Merge branch 'master' into pep8

CONTRIBUTORS.md

@@ -0,0 +1,6 @@
+Michael Fenton
+David Fagan
+James McDermott
+Stefan Forstenlechner
+Erik Hemberg
+Michael O'Neill

src/algorithm/parameters.py

@@ -301,6 +301,14 @@ def set_params(command_line_args, create_files=True):
             params['GENERATION_SIZE'] = params['POPULATION_SIZE'] - \
                                         params['ELITE_SIZE']
 
+        if (params['MUTATION_PROBABILITY'] is not None and
+            params['MUTATION_EVENTS'] != 1):
+            s = "operators.mutation.int_flip_per_codon\n" \
+                "Error: mutually exclusive parameters 'MUTATION_PROBABILITY'" \
+                "and 'MUTATION_EVENTS' have been explicitly set.\n" \
+                "Only one of these parameters can be used at a time."
+            raise Exception(s)
+
         # Initialise run lists and folders before we set imports.r
         initialise_run_params(create_files)
 

src/fitness/progsys.py

@@ -1,12 +1,13 @@
 import json
-import subprocess
-import sys
+from subprocess import Popen, PIPE
+from sys import maxsize, executable
 from os import path
 
 from algorithm.parameters import params
 from fitness.base_ff_classes.base_ff import base_ff
 
 
+
 class progsys(base_ff):
     """Fitness function for program synthesis problems. Grammars and datasets
     for 29 benchmark problems from doi.org/10.1145/2739480.2754769 are
@@ -66,10 +67,8 @@ def evaluate(self, ind, **kwargs):
     @staticmethod
     def create_eval_process():
         """create separate python process for evaluation"""
-        return subprocess.Popen(['python',
-                                 'scripts/python_script_evaluation.py'],
-                                stdout=subprocess.PIPE,
-                                stdin=subprocess.PIPE)
+        return Popen([sys.executable, 'scripts/python_script_evaluation.py'],
+                     stdout=PIPE, stdin=PIPE)
 
     def format_program(self, individual, header, footer):
         """formats the program by formatting the individual and adding

src/operators/mutation.py

@@ -71,20 +71,12 @@ def int_flip_per_codon(ind):
         # Linear mutation cannot be performed on this individual.
         return ind
 
-    # Set mutation probability. Default is 1 over the length of the genome.
-    if params['MUTATION_PROBABILITY'] and params['MUTATION_EVENTS'] == 1:
+    # Set mutation probability. 
+    if params['MUTATION_PROBABILITY'] is not None:
         p_mut = params['MUTATION_PROBABILITY']
-    elif params['MUTATION_PROBABILITY'] and params['MUTATION_EVENTS'] > 1:
-        s = "operators.mutation.int_flip_per_codon\n" \
-            "Error: mutually exclusive parameters for 'MUTATION_PROBABILITY'" \
-            "and 'MUTATION_EVENTS' have been explicitly set.\n" \
-            "       Only one of these parameters can be used at a time with" \
-            "int_flip_per_codon mutation."
-        raise Exception(s)
     else:
-        # Default mutation events per individual is 1. Raising this number
-        # will influence the mutation probability for each codon.
-        p_mut = params['MUTATION_EVENTS'] / eff_length
+        # Default is 1 divided by genome length.
+        p_mut = 1.0 / eff_length
 
     # Mutation probability works per-codon over the portion of the
     # genome as defined by the within_used flag.

src/representation/derivation.py

@@ -43,16 +43,10 @@ def generate_tree(tree, genome, output, method, nodes, depth, max_depth,
     available = legal_productions(method, remaining_depth, tree.root,
                                   productions['choices'])
 
-    # Randomly pick a production choice.
+    # Randomly pick a production choice and make a codon with it.
     chosen_prod = choice(available)
-
-    # Find the index of the chosen production and set a matching codon based
-    # on that index.
-    prod_index = productions['choices'].index(chosen_prod)
-    codon = randrange(productions['no_choices'],
-                      params['BNF_GRAMMAR'].codon_size,
-                      productions['no_choices']) + prod_index
-
+    codon = generate_codon(chosen_prod, productions)
+
     # Set the codon for the current node and append codon to the genome.
     tree.codon = codon
     genome.append(codon)
@@ -93,6 +87,30 @@ def generate_tree(tree, genome, output, method, nodes, depth, max_depth,
     return genome, output, nodes, depth, max_depth
 
