Implement Rotoselect Quantum Circuit Structure Learning Algorithm

Ref: https://arxiv.org/abs/1905.09692


Signed-off-by: Thien Nguyen <nguyentm@ornl.gov>

quantum/examples/qasm/CMakeLists.txt

@@ -24,4 +24,7 @@ target_link_libraries(ddcl_example PRIVATE xacc)
 
 
 add_executable(bell_noisy bell_readout_py_load.cpp)
-target_link_libraries(bell_noisy PRIVATE xacc)
\ No newline at end of file
+target_link_libraries(bell_noisy PRIVATE xacc)
+
+add_executable(rotoselect rotoselect_quantum_circuit_structure_learning.cpp)
+target_link_libraries(rotoselect PRIVATE xacc xacc-quantum-gate)

quantum/examples/qasm/rotoselect_quantum_circuit_structure_learning.cpp

+/*******************************************************************************
+ * Copyright (c) 2020 UT-Battelle, LLC.
+ * All rights reserved. This program and the accompanying materials
+ * are made available under the terms of the Eclipse Public License v1.0
+ * and Eclipse Distribution License v1.0 which accompanies this
+ * distribution. The Eclipse Public License is available at
+ * http://www.eclipse.org/legal/epl-v10.html and the Eclipse Distribution
+ *License is available at https://eclipse.org/org/documents/edl-v10.php
+ *
+ * Contributors:
+ *   Thien Nguyen - initial API and implementation
+ *******************************************************************************/
+#include "xacc.hpp"
+#include "xacc_observable.hpp"
+#include "xacc_service.hpp"
+
+// This example demonstrates XACC implementation of the Rotoselect quantum circuit structure learning.
+// Ref: https://arxiv.org/abs/1905.09692
+// The algorithm will *train* the circuit to prepare a quantum state that minimizes an objective function (aka Hamiltonian observable).
+// Inputs to the Rotoselect algorithm are:
+//  - accelerator: reference to a quantum processor accelerator (e.g. a simulator) which will evaluate circuit to get the energy.
+//  - observable: the objective function to minimize in the form of a XACC Observable object
+//  - layers: depth of the training circuit; each layer contains a rotation gate (Rx, Ry, Rz) on each qubit and CZ gates between neighboring qubits.
+//  - iterations: the number of iterations that we want to train the circuit. Each rotation gate update is considered as an iteration.
+// Note: the number of qubits is inferred automatically from the input acceleration buffer.
+// The optimal energy achieved after the training is embedded in the input acceleration buffer as "opt-val".
+int main(int argc, char **argv) {
+  xacc::Initialize(argc, argv);
+  // Use the Qpp simulator as the accelerator
+  auto acc = xacc::getAccelerator("qpp");
+  
+  // In this example, we try to find the ground state energy of the 5-qubit Heisenberg model on a 
+  // 1D lattice with periodic boundary conditions.
+  auto buffer = xacc::qalloc(5);
+  // The corresponding Hamiltonian is:
+  // Note: we set all scalar params to 1.0 for simplicity.
+  auto observable = xacc::quantum::getObservable(
+      "pauli",
+      std::string("X0 X1 + X1 X2 + X2 X3 + X3 X4 + X4 X0 + "
+                  "Y0 Y1 + Y1 Y2 + Y2 Y3 + Y3 Y4 + Y4 Y0 +"
+                  "Z0 Z1 + Z1 Z2 + Z2 Z3 + Z3 Z4 + Z4 Z0 + "
+                  "Z0 + Z1 + Z2 + Z3 + Z4"));
+
+  auto rotoselect = xacc::getService<xacc::Algorithm>("rotoselect");
+  // We train the circuit that has 6 layers for 1000 iterations.
+  // Note: we can repeat this training multiple times to collect the mean and standard deviation
+  // as reported in https://arxiv.org/abs/1905.09692
+  rotoselect->initialize({
+                          std::make_pair("accelerator", acc),
+                          std::make_pair("observable", observable),
+                          std::make_pair("layers", 6),
+                          std::make_pair("iterations", 1000),
+                        });
+  
+  rotoselect->execute(buffer);
+  // Expected value ~ -8.4 -> -8.5
+  // Ref: FIG. 2 of https://arxiv.org/abs/1905.09692
+  std::cout << "Energy: " << (*buffer)["opt-val"].as<double>() << "\n";
+}
\ No newline at end of file

