random_net.H

Go to the documentation of this file.

 
/*
                          Aleph_w
 
  Data structures & Algorithms
  version 2.0.0b
  https://github.com/lrleon/Aleph-w
 
  This file is part of Aleph-w library
 
  Copyright (c) 2002-2026 Leandro Rabindranath Leon
 
  Permission is hereby granted, free of charge, to any person obtaining a copy
  of this software and associated documentation files (the "Software"), to deal
  in the Software without restriction, including without limitation the rights
  to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  copies of the Software, and to permit persons to whom the Software is
  furnished to do so, subject to the following conditions:
 
  The above copyright notice and this permission notice shall be included in all
  copies or substantial portions of the Software.
 
  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  SOFTWARE.
*/
 
 
# ifndef RANDOM_NET_H
# define RANDOM_NET_H
 
# include <gsl/gsl_rng.h>
# include <gsl/gsl_randist.h>
 
# include <memory>
# include <stdexcept>
# include <algorithm>
# include <tpl_dynArray.H>
# include <tpl_dynSetTree.H>
# include <tpl_netcost.H>
# include <tpl_indexArc.H>
# include <ah-errors.H>
 
 
template <class Net>
class Random_Network_Flow
{
  typedef typename Net::Node Node;
  typedef typename Net::Arc Arc;
 
  DynArray<DynArray<Node *>> rank;
  gsl_rng *r;
  std::unique_ptr<Net> net;
 
public:
  Random_Network_Flow(const unsigned int seed = time(nullptr))
    : r(gsl_rng_alloc(gsl_rng_mt19937)), net(new Net)
  {
    gsl_rng_set(r, seed % gsl_rng_max(r));
  }
 
  ~Random_Network_Flow()
  {
    gsl_rng_free(r);
  }
 
private:
  void connect_ranks(size_t src_idx_rank,
                     size_t tgt_idx_rank,
                     const double & cap_mean,
                     const double & cap_sigma,
                     const double & density)
  {
    IndexArc<Net> arcs(*this->net.get());
    DynArray<Node *> src_rank = rank.access(src_idx_rank);
    DynArray<Node *> tgt_rank = rank.access(tgt_idx_rank);
    const size_t & num_nodes_src_rank = src_rank.size();
    const size_t & num_nodes_tgt_rank = tgt_rank.size();
    for (int i = 0; i < num_nodes_src_rank; ++i)
      {
        auto src = src_rank.access(i);
        const double lambda = density * num_nodes_tgt_rank;
        size_t num_arcs = gsl_ran_exponential(r, lambda);
        num_arcs = std::min(num_arcs, num_nodes_tgt_rank);
        for (int k = 0; k < num_arcs; ++k)
          {
          repeat:
            auto tgt_idx = gsl_rng_uniform_int(r, num_nodes_tgt_rank);
            auto tgt = tgt_rank.access(tgt_idx);
            if (arcs.search(src, tgt) != nullptr)
              goto repeat;
 
            const double cap = gsl_ran_gaussian(r, cap_sigma) + cap_mean;
            this->net->insert_arc(src, tgt, cap);
          }
      }
  }
 
  void create_forward_arcs_in_rank(size_t curr_rank_idx,
                                   const double & cap_mean,
                                   const double & cap_sigma,
                                   const double & forward_density)
  {
    assert(curr_rank_idx < rank.size() - 1);
    connect_ranks(curr_rank_idx, curr_rank_idx + 1,
                  cap_mean, cap_sigma, forward_density);
  }
 
  void create_backward_arcs_in_rank(const size_t curr_rank_idx,
                                    const double & cap_mean,
                                    const double & cap_sigma,
                                    const double & backward_density)
  {
    assert(curr_rank_idx > 0);
    connect_ranks(curr_rank_idx, curr_rank_idx - 1,
                  cap_mean, cap_sigma, backward_density);
  }
 
  // create num_ranks each one with num_nodes_by_rank according to the
  // mean of a normal with rank_sigma as deviation
  void create_ranks(const size_t num_ranks,
                    const size_t num_nodes_by_rank,
                    const double & rank_sigma)
  {
    ah_domain_error_if(rank_sigma < 0)
      << "Rank sigma " + std::to_string(rank_sigma) + " less than 0";
    ah_domain_error_if(rank_sigma > 1)
      << "Rank sigma " + std::to_string(rank_sigma) + " greater than 1";
    for (int i = 0; i < num_ranks; ++i)
      {
        const double mu = gsl_ran_gaussian(r, rank_sigma * num_nodes_by_rank);
        const size_t num_nodes = num_nodes_by_rank + mu;
        rank.touch(i).reserve(num_nodes);
        DynArray<Node *> & curr_rank = rank.access(i);
        for (int k = 0; k < num_nodes; ++k)
          curr_rank.access(k) = this->net->insert_node();
      }
  }
 
  void create_arcs(const double & cap_mean,
                   const double & cap_sigma,
                   const double & forward_density,
                   const double & backward_density)
  {
    ah_domain_error_if(forward_density < backward_density)
      << "forward density less than backward density";
    ah_domain_error_if(forward_density <= 0 or forward_density >= 1)
      << "forward density out of range";
    ah_domain_error_if(backward_density <= 0 or backward_density >= 1)
      << "backward density out of range";
 
    const size_t & N = rank.size();
    
    // Need at least 2 ranks to create inter-rank arcs
    if (N < 2)
      return;
      
    create_forward_arcs_in_rank(0, cap_mean, cap_sigma, forward_density);
    for (int i = 1; i < N - 1; ++i)
      {
        create_forward_arcs_in_rank(i, cap_mean, cap_sigma, forward_density);
        create_backward_arcs_in_rank(i, cap_mean, cap_sigma, backward_density);
      }
    create_backward_arcs_in_rank(N - 1, cap_mean, cap_sigma, backward_density);
  }
 
public:
  Net * operator ()(const size_t num_ranks,
                    const size_t num_nodes_by_rank,
                    const double & rank_sigma = 0.2,
                    const double & cap_mean = 100,
                    const double & cap_sigma = 0.9,
                    const double & forward_density = 0.3,
                    const double & backward_density = 0.01)
  {
    // Create fresh network for this generation
    this->net.reset(new Net);
    rank.empty();
    
    create_ranks(num_ranks, num_nodes_by_rank, rank_sigma);
    create_arcs(cap_mean, cap_sigma, forward_density, backward_density);
 
    return this->net.release();
  }
};
 
 
# endif // RANDOM_NET_H