtil. * Include algorithm header in BitUtil (sort). * Fix tidy issue. Fix failing test. * Make wires2Parity inline. * Update changelog. * Update pennylane_lightning/core/src/utils/BitUtil.hpp Co-authored-by: Amintor Dusko <[email protected]> * Add GateImplementationsLM::applyGeneratorDoubleExcitation(Minus/Plus) kernels. * Update changelog. * Reduce 4-qubit gate LM kernels. * Add parity2indices functions to reduce boilerplate. * Auto update version * trigger CI --------- Co-authored-by: Dev version update bot <github-actions[bot]@users.noreply.github.com> Co-authored-by: Amintor Dusko <[email protected]>
PennyLaneAI · Oct 11, 2023 · c37d79c · c37d79c
1 parent 5cb3a67
commit c37d79c
Show file tree

Hide file tree

Showing 5 changed files with 302 additions and 151 deletions.
diff --git a/.github/CHANGELOG.md b/.github/CHANGELOG.md
@@ -10,6 +10,9 @@
 
 ### Breaking changes
 
+* Implement `LM::GeneratorDoubleExcitation`, `LM::GeneratorDoubleExcitationMinus`, `LM::GeneratorDoubleExcitationPlus` kernels. L-Qubit default kernels are now strictly from the `LM` implementation, which requires less memory and is faster for large state vectors.  
+  [(#512)](https://github.com/PennyLaneAI/pennylane-lightning/pull/512)
+
 * Add workflows validating compatibility between PennyLane and Lightning's most recent stable releases and development (latest) versions.
   [(#507)](https://github.com/PennyLaneAI/pennylane-lightning/pull/507)
   [(#498)](https://github.com/PennyLaneAI/pennylane-lightning/pull/498)

diff --git a/pennylane_lightning/core/_version.py b/pennylane_lightning/core/_version.py
@@ -16,4 +16,4 @@
    Version number (major.minor.patch[-label])
 """
 
-__version__ = "0.33.0-dev17"
+__version__ = "0.33.0-dev18"
diff --git a/pennylane_lightning/core/src/simulators/lightning_qubit/gates/AssignKernelMap_Default.cpp b/pennylane_lightning/core/src/simulators/lightning_qubit/gates/AssignKernelMap_Default.cpp
@@ -203,13 +203,13 @@ void assignKernelsForGeneratorOp_Default() {
                                all_qubit_numbers, KernelType::LM);
     instance.assignKernelForOp(GeneratorOperation::DoubleExcitation,
                                all_threading, all_memory_model,
-                               all_qubit_numbers, KernelType::PI);
+                               all_qubit_numbers, KernelType::LM);
     instance.assignKernelForOp(GeneratorOperation::DoubleExcitationPlus,
                                all_threading, all_memory_model,
-                               all_qubit_numbers, KernelType::PI);
+                               all_qubit_numbers, KernelType::LM);
     instance.assignKernelForOp(GeneratorOperation::DoubleExcitationMinus,
                                all_threading, all_memory_model,
-                               all_qubit_numbers, KernelType::PI);
+                               all_qubit_numbers, KernelType::LM);
 
     instance.assignKernelForOp(GeneratorOperation::MultiRZ, all_threading,
                                all_memory_model, all_qubit_numbers,

diff --git a/...lightning/core/src/simulators/lightning_qubit/gates/cpu_kernels/GateImplementationsLM.cpp b/...lightning/core/src/simulators/lightning_qubit/gates/cpu_kernels/GateImplementationsLM.cpp
@@ -203,6 +203,116 @@ auto GateImplementationsLM::applyGeneratorSingleExcitationPlus(
     return -static_cast<PrecisionT>(0.5);
 }
 
