Commit 6e304b6b authored by Nguyen, Thien Minh's avatar Nguyen, Thien Minh
Browse files

Merge branch 'master' into tnguyen/ftqc-shots

parents cc350142 4b40a33c

examples/hybrid/deuteron_h2_qaoa.cpp

+0 −11
Original line number Diff line number Diff line
@@ -32,15 +32,4 @@ int main() { 


  std::cout << "<H> = " << energy << "\n";



  // note you could also call execute_async() -> Handle and

  // manually qcor::sync(handle)



  // or you could call execute(initial_params:vector<double>)

  // or you could call execute_async(initial_params:vector<double>)

  // or you could call execute(optimizer, initial_params:vector<double>)

  // or you could call execute(optimizer)



  // If you wanted to provide initial-parameters, you can

  // query the size of vector you need with

  // qaoa.n_parameters() (also have n_gamma(), n_beta(), and n_steps())

}
 
 No newline at end of file

examples/hybrid/deuteron_h2_vqe.cpp

+1 −8
Original line number Diff line number Diff line
@@ -60,14 +60,7 @@ int main() { 
  // Now do the same for the vector double ansatz, but

  // also demonstrate the async interface

  VQE vqe_vec(ansatz_vec, H);

  auto handle = vqe_vec.execute_async(std::vector<double>{0.0});



  // Can go do other work, quantum execution is happening on

  // separate thread



  // Get the energy, this call will kick off a wait if needed

  const auto [energy_vec, params_vec] = vqe_vec.sync(handle);



  const auto [energy_vec, params_vec] = vqe_vec.execute(std::vector<double>{0.0});

  std::cout << "<H>(" << params_vec[0] << ") = " << energy_vec << "\n";



  // Now run with the mixed language kernel,

examples/hybrid/parameter_sweep.cpp

0 → 100644
+26 −0
Original line number Diff line number Diff line 
#include "qcor_hybrid.hpp"



// Define one quantum kernel that takes

// double angle parameter

__qpu__ void ansatz(qreg q, double x) {

  X(q[0]);

  Ry(q[1], x);

  CX(q[1], q[0]);

}



int main() {



  // Define the Hamiltonian using the QCOR API

  auto H = 5.907 - 2.1433 * X(0) * X(1) - 2.1433 * Y(0) * Y(1) + .21829 * Z(0) -

           6.125 * Z(1);



  // Create a VQE instance, must give it

  // the parameterized ansatz functor and Observable

  VQE vqe(ansatz, H);

  for (auto [iter, x] :

       enumerate(linspace(-constants::pi, constants::pi, 20))) {

    std::cout << iter << ", " << x << ", " << vqe({x}) << "\n";

  }



  vqe.persist_data("param_sweep_data.json");

}
 
 No newline at end of file

lib/hybrid/qcor_hybrid.hpp

+33 −39
Original line number Diff line number Diff line
@@ -112,29 +112,13 @@ public: 
  // Execute the VQE task synchronously, assumes default optimizer

  VQEResultType execute(KernelArgs... initial_params) {

    auto optimizer = qcor::createOptimizer(DEFAULT_OPTIMIZER);

    auto handle = execute_async(optimizer, initial_params...);

    return this->sync(handle);

  }



  // Execute the VQE task asynchronously, default optimizer

  Handle execute_async(KernelArgs... initial_params) {

    auto optimizer = qcor::createOptimizer(DEFAULT_OPTIMIZER);

    return execute_async(optimizer, initial_params...);

    return execute(optimizer, initial_params...);

  }



  // Execute the VQE task synchronously, use provided Optimizer

  //   template <typename... Args>

  VQEResultType execute(std::shared_ptr<Optimizer> optimizer,

                        KernelArgs... initial_params) {

    auto handle = execute_async(optimizer, initial_params...);

    return this->sync(handle);

  }



  // Execute the VQE task asynchronously, use provided Optimizer

  //   template <typename... Args>

  Handle execute_async(std::shared_ptr<Optimizer> optimizer,

                       KernelArgs... initial_params) {



    auto ansatz_casted =

        reinterpret_cast<void (*)(std::shared_ptr<CompositeInstruction>, qreg,

                                  KernelArgs...)>(ansatz_ptr);
@@ -177,15 +161,34 @@ public: 
    }

    options.insert("observable", __internal__::qcor_as_shared(&observable));

    objective->set_options(options);

    return optimizer->optimize(*objective.get());

  }



    // Run TaskInitiate, kick of the VQE job asynchronously

    return qcor::taskInitiate(objective, optimizer);

  double operator()(const std::vector<double> x) {

    auto ansatz_casted =

        reinterpret_cast<void (*)(std::shared_ptr<CompositeInstruction>, qreg,

                                  KernelArgs...)>(ansatz_ptr);



    // Create the Arg Translator

    TranslationFunctor<qreg, KernelArgs...> arg_translator;

    if (translation_functor.has_value()) {

      arg_translator = std::any_cast<TranslationFunctor<qreg, KernelArgs...>>(

          translation_functor);