quantum/plugins/algorithms/CMakeLists.txt

@@ -13,6 +13,8 @@
 add_subdirectory(vqe)
 add_subdirectory(rdm)
 add_subdirectory(ml)
+add_subdirectory(rotoselect)
+
 file(GLOB PYDECORATORS ${CMAKE_CURRENT_SOURCE_DIR}/vqe/python/*.py
                        ${CMAKE_CURRENT_SOURCE_DIR}/ml/ddcl/python/*.py)
 install(FILES ${PYDECORATORS} DESTINATION ${CMAKE_INSTALL_PREFIX}/py-plugins)

quantum/plugins/algorithms/rotoselect/CMakeLists.txt

+# *******************************************************************************
+# Copyright (c) 2020 UT-Battelle, LLC.
+# All rights reserved. This program and the accompanying materials
+# are made available under the terms of the Eclipse Public License v1.0
+# and Eclipse Distribution License v.10 which accompany this distribution.
+# The Eclipse Public License is available at http://www.eclipse.org/legal/epl-v10.html
+# and the Eclipse Distribution License is available at
+# https://eclipse.org/org/documents/edl-v10.php
+#
+# Contributors:
+#   Thien Nguyen - initial API and implementation
+# *******************************************************************************/
+set(LIBRARY_NAME xacc-algorithm-rotoselect)
+
+file(GLOB SRC *.cpp)
+
+usfunctiongetresourcesource(TARGET ${LIBRARY_NAME} OUT SRC)
+usfunctiongeneratebundleinit(TARGET ${LIBRARY_NAME} OUT SRC)
+
+add_library(${LIBRARY_NAME} SHARED ${SRC})
+
+target_include_directories(
+  ${LIBRARY_NAME}
+  PUBLIC .)
+
+target_link_libraries(${LIBRARY_NAME} PUBLIC xacc CppMicroServices)
+
+set(_bundle_name xacc_algorithm_rotoselect)
+set_target_properties(${LIBRARY_NAME}
+                      PROPERTIES COMPILE_DEFINITIONS
+                                 US_BUNDLE_NAME=${_bundle_name}
+                                 US_BUNDLE_NAME
+                                 ${_bundle_name})
+
+usfunctionembedresources(TARGET
+                         ${LIBRARY_NAME}
+                         WORKING_DIRECTORY
+                         ${CMAKE_CURRENT_SOURCE_DIR}
+                         FILES
+                         manifest.json)
+
+if(APPLE)
+  set_target_properties(${LIBRARY_NAME}
+                        PROPERTIES INSTALL_RPATH "@loader_path/../lib")
+  set_target_properties(${LIBRARY_NAME}
+                        PROPERTIES LINK_FLAGS "-undefined dynamic_lookup")
+else()
+  set_target_properties(${LIBRARY_NAME}
+                        PROPERTIES INSTALL_RPATH "$ORIGIN/../lib")
+  set_target_properties(${LIBRARY_NAME} PROPERTIES LINK_FLAGS "-shared")
+endif()
+
+if(XACC_BUILD_TESTS)
+  add_subdirectory(tests)
+endif()
+
+install(TARGETS ${LIBRARY_NAME} DESTINATION ${CMAKE_INSTALL_PREFIX}/plugins)

quantum/plugins/algorithms/rotoselect/README.md

+This is a XACC algorithm plugin which implements the `Rotoselect` algorithm.
+
+Ref: Quantum circuit structure learning
+https://arxiv.org/abs/1905.09692
\ No newline at end of file

quantum/plugins/algorithms/rotoselect/RotoselectActivator.cpp

+/*******************************************************************************
+ * Copyright (c) 2020 UT-Battelle, LLC.
+ * All rights reserved. This program and the accompanying materials
+ * are made available under the terms of the Eclipse Public License v1.0
+ * and Eclipse Distribution License v1.0 which accompanies this
+ * distribution. The Eclipse Public License is available at
+ * http://www.eclipse.org/legal/epl-v10.html and the Eclipse Distribution
+ *License is available at https://eclipse.org/org/documents/edl-v10.php
+ *
+ * Contributors:
+ *   Thien Nguyen - initial API and implementation
+ *******************************************************************************/
+#include "rotoselect.hpp"
+
+#include "cppmicroservices/BundleActivator.h"
+#include "cppmicroservices/BundleContext.h"
+#include "cppmicroservices/ServiceProperties.h"
+
+using namespace cppmicroservices;
+
+namespace {
+
+class US_ABI_LOCAL RotoselectActivator : public BundleActivator {
+
+public:
+  RotoselectActivator() {}
+
+  void Start(BundleContext context) {
+    auto c = std::make_shared<xacc::algorithm::Rotoselect>();
+    context.RegisterService<xacc::Algorithm>(c);
+  }
+
+  void Stop(BundleContext /*context*/) {}
+};
+
+} // namespace
+
+CPPMICROSERVICES_EXPORT_BUNDLE_ACTIVATOR(RotoselectActivator)