+template <class PrecisionT>
+auto GateImplementationsLM::applyGeneratorDoubleExcitation(
+    std::complex<PrecisionT> *arr, std::size_t num_qubits,
+    const std::vector<std::size_t> &wires, [[maybe_unused]] bool adj)
+    -> PrecisionT {
+    using ComplexT = std::complex<PrecisionT>;
+    PL_ASSERT(wires.size() == 4);
+    constexpr ComplexT zero{};
+    constexpr ComplexT imag{0, 1};
+    constexpr std::size_t one{1};
+
+    const std::array<std::size_t, 4> rev_wires{
+        num_qubits - wires[3] - 1, num_qubits - wires[2] - 1,
+        num_qubits - wires[1] - 1, num_qubits - wires[0] - 1};
+
+    const std::array<std::size_t, 4> rev_wire_shifts{
+        one << rev_wires[0], one << rev_wires[1], one << rev_wires[2],
+        one << rev_wires[3]};
+
+    const auto parity = Pennylane::Util::revWireParity(rev_wires);
+
+    for (std::size_t k = 0; k < exp2(num_qubits - 4); k++) {
+        const auto indices =
+            GateImplementationsLM::parity2indices(k, parity, rev_wire_shifts);
+        const ComplexT v3 = arr[indices[0B0011]];
+        const ComplexT v12 = arr[indices[0B1100]];
+        for (const auto &i : indices) {
+            arr[i] = zero;
+        }
+        arr[indices[0B0011]] = -v12 * imag;
+        arr[indices[0B1100]] = v3 * imag;
+    }
+    // NOLINTNEXTLINE(readability - magic - numbers)
+    return -static_cast<PrecisionT>(0.5);
+}
+
+template <class PrecisionT>
+auto GateImplementationsLM::applyGeneratorDoubleExcitationMinus(
+    std::complex<PrecisionT> *arr, std::size_t num_qubits,
+    const std::vector<std::size_t> &wires, [[maybe_unused]] bool adj)
+    -> PrecisionT {
+    using ComplexT = std::complex<PrecisionT>;
+    PL_ASSERT(wires.size() == 4);
+    constexpr ComplexT imag{0, 1};
+    constexpr std::size_t one{1};
+
+    const std::array<std::size_t, 4> rev_wires{
+        num_qubits - wires[3] - 1, num_qubits - wires[2] - 1,
+        num_qubits - wires[1] - 1, num_qubits - wires[0] - 1};
+
+    const std::array<std::size_t, 4> rev_wire_shifts{
+        one << rev_wires[0], one << rev_wires[1], one << rev_wires[2],
+        one << rev_wires[3]};
+
+    const auto parity = Pennylane::Util::revWireParity(rev_wires);
+
+    for (std::size_t k = 0; k < exp2(num_qubits - 4); k++) {
+        const std::size_t i0000 =
+            ((k << 4U) & parity[4]) | ((k << 3U) & parity[3]) |
+            ((k << 2U) & parity[2]) | ((k << 1U) & parity[1]) | (k & parity[0]);
+        const std::size_t i0011 =
+            i0000 | rev_wire_shifts[1] | rev_wire_shifts[0];
+        const std::size_t i1100 =
+            i0000 | rev_wire_shifts[3] | rev_wire_shifts[2];
+
+        arr[i0011] *= imag;
+        arr[i1100] *= -imag;
+        swap(arr[i1100], arr[i0011]);
+    }
+    // NOLINTNEXTLINE(readability - magic - numbers)
+    return -static_cast<PrecisionT>(0.5);
+}
+
+template <class PrecisionT>
+auto GateImplementationsLM::applyGeneratorDoubleExcitationPlus(
+    std::complex<PrecisionT> *arr, std::size_t num_qubits,
+    const std::vector<std::size_t> &wires, [[maybe_unused]] bool adj)
+    -> PrecisionT {
+    using ComplexT = std::complex<PrecisionT>;
+    PL_ASSERT(wires.size() == 4);
+    constexpr ComplexT imag{0, 1};
+    constexpr std::size_t one{1};
+
+    const std::array<std::size_t, 4> rev_wires{
+        num_qubits - wires[3] - 1, num_qubits - wires[2] - 1,
+        num_qubits - wires[1] - 1, num_qubits - wires[0] - 1};
+
+    const std::array<std::size_t, 4> rev_wire_shifts{
+        one << rev_wires[0], one << rev_wires[1], one << rev_wires[2],
+        one << rev_wires[3]};
+
+    const auto parity = Pennylane::Util::revWireParity(rev_wires);
+
+    for (std::size_t k = 0; k < exp2(num_qubits - 4); k++) {
+        const std::size_t i0000 =
+            ((k << 4U) & parity[4]) | ((k << 3U) & parity[3]) |
+            ((k << 2U) & parity[2]) | ((k << 1U) & parity[1]) | (k & parity[0]);
+        const std::size_t i0011 =
+            i0000 | rev_wire_shifts[1] | rev_wire_shifts[0];
+        const std::size_t i1100 =
+            i0000 | rev_wire_shifts[3] | rev_wire_shifts[2];
+
+        arr[i0011] *= -imag;
+        arr[i1100] *= imag;
+        swap(arr[i1100], arr[i0011]);
+    }
+    // NOLINTNEXTLINE(readability - magic - numbers)
+    return static_cast<PrecisionT>(0.5);
+}
+
 // Explicit instantiation starts
 /* Matrix operations */
 template void GateImplementationsLM::applySingleQubitOp<float>(
@@ -571,6 +681,31 @@ template auto GateImplementationsLM::applyGeneratorSingleExcitationPlus<double>(
     std::complex<double> *arr, size_t num_qubits,
     const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> double;
 
+template auto GateImplementationsLM::applyGeneratorDoubleExcitation<float>(
+    std::complex<float> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> float;
+
+template auto GateImplementationsLM::applyGeneratorDoubleExcitation<double>(
+    std::complex<double> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> double;
+
+template auto GateImplementationsLM::applyGeneratorDoubleExcitationMinus<float>(
+    std::complex<float> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> float;
+
+template auto
+GateImplementationsLM::applyGeneratorDoubleExcitationMinus<double>(
+    std::complex<double> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> double;
+
+template auto GateImplementationsLM::applyGeneratorDoubleExcitationPlus<float>(
+    std::complex<float> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> float;
+
+template auto GateImplementationsLM::applyGeneratorDoubleExcitationPlus<double>(
+    std::complex<double> *arr, size_t num_qubits,
+    const std::vector<size_t> &wires, [[maybe_unused]] bool adj) -> double;
+
 // Explicit instantiations ends
 
 } // namespace Pennylane::LightningQubit::Gates