Merge pull request #28 from seldon-code/develop

Network refactor and reluctant Activity driven model

.github/workflows/ci.yml

@@ -34,4 +34,4 @@ jobs:
     - name: Test with meson
       shell: micromamba-shell {0}
       run: |
-        meson test -C build
+        meson test -C build --verbose

.gitmessage

@@ -0,0 +1,19 @@
+# Title: Summary, imperative, start upper case, don't end with a period
+# No more than 50 chars. #### 50 chars is here:  #
+
+# Remember blank line between title and body.
+
+# Body: Explain *what* and *why* (not *how*).
+# Wrap at 72 chars. ################################## which is here:  #
+
+# At the end: Include Co-authored-by for all contributors. 
+# Include at least one empty line before it. Format: 
+# Co-authored-by: name <[email protected]>
+
+# 1. Separate subject from body with a blank line
+# 2. Limit the subject line to 50 characters
+# 3. Capitalize the subject line
+# 4. Do not end the subject line with a period
+# 5. Use the imperative mood in the subject line
+# 6. Wrap the body at 72 characters
+# 7. Use the body to explain what and why vs. how

CONTRIBUTING.md

@@ -52,3 +52,12 @@ A good commit should have:
 <!-- * Follow our [style guide][style]. -->
 
 <!-- [style]: https://github.com/thoughtbot/guides/tree/master/style -->
+
+### Commit template
+If you are not already using a commit template, consider doing so. The Seldon repository includes a template under `.gitmessage`.
+
+```
+git config --global commit.template .gitmessage
+```
+
+You can omit the `--global` to only use the template when committing to Seldon.

examples/ActivityDriven/conf.toml

@@ -6,6 +6,8 @@ model = "ActivityDriven"
 n_output_network = 20 # Write the network every 20 iterations
 n_output_agents = 1 # Write the opinions of agents after every iteration
 print_progress = true # Print the iteration time ; if not set, then always print
+print_initial = true # Print the initial opinions and network file from step 0. If not set, this is true by default.
+start_output = 2 # Start writing out opinions and/or network files from this iteration. If not set, this is 1.
 
 [model]
 max_iterations = 500 # If not set, max iterations is infinite

examples/ActivityDrivenBot/conf.toml

@@ -6,6 +6,8 @@ rng_seed = 120           # Leaving this empty will pick a random seed
 n_output_network = 1  # Write the network every 20 iterations
 n_output_agents = 1 # Write the opinions of agents after every iteration
 print_progress = true # Print the iteration time; if not set, then always print
+print_initial = true # Print the initial opinions and network file from step 0. If not set, this is true by default.
+start_output = 1 # Start writing out opinions and/or network files from this iteration. If not set, this is 1.
 
 [model]
 max_iterations = 1000 # If not set, max iterations is infinite

examples/ActivityDrivenMeanField/conf.toml

@@ -6,6 +6,8 @@ rng_seed = 12345678 # Leaving this empty will pick a random seed
 n_output_network = 20 # Write the network every 20 iterations
 n_output_agents = 1 # Write the opinions of agents after every iteration
 print_progress = true # Print the iteration time ; if not set, then always print
+print_initial = true # Print the initial opinions and network file from step 0. If not set, this is true by default.
+start_output = 1 # Start writing out opinions and/or network files from this iteration. If not set, this is 1.
 
 [model]
 max_iterations = 2000 # If not set, max iterations is infinite

examples/ActivityDrivenReluctance/conf.toml

@@ -0,0 +1,32 @@
+[simulation]
+model = "ActivityDriven"
+# rng_seed = 120 # Leaving this empty will pick a random seed
+
+[io]
+n_output_network = 20 # Write the network every 20 iterations
+n_output_agents = 1   # Write the opinions of agents after every iteration
+print_progress = true # Print the iteration time ; if not set, then always print
+
+[model]
+max_iterations = 500 # If not set, max iterations is infinite
+
+[ActivityDriven]
+dt = 0.01               # Timestep for the integration of the coupled ODEs
+m = 10                  # Number of agents contacted, when the agent is active
+eps = 0.01              # Minimum activity epsilon; a_i belongs to [epsilon,1]
+gamma = 2.1             # Exponent of activity power law distribution of activities
+reciprocity = 0.65      # probability that when agent i contacts j via weighted reservoir sampling, j also sends feedback to i. So every agent can have more than m incoming connections
+homophily = 1.0         # aka beta. if zero, agents pick their interaction partners at random
+alpha = 3.0             # Controversialness of the issue, must be greater than 0.
+K = 2.0
+mean_activities = false # Use the mean value of the powerlaw distribution for the activities of all agents
+mean_weights = false    # Use the meanfield approximation of the network edges
+
+reluctances = true       # Assigns a "reluctance" (m_i) to each agent. By default; false and every agent has a reluctance of 1
+reluctance_mean = 1.0    # Mean of distribution before drawing from a truncated normal distribution (default set to 1.0)
+reluctance_sigma = 0.25  # Width of normal distribution (before truncating)
+reluctance_eps = 0.01    # Minimum such that the normal distribution is truncated at this value
+
+[network]
+number_of_agents = 1000
+connections_per_agent = 10

