Fix counting of mutations in genome.mutate

cgp/genome.py

@@ -361,6 +361,10 @@ def mutate(self, mutation_rate: float, rng: np.random.RandomState):
         bool
             True if only inactive regions of the genome were mutated, False otherwise.
         """
+
+        def count_dna_differences(dna0: List[int], dna1: List[int]) -> int:
+            return len([1 for gene0, gene1 in zip(dna0, dna1) if gene0 != gene1])
+
         n_mutations = int(mutation_rate * len(self.dna))
         assert n_mutations > 0
 
@@ -369,9 +373,8 @@ def mutate(self, mutation_rate: float, rng: np.random.RandomState):
 
         dna = list(self._dna)
 
-        successful_mutations = 0
         only_silent_mutations = True
-        while successful_mutations < n_mutations:
+        while count_dna_differences(self.dna, dna) < n_mutations:
 
             gene_idx = rng.randint(0, self._n_genes)
             region_idx = gene_idx // self._length_per_region
@@ -380,31 +383,30 @@ def mutate(self, mutation_rate: float, rng: np.random.RandomState):
                 continue  # nothing to do here
 
             elif self._is_output_region(region_idx):
-                success = self._mutate_output_region(dna, gene_idx, rng)
-                if success:
-                    silent = False
-                    only_silent_mutations = only_silent_mutations and silent
-                    successful_mutations += 1
+                silent = self._mutate_output_region(dna, gene_idx, rng)
 
             elif self._is_hidden_region(region_idx):
                 silent = self._mutate_hidden_region(dna, gene_idx, active_regions, rng)
-                only_silent_mutations = only_silent_mutations and silent
 
             else:
                 assert False  # should never be reached
 
+            only_silent_mutations = only_silent_mutations and silent
+
         self.dna = dna
         return only_silent_mutations
 
-    def _mutate_output_region(self, dna: List[int], gene_idx: int, rng: np.random.RandomState):
+    def _mutate_output_region(
+        self, dna: List[int], gene_idx: int, rng: np.random.RandomState
+    ) -> bool:
         assert self._is_gene_in_output_region(gene_idx)
 
         if not self._is_function_gene(gene_idx) and self._is_active_input_gene(gene_idx):
             permissible_inputs = self._permissible_inputs_for_output_region()
             dna[gene_idx] = rng.choice(permissible_inputs)
-            return True
-        else:
             return False
+        else:
+            return True
 
     def _mutate_hidden_region(
         self, dna: List[int], gene_idx: int, active_regions: List[int], rng: np.random.RandomState

examples/example_evo_regression.py

@@ -141,7 +141,8 @@ def recording_callback(pop):
 
         f_graph = champion.to_func()
         x_0_range = np.linspace(-5.0, 5.0, 20)
-        x_1_range = np.ones_like(x_0_range) * 2.0  # fix x_1 such than 1d plot makes sense
+        x_1_range = np.ones_like(x_0_range) * 2.0
+        # fix x_1 such than 1d plot makes sense
         y = [f_graph([x_0, x_1_range[0]]) for x_0 in x_0_range]
         y_target = target_function(np.hstack([x_0_range.reshape(-1, 1), x_1_range.reshape(-1, 1)]))
 

test/test_genome.py

@@ -404,6 +404,31 @@ def test_mutate_output_region(rng_seed):
         ID_NON_CODING_GENE,
     ]
 
-    assert genome._mutate_output_region(list(dna), 9, rng) is False
-    assert genome._mutate_output_region(list(dna), 10, rng) is True
-    assert genome._mutate_output_region(list(dna), 11, rng) is False
+    assert genome._mutate_output_region(list(dna), 9, rng) is True
+    assert genome._mutate_output_region(list(dna), 10, rng) is False
+    assert genome._mutate_output_region(list(dna), 11, rng) is True
+
+
+@pytest.mark.parametrize("mutation_rate", [0.02, 0.05, 0.2])
+def test_correct_number_of_mutations(mutation_rate, rng_seed):
+
+    n_inputs = 2
+    n_outputs = 1
+    n_columns = 10
+    n_rows = 2
+    levels_back = 5
+    primitives = (cgp.Add, cgp.Sub, cgp.Mul, cgp.Div, cgp.ConstantFloat)
+
+    rng = np.random.RandomState(rng_seed)
+    genome = cgp.Genome(n_inputs, n_outputs, n_columns, n_rows, levels_back, primitives)
+    genome.randomize(rng)
+
+    n_mutations = 0
+    genome_new = genome.clone()
+    genome_new.mutate(mutation_rate, rng)
+    for (gene_0, gene_1) in zip(genome.dna, genome_new.dna):
+        if gene_0 != gene_1:
+            n_mutations += 1
+
+    n_mutations_expected = int(mutation_rate * len(genome.dna))
+    assert n_mutations == n_mutations_expected

test/test_population.py

@@ -57,7 +57,7 @@ def test_crossover_two_offspring(population_simple_fitness):
 
 
 def test_mutate(population_params, genome_params):
-    population_params["mutation_rate"] = 0.999999
+    population_params["mutation_rate"] = 0.5
     pop = cgp.Population(**population_params, genome_params=genome_params)
 
     offspring = pop.parents