quantum/plugins/algorithms/rotoselect/manifest.json

+{
+  "bundle.symbolic_name" : "xacc_algorithm_rotoselect",
+  "bundle.activator" : true,
+  "bundle.name" : "XACC Rotoselect Algorithm",
+  "bundle.description" : "Quantum Circuit Structure Learning based on Rotoselect algorithm"
+}

quantum/plugins/algorithms/rotoselect/rotoselect.cpp

+/*******************************************************************************
+ * Copyright (c) 2020 UT-Battelle, LLC.
+ * All rights reserved. This program and the accompanying materials
+ * are made available under the terms of the Eclipse Public License v1.0
+ * and Eclipse Distribution License v1.0 which accompanies this
+ * distribution. The Eclipse Public License is available at
+ * http://www.eclipse.org/legal/epl-v10.html and the Eclipse Distribution
+ *License is available at https://eclipse.org/org/documents/edl-v10.php
+ *
+ * Contributors:
+ *   Thien Nguyen - initial API and implementation
+ *******************************************************************************/
+
+#include "rotoselect.hpp"
+#include "xacc.hpp"
+#include "xacc_service.hpp"
+#include <random>
+#include <algorithm>
+#include <iterator>
+#include <iostream>
+#include <functional>
+#include <cassert>
+#include <cmath>
+
+using namespace xacc;
+
+namespace {
+  enum class PauliType { X = 0, Y = 1, Z = 2 };
+
+  std::vector<double> initializeThetaVec(int in_nbRotationGates)
+  {
+    std::random_device rndDevice;
+    std::mt19937 rndEngine {rndDevice()}; 
+    // Theta angle is in (-pi, pi] range 
+    std::uniform_real_distribution<double> dist {-M_PI, M_PI};
+
+    auto gen = [&dist, &rndEngine]() {
+      return dist(rndEngine);
+    };
+
+    std::vector<double> vec(in_nbRotationGates);
+    std::generate(vec.begin(), vec.end(), gen);
+    return vec;
+  }
+
+  std::vector<PauliType> initializePauliTypeVec(int in_nbRotationGates)
+  {
+    std::random_device rndDevice;
+    std::mt19937 rndEngine {rndDevice()}; 
+    // Pauli type is X, Y, Z (0->2) 
+    std::uniform_int_distribution<int> dist {0, 2};
+
+    auto gen = [&dist, &rndEngine]() -> PauliType {
+      const auto randVal = dist(rndEngine);
+      return static_cast<PauliType>(randVal);
+    };
+
+    std::vector<PauliType> vec(in_nbRotationGates);
+    std::generate(vec.begin(), vec.end(), gen);
+    return vec;
+  }
+}
+
+
+namespace xacc {
+namespace algorithm {
+
+bool Rotoselect::initialize(const HeterogeneousMap& in_parameters) 
+{
+  bool validationResult = true;
+  if (!in_parameters.keyExists<std::shared_ptr<Observable>>("observable")) 
+  {
+    std::cout << "Observable is required. Please provide 'observable' parameter.\n";
+    validationResult = false;
+  }
+  
+  if (!in_parameters.keyExists<std::shared_ptr<Accelerator>>("accelerator")) 
+  {
+    std::cout << "Accelerator is required. Please provide 'accelerator' parameter.\n";
+    validationResult = false;
+  }
+
+
+  if (!in_parameters.keyExists<int>("iterations")) 
+  {
+    std::cout << "Please specify the number of iterations in the 'iterations' parameter.\n";
+    validationResult = false;
+  }
+  else
+  {
+    const int nbIterations = in_parameters.get<int>("iterations");
+    if (nbIterations < 1)
+    {
+      std::cout << "Invalid 'iterations' parameter. This must be a positive number.\n";
+      validationResult = false;
+    }
+  }  
+
+  if (!in_parameters.keyExists<int>("layers")) 
+  {
+    std::cout << "Please specify the number of circuit layers in the 'layers' parameter.\n";
+    validationResult = false;
+  }
+  else
+  {
+    const int nbLayers = in_parameters.get<int>("layers");
+    if (nbLayers < 1)
+    {
+      std::cout << "Invalid 'layers' parameter. This must be a positive number.\n";
+      validationResult = false;
+    }
+  }  
+
+  if (!validationResult)
+  {
+    xacc::error("Input parameter validation failed.\n");