 
+def generate_codon(chosen_prod, productions):
+    """
+    Generate a single codon
+    
+    :param chosen_prod: the specific production to build a codon for
+    :param productions: productions possible from the current root
+    :return: a codon integer
+    
+    """
+
+    # Find the index of the chosen production
+    production_index = productions['choices'].index(chosen_prod)
+
+    # Choose a random offset with guarantee that (offset + production_index) < codon_size
+    offset = randrange(
+        start=0,
+        stop=params['BNF_GRAMMAR'].codon_size - productions['no_choices'] + 1,
+        step=productions['no_choices']
+    )
+
+    codon = offset + production_index
+    return codon
+
+
 def legal_productions(method, depth_limit, root, productions):
     """
     Returns the available production choices for a node given a specific
@@ -114,8 +132,8 @@ def legal_productions(method, depth_limit, root, productions):
 
     if method == "random":
         # Randomly build a tree.
-
-        if not depth_limit:
+        
+        if depth_limit is None:
             # There is no depth limit, any production choice can be used.
             available = productions
 
@@ -224,15 +242,9 @@ def pi_random_derivation(tree, max_depth):
         available = legal_productions("random", remaining_depth, node.root,
                                       productions['choices'])
 
-        # Randomly pick a production choice.
+        # Randomly pick a production choice and make a codon with it.
         chosen_prod = choice(available)
-
-        # Find the index of the chosen production and set a matching codon
-        # based on that index.
-        prod_index = productions['choices'].index(chosen_prod)
-        codon = randrange(productions['no_choices'],
-                          params['BNF_GRAMMAR'].codon_size,
-                          productions['no_choices']) + prod_index
+        codon = generate_codon(chosen_prod, productions)
 
         # Set the codon for the current node and append codon to the genome.
         node.codon = codon
@@ -333,15 +345,9 @@ def pi_grow(tree, max_depth):
             available = legal_productions("random", remaining_depth, node.root,
                                           productions['choices'])
 
-        # Randomly pick a production choice.
+        # Randomly pick a production choice and make a codon with it.
         chosen_prod = choice(available)
-
-        # Find the index of the chosen production and set a matching codon
-        # based on that index.
-        prod_index = productions['choices'].index(chosen_prod)
-        codon = randrange(productions['no_choices'],
-                          params['BNF_GRAMMAR'].codon_size,
-                          productions['no_choices']) + prod_index
+        codon = generate_codon(chosen_prod, productions)
 
         # Set the codon for the current node and append codon to the genome.
         node.codon = codon

src/scripts/experiment_manager.py

@@ -3,8 +3,8 @@
 
     Copyright (c) 2014 Michael Fenton
     Hereby licensed under the GNU GPL v3."""
-from sys import path
 
+from sys import path, executable
 path.append("../src")
 
 from utilities.algorithm.general import check_python_version
@@ -26,7 +26,7 @@ def execute_run(seed):
     :return: Nothing.
     """
 
-    exec_str = "python3 ponyge.py " \
+    exec_str = executable + " ponyge.py " \
                "--random_seed " + str(seed) + " " + " ".join(sys.argv[1:])
 
     call(exec_str, shell=True)

src/utilities/fitness/error_metric.py

@@ -1,7 +1,7 @@
 import warnings
 
 import numpy as np
-from sklearn.metrics.classification import f1_score as sklearn_f1_score
+from sklearn.metrics import f1_score as sklearn_f1_score
 
 
 def mae(y, yhat):

src/utilities/stats/file_io.py

@@ -125,7 +125,7 @@ def generate_folders_and_files():
 
         if not path.isdir(path_1):
             # Create results folder.
-            makedirs(path_1)
+            makedirs(path_1, exist_ok=True)
 
         # Set file path to include experiment name.
         params['FILE_PATH'] = path.join(path_1, params['EXPERIMENT_NAME'])
@@ -136,12 +136,12 @@ def generate_folders_and_files():
 
     # Generate save folders
     if not path.isdir(params['FILE_PATH']):
-        makedirs(params['FILE_PATH'])
+        makedirs(params['FILE_PATH'], exist_ok=True)
 
     if not path.isdir(path.join(params['FILE_PATH'],
                                 str(params['TIME_STAMP']))):
         makedirs(path.join(params['FILE_PATH'],
-                           str(params['TIME_STAMP'])))
+                           str(params['TIME_STAMP'])), exist_ok=True)
 
     params['FILE_PATH'] = path.join(params['FILE_PATH'],
                                     str(params['TIME_STAMP']))