xref: /dports/science/libint2/libint-2.7.1/src/bin/libint/tactic.h

/*
 *  Copyright (C) 2004-2021 Edward F. Valeev
 *
 *  This file is part of Libint.
 *
 *  Libint is free software: you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation, either version 3 of the License, or
 *  (at your option) any later version.
 *
 *  Libint is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with Libint.  If not, see <http://www.gnu.org/licenses/>.
 *
 */

#include <cmath>
#include <cstdlib>
#include <ctime>
#include <smart_ptr.h>
#include <time.h>

#ifndef _libint2_src_bin_libint_tactic_h_
#define _libint2_src_bin_libint_tactic_h_

namespace libint2 {

  class DirectedGraph;
  class RecurrenceRelation;

  class DummyRandomizePolicy;
  class StdRandomizePolicy;

  /** Tactic is used to choose the optimal (in some sense) recurrence relation to reduce
      a vertex.
  */
  class Tactic {
    public:
    typedef SafePtr<RecurrenceRelation> RR;
    typedef std::vector<RR> rr_stack;

    Tactic() {}
    virtual ~Tactic() {}

    virtual RR optimal_rr(const rr_stack& stack) const =0;

    // make return true if need to debug Tactic classes
    static constexpr bool class_debug() {
      return false;
    }
  };

  /** FirstChoiceTactic simply chooses the first RR
    */
  template <class RandomizePolicy = DummyRandomizePolicy>
    class FirstChoiceTactic : public Tactic {
    public:
    FirstChoiceTactic(const SafePtr<RandomizePolicy>& rpolicy = SafePtr<RandomizePolicy>(new RandomizePolicy)) : Tactic(), rpolicy_(rpolicy) {}
    virtual ~FirstChoiceTactic() {}

    RR optimal_rr(const rr_stack& stack) const {
      if (!stack.empty())
        return stack[0 + rpolicy_->noise(stack.size())];
      else
        return RR();
    }

    private:
    SafePtr<RandomizePolicy> rpolicy_;
  };

  /** FewestNewVerticesTactic chooses RR which adds fewest new vertices to
      DirectedGraph dg
    */
  class FewestNewVerticesTactic : public Tactic {
    public:
    FewestNewVerticesTactic(const SafePtr<DirectedGraph>& dg) : Tactic(), dg_(dg) {}
    virtual ~FewestNewVerticesTactic() {}

    RR optimal_rr(const rr_stack& stack) const;

    private:
    SafePtr<DirectedGraph> dg_;
  };

  /** ZeroNewVerticesTactic chooses first RR which adds no new vertices on
      DirectedGraph dg
    */
  class ZeroNewVerticesTactic : public Tactic {
    public:
    ZeroNewVerticesTactic(const SafePtr<DirectedGraph>& dg) : Tactic(), dg_(dg) {}
    virtual ~ZeroNewVerticesTactic() {}

    RR optimal_rr(const rr_stack& stack) const;

    private:
    SafePtr<DirectedGraph> dg_;
  };

  /** RandomChoiceTactic chooses randomly among the applicable RRs
    */
  class RandomChoiceTactic : public Tactic {
    public:
    RandomChoiceTactic();
    virtual ~RandomChoiceTactic() {}

    RR optimal_rr(const rr_stack& stack) const;
  };

  /** NullTactic always returns null RecurrenceRelation
    */
  class NullTactic : public Tactic {
    public:
    NullTactic() : Tactic() {}
    virtual ~NullTactic() {}

    RR optimal_rr(const rr_stack& stack) const;
  };

  /**
   * ParticleDirectionTactic returns the first RR that transfers the quantum numbers between particles in
   * the desired direction.
   */
  class ParticleDirectionTactic : public Tactic {
    public:
      /**
       * @param increasing if true, quanta should be transferred from lower to higher particle indices.
       */
      ParticleDirectionTactic(bool increase) : Tactic(), increase_(increase) {}
      virtual ~ParticleDirectionTactic() {}

      RR optimal_rr(const rr_stack& stack) const;
    private:
      bool increase_;
  };

  /**
   * TwoCenter_OS_Tactic decides graph build for <bra0|ket0>
   */
  class TwoCenter_OS_Tactic : public Tactic {
    public:
      /**
       * @param lbra0
       * @param lket0
       */
    TwoCenter_OS_Tactic(unsigned lbra0,
                        unsigned lket0) : Tactic(), lbra0_(lbra0), lket0_(lket0) {}
      virtual ~TwoCenter_OS_Tactic() {}

      RR optimal_rr(const rr_stack& stack) const;
    private:
      unsigned lbra0_;
      unsigned lket0_;
  };

  /**
   * FourCenter_OS_Tactic decides graph build for (bra0 ket0| bra1 ket1) = <bra0 bra1|ket0 ket1>
   */
  class FourCenter_OS_Tactic : public Tactic {
    public:
      /**
       * @param lbra0
       * @param lbra1
       * @param lket0
       * @param lket1
       */
      FourCenter_OS_Tactic(unsigned lbra0,
                           unsigned lket0,
                           unsigned lbra1,
                           unsigned lket1) : Tactic(), lbra0_(lbra0), lket0_(lket0),
                           lbra1_(lbra1), lket1_(lket1) {}
      virtual ~FourCenter_OS_Tactic() {}

      RR optimal_rr(const rr_stack& stack) const;
    private:
      unsigned lbra0_;
      unsigned lket0_;
      unsigned lbra1_;
      unsigned lket1_;
  };

  /////////////////////////////////

  struct DummyRandomizePolicy {
    unsigned int noise(unsigned int nrrs) const { return 0; }
  };

  /** The shift parameter is computed as follows:
      delta = floor(nrrs*scale*random()/RAND_MAX)
      where nrrs is the number of possibilities, scale
      is the user-specified parameter.
  */
  class StdRandomizePolicy {
    public:
    StdRandomizePolicy(double scale) : scale_(scale) {
      // Initialize state randomly
      time_t t;
      srandom(time(&t));
    }

    unsigned int noise(unsigned int nrrs) const {
      unsigned long rand = random();
      const unsigned long range = RAND_MAX;
      const unsigned int result = static_cast<unsigned int>(std::floor(nrrs*scale_*rand/range));
      return result;
    }

    private:
    double scale_;
  };


};

#endif