Add experiments script

.gitignore

@@ -3,7 +3,6 @@
 **.csv
 **.spec
 **.tex
-**.sh
 **.pkl
 **/dist
 **/build

batch-run.py

@@ -19,11 +19,11 @@ def main():
                         help='The identifier of this execution, to be used for naming the output files.')
     parser.add_argument('--seed', default=42,
                         help='Seed for reproducibility (set seed=None if reproducibility is not required). Default is 42.')
-    parser.add_argument('--max_iterations', type=int, default=1000,
+    parser.add_argument('--max_iterations', type=int, default=5000,
                         help='The maximum number of iterations of the system. Default is 1000.')
     parser.add_argument('--n', nargs="+", type=int, default=1000,
                         help='The number of players (natural number). Default is 100.')
-    parser.add_argument('--k', nargs="+", type=int, default=[95, 110, 2],
+    parser.add_argument('--k', nargs="+", type=int, default=[100, 200, 2],
                         help='The k value of the system (natural number). Default is 10.')
     parser.add_argument('--alpha', nargs="+", type=float, default=0.3,
                         help='The alpha value of the system (decimal number between 0 and 1). Default is 0.3')
@@ -66,6 +66,7 @@ def main():
         "median_pools_per_operator": sim.get_median_pools_per_operator,
         "avgSatRate": sim.get_avg_sat_rate,
         "nakamotoCoeff": sim.get_nakamoto_coefficient,
+        "StatisticalDistance": sim.get_controlled_stake_distr_stat_dist,
         # "NCR": sim.get_NCR,
         #"MinAggregatePledge": sim.get_min_aggregate_pledge,
         # "pledge_rate": sim.get_pledge_rate,
@@ -93,15 +94,15 @@ def main():
     run_data_MP = batch_run_MP.get_model_vars_dataframe()
     # print(run_data_MP.head())
 
-    output_dir = "output"
+    output_dir = "output/19-11-21"
     pickled_batch_run_data = output_dir + "/batch-run-data.pkl"
     with open(pickled_batch_run_data, "wb") as pkl_file:
         pkl.dump(run_data_MP, pkl_file)
 
     variable_param = list(variable_params.keys())[0]
     colours = [np.random.rand(3, ) for i in range(len(model_reporters))]
     for i, model_reporter in enumerate(model_reporters):
-        plot_aggregate_data(run_data_MP, variable_param, model_reporter, colours[i], args_dict["execution_id"])
+        plot_aggregate_data(run_data_MP, variable_param, model_reporter, colours[i], args_dict["execution_id"], output_dir)
 
     # ordered dicts with data from each step of each run (the combinations of variable params act as the keys)
     # for example data_collector_model[(0.1, 0.02, 1)] shows the values of the parameters collected at model level
@@ -112,8 +113,8 @@ def main():
     plt.show()
 
 
-def plot_aggregate_data(df, variable_param, model_reporter, color, exec_id):
-    figures_dir = "output/figures/"
+def plot_aggregate_data(df, variable_param, model_reporter, color, exec_id, output_dir):
+    figures_dir = output_dir + "/figures/"
     plt.figure()
     plt.scatter(df[variable_param], df[model_reporter], color=color)
     plt.xlabel(variable_param)

experiments.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+python3.9 main.py --simulation_id=baseline-100-10 --n=100 --k=10 --alpha=0.3
+
+python3.9 main.py --simulation_id=baseline-1000-100 --n=1000 --k=100 --alpha=0.3
+
+python3.9 main.py --simulation_id=abstention-30-100-10 --n=100 --k=10 --alpha=0.3 --abstention_rate=0.3
+
+python3.9 main.py --simulation_id=parameter-change-k-100 --n=100 --k 10 20 --alpha=0.3
+
+python3.9 main.py --simulation_id=parameter-change-alpha-100-10 --n=100 --k=10 --alpha 0.03 0.3
+
+python3.9 main.py --simulation_id=parameter-change-k-1000 --n=1000 --k 100 200 --alpha=0.3
+
+python3.9 main.py --simulation_id=parameter-change-alpha-1000-100 --n=1000 --k=100 --alpha 0.03 0.3
+
+python3.9 main.py --simulation_id=alpha-zero-100-10 --n=100 --k=10 --alpha=0
+
+python3.9 main.py --simulation_id=alpha-zero-1000-100 --n=1000 --k=100 --alpha=0
+
+python3.9 batch-run.py --execution_id=varying-k-100 --n=100 --k 1 51 1 --alpha=0.3
+
+python3.9 batch-run.py --execution_id=varying-alpha-1000-100 --n=1000 --k=100 --alpha 0 10 0.25
+
+python3.9 batch-run.py --execution_id=varying-abstention-rate-1000-100 --n=1000 --k=100 --alpha=0.3 --abstention_rate 0 0.91 0.01
+
+python3.9 batch-run.py --execution_id=varying-k-1000 --n=1000 --k 1 251 2 --alpha=0.3
+
+
+
+

logic/model_reporters.py

@@ -164,9 +164,11 @@ def get_controlled_stake_mean_abs_diff(model):
 def get_controlled_stake_distr_stat_dist(model):
     """
     :param model:
-    :return: the statistical difference of the distributions of the stake that players control
+    :return: the statistical distance of the distributions of the stake that players control
                 (how they started vs how they ended up)
     """
+    if not model.has_converged():
+        return -1
     active_players = {player_id: player for player_id, player in model.get_players_dict().items() if
                       not player.abstains}
     pools = model.get_pools_list()

logic/sim.py

@@ -245,7 +245,7 @@ def dump_state_to_csv(self):
               "Private" if pool.is_private else "Public"]
              for pool in pools])
 