+    return false;
+  }
+
+  m_parameters = in_parameters;
+  return true;
+}
+
+const std::vector<std::string> Rotoselect::requiredParameters() const 
+{
+  return { "observable", "accelerator", "layers", "iterations" };
+}
+
+void Rotoselect::execute(const std::shared_ptr<AcceleratorBuffer> buffer) const 
+{
+  auto gateRegistry = xacc::getService<IRProvider>("quantum");
+  auto accelerator = m_parameters.get<std::shared_ptr<Accelerator>>("accelerator");
+
+  const auto nbQubits = buffer->size();
+  const auto nbLayers = m_parameters.get<int>("layers");
+  // D params
+  const auto nbRotationGates = nbQubits * nbLayers;
+  // Tracking of all rotation gates (angle and type) 
+  std::vector<double> thetaVec = initializeThetaVec(nbRotationGates);
+  std::vector<PauliType> pauliTypeVec = initializePauliTypeVec(nbRotationGates);
+  std::unordered_map<size_t, size_t> pauliGateIdxToCircuitIdx;
+  
+  auto rotoselectKernel = gateRegistry->createComposite("rotoselectKernel");
+  
+  const auto createPauliGate = [&gateRegistry](size_t in_qubitIdx, PauliType in_pauliType, double in_angle) {
+    switch (in_pauliType)
+    {
+      case PauliType::X:
+      {
+        return gateRegistry->createInstruction("Rx", { in_qubitIdx}, { in_angle });
+      }         
+      case PauliType::Y:
+      {
+        return gateRegistry->createInstruction("Ry", { in_qubitIdx }, { in_angle });
+      }
+      case PauliType::Z:
+      {
+        return gateRegistry->createInstruction("Rz", { in_qubitIdx }, { in_angle });
+      }         
+    };
+  };  
+  
+  // Step 1: initialize the quantum circuit
+  for (size_t i = 0; i < nbLayers; ++i)
+  {
+    // Pauli rotation layer:
+    for (size_t j = 0; j < nbQubits; ++j)
+    {
+      const auto pauliType = pauliTypeVec[i*nbQubits + j];
+      const auto angle = thetaVec[i*nbQubits + j];
+      rotoselectKernel->addInstruction(createPauliGate(j, pauliType, angle));
+      const auto circuitGateIdx = rotoselectKernel->nInstructions() - 1;
+      // Track the mapping between Pauli rotation gate and absolute index within the circuit
+      // Note: we have layers of CZ gates in between rotation layers.
+      pauliGateIdxToCircuitIdx.emplace(i*nbQubits + j, circuitGateIdx);
+    }
+
+    // CZ layer
+    for (size_t j = 0; j < nbQubits; j += 2)
+    {
+      if (j + 1 < nbQubits)
+      {
+        auto cz = gateRegistry->createInstruction("CZ", { j, j + 1 });
+        rotoselectKernel->addInstruction(cz);
+      }
+    }
+  }
+  
+  if (xacc::verbose)
+  {
+    // Debug:
+    std::cout << "Circuit: \n" <<  rotoselectKernel->toString();
+  }
+  
+  const auto observable = m_parameters.get_with_throw<std::shared_ptr<Observable>>("observable");
+  // Step 2: Optimization loop
+  bool stopCriteriaMet = false;
+  // Currently, we only use trial count as the stopping criteria.
+  // We can implement other criteria, such as convergence rate, etc.
+  const int nbIterations = m_parameters.get<int>("iterations");
+  int iterationCount = 0;
+  
+  // Optimization note:
+  // This evaluation function is the main runner of the training loop (7 evaluations per training step/iteration).
+  // We can run all evaluations per step in parallel. In order to do so, we need:
+  // (1) Clone the rotoselectKernel for each thread, i.e. each thread has an independent copy (because it will swap a gate for evaluation).
+  // (2) Clone the accelerator (simulator) which will run the circuit to evaluate the energy for this config.
+  // This will give us good speedup in the case that the simulator is not utilize all the CPU resources (i.e. if the simulator is not multithreaded by itself).
+  const auto evaluateFn = [&](size_t in_pauliGateIdx, PauliType in_pauliType, double in_angle) -> double {
+    // Find the gate:
+    auto iter = pauliGateIdxToCircuitIdx.find(in_pauliGateIdx);