    } else {

      __internal__::TranslationFunctorAutoGenerator auto_gen;

      arg_translator = auto_gen(q, std::tuple<KernelArgs...>());

    }



  // Sync up the results with the host thread

  VQEResultType sync(Handle &h) {

    auto results = qcor::sync(h);

    return std::make_pair(results.opt_val, results.opt_params);

    // Get the VQE ObjectiveFunction and set the qreg

    auto objective = qcor::createObjectiveFunction(

        ansatz_casted, observable,

        std::make_shared<ArgsTranslator<qreg, KernelArgs...>>(arg_translator),

        q, x.size());



    options.insert("observable", __internal__::qcor_as_shared(&observable));

    objective->set_options(options);



    return (*objective)(x);

  }



  // Return all unique parameter sets this VQE run used
@@ -210,6 +213,11 @@ public: 
    }

    return ret;

  }



  void persist_data(const std::string& filename) {

      q.write_file(filename);

  }



};



namespace __internal__ {
@@ -371,15 +379,6 @@ public: 
  // initial parameters Will fail if initial_parameters.size() != n_parameters()

  QAOAResultType execute(std::shared_ptr<Optimizer> optimizer,

                       const std::vector<double> initial_parameters = {}) {

    auto handle = execute_async(optimizer, initial_parameters);

    auto results = this->sync(handle);

    return std::make_pair(results.first, results.second);

  }



  // Execute the algorithm asynchronously, provding an Optimizer and optionally

  // initial parameters Will fail if initial_parameters.size() != n_parameters()

  Handle execute_async(std::shared_ptr<Optimizer> optimizer,

                       const std::vector<double> initial_parameters = {}) {



    q = qalloc(cost.nBits());

    auto n_params = nSteps * (n_gamma() + n_beta());
@@ -422,14 +421,9 @@ public: 
    objective->set_options(options);

    optimizer->appendOption("initial-parameters", init_params);



    return qcor::taskInitiate(objective, optimizer);

    return optimizer->optimize(*objective.get());// qcor::taskInitiate(objective, optimizer);

  }



  // Sync up the results with the host thread

  QAOAResultType sync(Handle &h) {

    auto results = qcor::sync(h);

    return std::make_pair(results.opt_val, results.opt_params);

  }

}; // namespace qcor



void execute_adapt(qreg q, const HeterogeneousMap &&m);

runtime/objectives/vqe/vqe.cpp

+44 −0
Original line number Diff line number Diff line
@@ -13,7 +13,20 @@ using namespace cppmicroservices; 
#include "xacc_internal_compiler.hpp"

#include "xacc_service.hpp"



namespace {



template <typename T>

std::ostream &operator<<(std::ostream &os, const std::vector<T> &v) {

  os << "[";

  for (int i = 0; i < v.size(); ++i) {

    os << v[i];

    if (i != v.size() - 1)

      os << ",";

  }

  os << "]";

  return os;

}

} // namespace

namespace qcor {



class VQEObjective : public ObjectiveFunction {
@@ -35,13 +48,43 @@ public: 


    auto tmp_child = qalloc(qreg.size());

    auto val = vqe->execute(xacc::as_shared_ptr(tmp_child.results()), {})[0];

    double std_dev = 0.0;

    if (options.keyExists<int>("vqe-gather-statistics")) {

      std::vector<double> all_energies;

      all_energies.push_back(val);

      auto n = options.get<int>("vqe-gather-statistics");

      for (int i = 1; i < n; i++) {

        auto tmp_child = qalloc(qreg.size());

        auto local_val =

            vqe->execute(xacc::as_shared_ptr(tmp_child.results()), {})[0];

        all_energies.push_back(local_val);

      }

      auto sum = std::accumulate(all_energies.begin(), all_energies.end(), 0.);

      val = sum / all_energies.size();

      double sq_sum = std::inner_product(

          all_energies.begin(), all_energies.end(), all_energies.begin(), 0.0);

      std_dev = std::sqrt(sq_sum / all_energies.size() - val * val);

    }



    std::cout << "<H>(" << this->current_iterate_parameters << ") = " << val;

    if (std::fabs(std_dev) > 1e-12) {

      std::cout << " +- " << std_dev << "\n";

    } else {

      std::cout << std::endl;

    }

    

    // want to store parameters, have to do it here

    for (auto &child : tmp_child.results()->getChildren()) {

      child->addExtraInfo("parameters", current_iterate_parameters);

      auto tmp = current_iterate_parameters;

      tmp.push_back(val);

      child->addExtraInfo("qcor-params-energy", tmp);

      if (std::fabs(std_dev) > 1e-12) {

        child->addExtraInfo("qcor-energy-stddev", std_dev);

      }

      child->addExtraInfo("iteration", current_iteration);

    }

    current_iteration++;

    qreg.addChild(tmp_child);



    if (!dx.empty() && options.stringExists("gradient-strategy")) {
@@ -68,6 +111,7 @@ public: 
  }



public:

  int current_iteration = 0;

  const std::string name() const override { return "vqe"; }

  const std::string description() const override { return ""; }

};