examples/DeGroot/conf.toml

@@ -6,6 +6,8 @@ model = "DeGroot"
 n_output_network = 20 # Write the network every 20 iterations
 n_output_agents = 1 # Write the opinions of agents after every iteration
 print_progress = false # Print the iteration time ; if not set, then always prints
+print_initial = true # Print the initial opinions and network file from step 0. If not set, this is true by default.
+start_output = 1 # Start writing out opinions and/or network files from this iteration. If not set, this is 1.
 
 [model]
 max_iterations = 20 # If not set, max iterations is infinite

include/agent.hpp

@@ -1,5 +1,4 @@
 #pragma once
-#include "agent_base.hpp"
 #include <fmt/format.h>
 namespace Seldon
 {
@@ -8,17 +7,18 @@ namespace Seldon
 (which contains an opinion and perhaps some other things),
 it needs to implement to_string and from_string*/
 template<typename T>
-class Agent : public AgentBase
+class Agent
 {
 public:
     using data_t = T;
     data_t data;
     Agent() = default;
     Agent( data_t data ) : data( data ) {}
+    virtual ~Agent() = default;
 
     void from_string( const std::string & str );
 
-    std::string to_string() const override
+    virtual std::string to_string() const
     {
         return fmt::format( "{:.16f}", data );
     }

include/agent_generation.hpp

@@ -26,7 +26,7 @@ std::vector<AgentT> generate_from_file( const std::string & file )
         // Get the current line as a substring
         auto line     = file_contents.substr( start_of_line, end_of_line - start_of_line );
         start_of_line = end_of_line + 1;
-        // TODO: check if empty or comment
+
         if( line.empty() )
         {
             break;

include/config_parser.hpp

@@ -32,6 +32,8 @@ struct OutputSettings
     std::optional<size_t> n_output_agents  = std::nullopt;
     std::optional<size_t> n_output_network = std::nullopt;
     bool print_progress                    = true; // Print the iteration time, by default always prints
+    bool print_initial                     = true; // Output initial opinions and network, by default always outputs.
+    int start_output = 1; // Start printing opinion and/or network files from this iteration number
 };
 
 struct DeGrootSettings
@@ -58,6 +60,11 @@ struct ActivityDrivenSettings
     std::vector<double> bot_activity  = std::vector<double>( 0 );
     std::vector<double> bot_opinion   = std::vector<double>( 0 );
     std::vector<double> bot_homophily = std::vector<double>( 0 );
+    bool use_reluctances              = false;
+    double reluctance_mean            = 1.0;
+    double reluctance_sigma           = 0.25;
+    double reluctance_eps             = 0.01;
+    double covariance_factor          = 0.0;
 };
 
 struct InitialNetworkSettings

include/model.hpp

@@ -1,5 +1,5 @@
 #pragma once
-#include "model_base.hpp"
+#include <cstddef>
 #include <optional>
 
 namespace Seldon
@@ -8,37 +8,44 @@ namespace Seldon
 /* Model<T> is a base class from which the acutal models would derive. They have efficient access to a vector of AgentT,
  * without any pointer indirections */
 template<typename AgentT_>
-class Model : public ModelBase
+class Model
 {
 public:
     using AgentT = AgentT_;
-    std::vector<AgentT> agents;
 
-    std::optional<int> max_iterations = std::nullopt;
-    Model( size_t n_agents ) : agents( std::vector<AgentT>( int( n_agents ), AgentT() ) ) {}
-    Model( std::vector<AgentT> && agents ) : agents( agents ) {}
+    std::optional<size_t> max_iterations = std::nullopt;
 
-    void iteration() override
+    virtual void initialize_iterations()
     {
-        n_iterations++;
+        _n_iterations = 0;
+    }
+
+    virtual void iteration()
+    {
+        _n_iterations++;
     };
 
-    bool finished() override
+    size_t n_iterations()
+    {
+        return _n_iterations;
+    }
+
+    virtual bool finished()
     {
         if( max_iterations.has_value() )
         {
-            return max_iterations.value() <= n_iterations;
+            return max_iterations.value() <= n_iterations();
         }
         else
         {
             return false;
         }
     };
 