-        output_dir = "output/"
+        output_dir = "output/19-11-21/"
         path = pathlib.Path.cwd() / output_dir
         pathlib.Path(path).mkdir(parents=True, exist_ok=True)
         filename = (path / (self.simulation_id + '-final_configuration.csv')) \

logic/stakeholder.py

@@ -173,7 +173,7 @@ def calculate_delegator_utility(self, pool, stake_allocation):
         m_factor = (1 - pool.margin) * q
         u_0 = (r - pool.cost)
         u = m_factor * u_0
-        utility = max(0, u)
+        utility = u if u > 0 else 0
         return utility
 
     # how does a myopic player decide whether to open a pool or not? -> for now we assume that all players play non-myopically when it comes to pool moves

main.py

@@ -10,9 +10,9 @@ def main():
     print("Let the Pooling Games begin!")
 
     parser = argparse.ArgumentParser(description='Pooling Games')
-    parser.add_argument('--n', type=int, default=1000,
+    parser.add_argument('--n', type=int, default=100,
                         help='The number of players (natural number). Default is 100.')
-    parser.add_argument('--k', nargs="+", type=int, default=40,
+    parser.add_argument('--k', nargs="+", type=int, default=10,
                         help='The k value of the system (natural number). Default is 10.')
     parser.add_argument('--alpha', nargs="+", type=float, default=0.3,
                         help='The alpha value of the system (decimal number between 0 and 1). Default is 0.3')
@@ -91,8 +91,8 @@ def main():
     with open(pickled_simulation_filename, "wb") as pkl_file:
         pkl.dump(sim, pkl_file)
 
-    output_dir = "output/"
-    figures_dir = "output/figures/"
+    output_dir = "output/19-11-21/"
+    figures_dir = output_dir + "figures/"
     path = pathlib.Path.cwd() / figures_dir
     pathlib.Path(path).mkdir(parents=True, exist_ok=True)
 
@@ -112,16 +112,16 @@ def main():
     pivot_steps = sim.pivot_steps
 
     plot_line(simulation_id, sim_df["#Pools"], 'C0', "Number of pools over time", "Round",
-              "#Pools", "poolCount", equilibrium_steps, pivot_steps, True)
+              "#Pools", "poolCount", equilibrium_steps, pivot_steps, figures_dir, True)
 
     plot_line(simulation_id, sim_df["AvgPledge"], 'red', "Average pledge over time", "Round",
-              "Average pledge", "avgPledge", equilibrium_steps, pivot_steps, True)
+              "Average pledge", "avgPledge", equilibrium_steps, pivot_steps, figures_dir, True)
 
     plot_line(simulation_id, sim_df["TotalPledge"], 'purple', "Total pledge over time", "Round",
-              "Total pledge", "totalPledge", equilibrium_steps, pivot_steps, True)
+              "Total pledge", "totalPledge", equilibrium_steps, pivot_steps, figures_dir, True)
 
     plot_line(simulation_id, sim_df["MeanAbsDiff"], 'green', "Mean Absolute Difference of Controlled Stake", "Round",
-              "Mean abs diff", "meanAbsDiff", equilibrium_steps, pivot_steps, False)
+              "Mean abs diff", "meanAbsDiff", equilibrium_steps, pivot_steps, figures_dir, False)
 
     '''pool_sizes_by_step = sim_df["PoolSizes"]  # todo fix
         # print(pool_sizes_by_step)
@@ -136,8 +136,7 @@ def main():
 
 
 def plot_line(simulation_id, data, color, title, x_label, y_label, filename, equilibrium_steps, pivot_steps,
-              show_equilibrium=False):
-    figures_dir = "output/figures/"
+              figures_dir, show_equilibrium=False):
     path = pathlib.Path.cwd() / figures_dir
     pathlib.Path(path).mkdir(parents=True, exist_ok=True)
 
@@ -153,7 +152,7 @@ def plot_line(simulation_id, data, color, title, x_label, y_label, filename, equ
     for i, step in enumerate(pivot_steps):
         label = "Parameter change" if i == 0 else ""
         plt.plot(step, data[step], 'x', label=label, c=pivot_colour)
-    plt.title(title)
+    #plt.title(title)
     plt.xlabel(x_label)
     plt.ylabel(y_label)
     plt.legend()