+    assert(iter != pauliGateIdxToCircuitIdx.end());
+
+    const auto currentGate = rotoselectKernel->getInstruction(iter->second);
+    assert(currentGate->nRequiredBits() == 1);
+    const auto qubitIdx = currentGate->bits()[0];
+    auto newGate = createPauliGate(qubitIdx, in_pauliType, in_angle);
+    rotoselectKernel->replaceInstruction(iter->second, newGate);
+    // Calculate the observable
+    auto kernels = observable->observe(rotoselectKernel);
+    double energy = 0.0;
+
+    {
+      std::vector<double> coefficients;
+      std::vector<std::string> kernelNames;
+      std::vector<std::shared_ptr<CompositeInstruction>> fsToExec;
+
+      double identityCoeff = 0.0;
+      for (auto &f : kernels) 
+      {
+        kernelNames.push_back(f->name());
+        std::complex<double> coeff = f->getCoefficient();
+        int nFunctionInstructions = 0;
+        if (f->getInstruction(0)->isComposite()) 
+        {
+          nFunctionInstructions = rotoselectKernel->nInstructions() + f->nInstructions() - 1;
+        } 
+        else 
+        {
+          nFunctionInstructions = f->nInstructions();
+        }
+
+        if (nFunctionInstructions > rotoselectKernel->nInstructions()) 
+        {
+          fsToExec.push_back(f);
+          coefficients.push_back(std::real(coeff));
+        } 
+        else 
+        {
+          identityCoeff += std::real(coeff);
+        }
+      }
+
+      auto tmpBuffer = xacc::qalloc(buffer->size());
+      accelerator->execute(tmpBuffer, fsToExec);
+      auto buffers = tmpBuffer->getChildren();
+
+      energy = identityCoeff;
+      auto idBuffer = xacc::qalloc(buffer->size());
+      idBuffer->addExtraInfo("coefficient", identityCoeff);
+      idBuffer->setName("I");
+      idBuffer->addExtraInfo("kernel", "I");
+      idBuffer->addExtraInfo("exp-val-z", 1.0);
+      buffer->appendChild("I", idBuffer);
+
+      for (int i = 0; i < buffers.size(); i++) 
+      {
+        auto expval = buffers[i]->getExpectationValueZ();
+        energy += expval * coefficients[i];
+        buffers[i]->addExtraInfo("coefficient", coefficients[i]);
+        buffers[i]->addExtraInfo("kernel", fsToExec[i]->name());
+        buffers[i]->addExtraInfo("exp-val-z", expval);
+        buffer->appendChild(fsToExec[i]->name(), buffers[i]);
+      } 
+
+    }
+
+    // Return back the original gate:
+    rotoselectKernel->replaceInstruction(iter->second, currentGate);
+    return energy;
+  };
+  
+  // We keep track of all the min energy value achieved at each iteration.
+  // This can be used to determine stopping conditions (in addition to just iteration loop count)
+  std::vector<double> minEnergyVec;
+  minEnergyVec.reserve(nbIterations);
+  // The training result (i.e. the min energy that is encountered so far)
+  double trainingResult = std::numeric_limits<double>::max();
+  while (!stopCriteriaMet)
+  {
+    for (size_t i = 0; i < nbLayers; ++i)
+    {
+      for (size_t j = 0; j < nbQubits; ++j)
+      {
+        const auto pauliGateIdx =  i*nbQubits + j;
+        // We need 7 params to evaluate optimal params for this rotation gate
+        const double M_X_PI_2 = evaluateFn(pauliGateIdx, PauliType::X, M_PI_2);
+        const double M_X_M_PI_2 = evaluateFn(pauliGateIdx, PauliType::X, -M_PI_2);
+        const double M_Y_PI_2 = evaluateFn(pauliGateIdx, PauliType::Y, M_PI_2);
+        const double M_Y_M_PI_2 = evaluateFn(pauliGateIdx, PauliType::Y, -M_PI_2);
+        const double M_Z_PI_2 = evaluateFn(pauliGateIdx, PauliType::Z, M_PI_2);
+        const double M_Z_M_PI_2 = evaluateFn(pauliGateIdx, PauliType::Z, -M_PI_2);
+        const double M_ZERO = evaluateFn(pauliGateIdx, PauliType::X, 0.0);
+
+        // Compute the minimum for each Pauli type
+        const auto minCalc = [](double in_m0, double in_mPiOver2, double in_mMinusPiOver2){