-    AgentBase * get_agent( int idx ) override // For this to work AgentT needs to be a subclass of AgentBase
-    {
-        return &agents[idx];
-    }
+    virtual ~Model() = default;
+
+private:
+    size_t _n_iterations{};
 };
 
 } // namespace Seldon

include/models/ActivityDrivenModel.hpp

@@ -6,7 +6,7 @@
 #include <cstddef>
 #include <random>
 #include <set>
-#include <stdexcept>
+#include <string>
 #include <utility>
 #include <vector>
 
@@ -15,32 +15,49 @@ namespace Seldon
 
 struct ActivityAgentData
 {
-    double opinion  = 0; // x_i
-    double activity = 0; // a_i
+    double opinion    = 0;   // x_i
+    double activity   = 0;   // a_i
+    double reluctance = 1.0; // m_i
 };
 
 template<>
 inline std::string Agent<ActivityAgentData>::to_string() const
 {
-    return fmt::format( "{}, {}", data.opinion, data.activity );
+    return fmt::format( "{}, {}, {}", data.opinion, data.activity, data.reluctance );
 };
 
 template<>
 inline void Agent<ActivityAgentData>::from_string( const std::string & str )
 {
-    auto pos_comma = str.find_first_of( ',' );
-    data.opinion   = std::stod( str.substr( 0, pos_comma ) );
-    data.activity  = std::stod( str.substr( pos_comma + 1, str.size() ) );
+    auto pos_comma      = str.find_first_of( ',' );
+    auto pos_next_comma = str.find( ',', pos_comma + 1 );
+
+    data.opinion = std::stod( str.substr( 0, pos_comma ) );
+
+    if( pos_next_comma == std::string::npos )
+    {
+        data.activity = std::stod( str.substr( pos_comma + 1, str.size() ) );
+    }
+    else
+    {
+        data.activity = std::stod( str.substr( pos_comma + 1, pos_next_comma ) );
+    }
+
+    if( pos_next_comma != std::string::npos )
+    {
+        data.reluctance = std::stod( str.substr( pos_next_comma + 1, str.size() ) );
+    }
 };
 
-class ActivityAgentModel : public Model<Agent<ActivityAgentData>>
+class ActivityDrivenModel : public Model<Agent<ActivityAgentData>>
 {
 public:
-    using AgentT = Agent<ActivityAgentData>;
+    using AgentT   = Agent<ActivityAgentData>;
+    using NetworkT = Network<AgentT>;
 
 private:
-    Network & network;
-    std::vector<std::vector<Network::WeightT>> contact_prob_list; // Probability of choosing i in 1 to m rounds
+    NetworkT & network;
+    std::vector<std::vector<NetworkT::WeightT>> contact_prob_list; // Probability of choosing i in 1 to m rounds
     // Random number generation
     std::mt19937 & gen; // reference to simulation Mersenne-Twister engine
     std::set<std::pair<size_t, size_t>> reciprocal_edge_buffer{};
@@ -68,7 +85,8 @@ class ActivityAgentModel : public Model<Agent<ActivityAgentData>>
             for( size_t j = 0; j < neighbour_buffer.size(); j++ )
             {
                 j_index = neighbour_buffer[j];
-                k_buffer[idx_agent] += K * weight_buffer[j] * std::tanh( alpha * opinion( j_index ) );
+                k_buffer[idx_agent] += 1.0 / network.agents[idx_agent].data.reluctance * K * weight_buffer[j]
+                                       * std::tanh( alpha * opinion( j_index ) );
             }
             // Multiply by the timestep
             k_buffer[idx_agent] *= dt;
@@ -100,6 +118,12 @@ class ActivityAgentModel : public Model<Agent<ActivityAgentData>>
 
     double convergence_tol = 1e-12; // TODO: ??
 
+    bool use_reluctances     = false;
+    double reluctance_mean   = 1.0;
+    double reluctance_sigma  = 0.25;
+    double reluctance_eps    = 0.01;
+    double covariance_factor = 0.0;
+
     // bot @TODO: less hacky
 
     size_t n_bots                     = 0; // The first n_bots agents are bots
@@ -113,14 +137,11 @@ class ActivityAgentModel : public Model<Agent<ActivityAgentData>>
         return n_bots > 0;
     }
 
-    ActivityAgentModel( int n_agents, Network & network, std::mt19937 & gen );
+    ActivityDrivenModel( NetworkT & network, std::mt19937 & gen );
 
     void get_agents_from_power_law(); // This needs to be called after eps and gamma have been set
 
     void iteration() override;
-
-    // bool finished() overteration() override;
-    // bool finished() override;
 };
 
 } // namespace Seldon