/**
 * @file padam_update.hpp
 * @author Marcus Edel
 *
 * Implementation of Partially adaptive momentum estimation method (Padam).
 *
 * ensmallen is free software; you may redistribute it and/or modify it under
 * the terms of the 3-clause BSD license.  You should have received a copy of
 * the 3-clause BSD license along with ensmallen.  If not, see
 * http://www.opensource.org/licenses/BSD-3-Clause for more information.
 */
#ifndef ENSMALLEN_PADAM_PADAM_UPDATE_HPP
#define ENSMALLEN_PADAM_PADAM_UPDATE_HPP

namespace ens {

/**
 * Partially adaptive momentum estimation method (Padam),
 * adopts historical gradient information to automatically adjust the
 * learning rate.
 *
 * For more information, see the following.
 *
 * @code
 * @article{
 *   title   = {Closing the Generalization Gap of Adaptive Gradient Methods in
 *              Training Deep Neural Networks},
 *   author  = {{Chen}, J. and {Gu}, Q.},
 *   journal = {ArXiv e-prints},
 *   url     = {https://arxiv.org/abs/1806.06763}
 *   year    = {2018}
 * }
 * @endcode
 */
class PadamUpdate
{
 public:
  /**
   * Construct the Padam update policy with the given parameters.
   *
   * @param epsilon Epsilon is the minimum allowed gradient.
   * @param beta1 The smoothing parameter.
   * @param beta2 The second moment coefficient.
   * @param partial Partially adaptive parameter.
   */
  PadamUpdate(const double epsilon = 1e-8,
              const double beta1 = 0.9,
              const double beta2 = 0.999,
              const double partial = 0.25) :
      epsilon(epsilon),
      beta1(beta1),
      beta2(beta2),
      partial(partial),
      iteration(0)
  {
    // Nothing to do.
  }

  //! Get the value used to initialise the squared gradient parameter.
  double Epsilon() const { return epsilon; }
  //! Modify the value used to initialise the squared gradient parameter.
  double& Epsilon() { return epsilon; }

  //! Get the smoothing parameter.
  double Beta1() const { return beta1; }
  //! Modify the smoothing parameter.
  double& Beta1() { return beta1; }

  //! Get the second moment coefficient.
  double Beta2() const { return beta2; }
  //! Modify the second moment coefficient.
  double& Beta2() { return beta2; }

  //! Get the partial adaptive parameter.
  double Partial() const { return partial; }
  //! Modify the partial adaptive parameter.
  double& Partial() { return partial; }

  //! Get the current iteration number.
  size_t Iteration() const { return iteration; }
  //! Modify the current iteration number.
  size_t& Iteration() { return iteration; }

  /**
   * The UpdatePolicyType policy classes must contain an internal 'Policy'
   * template class with two template arguments: MatType and GradType.  This is
   * instantiated at the start of the optimization, and holds parameters
   * specific to an individual optimization.
   */
  template<typename MatType, typename GradType>
  class Policy
  {
   public:
    /**
     * This constructor is called by the SGD Optimize() method before the start
     * of the iteration update process.
     *
     * @param parent Instantiated PadamUpdate parent object.
     * @param rows Number of rows in the gradient matrix.
     * @param cols Number of columns in the gradient matrix.
     */
    Policy(PadamUpdate& parent, const size_t rows, const size_t cols) :
        parent(parent)
    {
      m.zeros(rows, cols);
      v.zeros(rows, cols);
      vImproved.zeros(rows, cols);
    }

    /**
     * Update step for Padam.
     *
     * @param iterate Parameters that minimize the function.
     * @param stepSize Step size to be used for the given iteration.
     * @param gradient The gradient matrix.
     */
    void Update(MatType& iterate,
                const double stepSize,
                const GradType& gradient)
    {
      // Increment the iteration counter variable.
      ++parent.iteration;

      // And update the iterate.
      m *= parent.beta1;
      m += (1 - parent.beta1) * gradient;

      v *= parent.beta2;
      v += (1 - parent.beta2) * (gradient % gradient);

      const double biasCorrection1 = 1.0 - std::pow(parent.beta1,
          parent.iteration);
      const double biasCorrection2 = 1.0 - std::pow(parent.beta2,
          parent.iteration);

      // Element wise maximum of past and present squared gradients.
      vImproved = arma::max(vImproved, v);

      iterate -= (stepSize * std::sqrt(biasCorrection2) / biasCorrection1) *
          m / arma::pow(vImproved + parent.epsilon, parent.partial);
    }

   private:
    //! Instantiated parent object.
    PadamUpdate& parent;

    //! The exponential moving average of gradient values.
    GradType m;

    //! The exponential moving average of squared gradient values.
    GradType v;

    //! The optimal sqaured gradient value.
    GradType vImproved;
  };

 private:
  //! The epsilon value used to initialise the squared gradient parameter.
  double epsilon;

  //! The smoothing parameter.
  double beta1;

  //! The second moment coefficient.
  double beta2;

  //! Partial adaptive parameter.
  double partial;

  //! The number of iterations.
  size_t iteration;
};

} // namespace ens

#endif