+          double result = -M_PI_2 - atan2(2*in_m0 - in_mPiOver2 - in_mMinusPiOver2, in_mPiOver2 - in_mMinusPiOver2);
+          // Normalize to the [-pi,+pi] range
+          if (result < -M_PI)
+          {
+            result += 2*M_PI;
+          }  
+          assert((result >= -M_PI && result <= M_PI));
+          return result;
+        };
+
+        const double min_Theta_X = minCalc(M_ZERO, M_X_PI_2, M_X_M_PI_2);
+        const double min_Theta_Y = minCalc(M_ZERO, M_Y_PI_2, M_Y_M_PI_2);
+        const double min_Theta_Z = minCalc(M_ZERO, M_Z_PI_2, M_Z_M_PI_2);
+        
+        // Which Pauli give us the minimum?
+        const double min_X = evaluateFn(pauliGateIdx, PauliType::X, min_Theta_X);
+        const double min_Y = evaluateFn(pauliGateIdx, PauliType::Y, min_Theta_Y);
+        const double min_Z = evaluateFn(pauliGateIdx, PauliType::Z, min_Theta_Z);
+        double minEnergy = min_X;
+        PauliType minPauli = PauliType::X;
+        double minAngle = min_Theta_X;
+        if (min_Y < minEnergy)
+        {
+          minEnergy = min_Y;
+          minPauli = PauliType::Y;
+          minAngle = min_Theta_Y;
+        }
+
+        if (min_Z < minEnergy)
+        {
+          minEnergy = min_Z;
+          minPauli = PauliType::Z;
+          minAngle = min_Theta_Z;
+        }
+
+        // Update the rotation gate:
+        ++iterationCount;
+        auto iter = pauliGateIdxToCircuitIdx.find(pauliGateIdx);
+        assert(iter != pauliGateIdxToCircuitIdx.end());
+        const auto currentGate = rotoselectKernel->getInstruction(iter->second);
+        assert(currentGate->nRequiredBits() == 1);
+        const auto qubitIdx = currentGate->bits()[0];
+        auto newGate = createPauliGate(qubitIdx, minPauli, minAngle);
+        rotoselectKernel->replaceInstruction(iter->second, newGate);
+        
+        if (xacc::verbose)
+        {
+          // Debug:
+          std::cout << "Replace " << currentGate->name() << "(" << currentGate->getParameter(0).as<double>() << ")" << " with " << newGate->name() << "(" << minAngle << ")" << " @ q[" << qubitIdx << "] \n";
+          std::cout << "Min Energy: " << minEnergy << "\n";
+          std::cout << "New circuit: \n" <<  rotoselectKernel->toString();
+        }
+        
+        minEnergyVec.emplace_back(minEnergy);
+        if (minEnergy < trainingResult)
+        {
+          // Update the new result
+          trainingResult = minEnergy;
+        }
+        // Training logging:
+        std::cout << "Rotoselect (iteration = " << iterationCount << "): Objective function value = " << trainingResult << "\n";
+      }
+    }
+
+    // For now, just use loop count as the stopping condition
+    // We can implement more intelligent stopping condition based on data in minEnergyVec 
+    // (i.e. the history of all min energy values attained in all previous iterations)
+    stopCriteriaMet = iterationCount >= nbIterations;
+  }  
+
+  // Done, add info to the buffer
+  buffer->addExtraInfo("opt-val", trainingResult);
+}
+
+} // namespace algorithm
+} // namespace xacc
\ No newline at end of file

quantum/plugins/algorithms/rotoselect/rotoselect.hpp

+/*******************************************************************************
+ * Copyright (c) 2020 UT-Battelle, LLC.
+ * All rights reserved. This program and the accompanying materials
+ * are made available under the terms of the Eclipse Public License v1.0
+ * and Eclipse Distribution License v1.0 which accompanies this
+ * distribution. The Eclipse Public License is available at
+ * http://www.eclipse.org/legal/epl-v10.html and the Eclipse Distribution
+ *License is available at https://eclipse.org/org/documents/edl-v10.php
+ *
+ * Contributors:
+ *   Thien Nguyen - initial API and implementation
+ *******************************************************************************/
+#pragma once
+
+#include "Algorithm.hpp"
+
+namespace xacc {