Drop Python 3.6 and 3.7 support, made minor internal modernizations.

Author: CodeReclaimers

CHANGELOG.md

@@ -41,6 +41,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - Includes test/visualization script for trained controllers
   - Full documentation in example README with usage tips
 
+### Changed
+- Dropped support for Python 3.6 and 3.7; neat-python now requires Python 3.8 or newer.
+- Modernized internal implementation in `neat/` and `examples/` to use Python 3 features
+  such as f-strings, comprehensions, dataclasses for internal helpers, and type hints,
+  without changing the public API.
+
 ### Fixed
 - **Population Size Drift**: Fixed small mismatches between actual population size and configured `pop_size`
   - `DefaultReproduction.reproduce()` now strictly enforces `len(population) == config.pop_size` for every non-extinction generation

examples/inverted-double-pendulum/evolve-feedforward.py

@@ -84,7 +84,7 @@ def run(config_file):
     with open('winner-feedforward.pickle', 'wb') as f:
         pickle.dump(winner, f)
 
-    print('\n\nBest genome:\n{!s}'.format(winner))
+    print(f'\n\nBest genome:\n{winner!s}')
 
     # Visualize the results
     visualize.plot_stats(stats, ylog=False, view=True, filename="feedforward-fitness.svg")

examples/inverted-double-pendulum/visualize.py

@@ -58,7 +58,7 @@ def plot_spikes(spikes, view=False, filename=None, title=None):
     if title is None:
         plt.title("Izhikevich's spiking neuron model")
     else:
-        plt.title("Izhikevich's spiking neuron model ({0!s})".format(title))
+        plt.title(f"Izhikevich's spiking neuron model ({title!s})")
 
     plt.subplot(3, 1, 2)
     plt.ylabel("Recovery (u)")

examples/memory-fixed/evolve.py

@@ -98,28 +98,28 @@ def run():
     stats.save()
 
     # Show output of the most fit genome against a random input.
-    print('\nBest genome:\n{!s}'.format(winner))
+    print(f'\nBest genome:\n{winner!s}')
     print('\nOutput:')
     winner_net = neat.nn.RecurrentNetwork.create(winner, config)
     num_correct = 0
     for n in range(num_tests):
-        print('\nRun {0} output:'.format(n))
+        print(f'\nRun {n} output:')
         seq = [random.choice((0.0, 1.0)) for _ in range(N)]
         winner_net.reset()
         for s in seq:
             inputs = [s, 0.0]
             winner_net.activate(inputs)
-            print('\tseq {0}'.format(inputs))
+            print(f'\tseq {inputs}')
 
         correct = True
         for s in seq:
             output = winner_net.activate([0, 1])
-            print("\texpected {0:1.5f} got {1:1.5f}".format(s, output[0]))
+            print(f"\texpected {s:1.5f} got {output[0]:1.5f}")
             correct = correct and round(output[0]) == s
         print("OK" if correct else "FAIL")
         num_correct += 1 if correct else 0
 
-    print("{0} of {1} correct {2:.2f}%".format(num_correct, num_tests, 100.0 * num_correct / num_tests))
+    print(f"{num_correct} of {num_tests} correct {100.0 * num_correct / num_tests:.2f}%")
 
     node_names = {-1: 'input', -2: 'gate', 0: 'output'}
     visualize.draw_net(config, winner, True, node_names=node_names)

examples/memory-fixed/visualize.py

@@ -57,7 +57,7 @@ def plot_spikes(spikes, view=False, filename=None, title=None):
     if title is None:
         plt.title("Izhikevich's spiking neuron model")
     else:
-        plt.title("Izhikevich's spiking neuron model ({0!s})".format(title))
+        plt.title(f"Izhikevich's spiking neuron model ({title!s})")
 
     plt.subplot(3, 1, 2)
     plt.ylabel("Recovery (u)")

examples/memory-variable/evolve.py

@@ -85,12 +85,12 @@ def run():
     stats.save()
 
     # Show output of the most fit genome against a random input.
-    print('\nBest genome:\n{!s}'.format(winner))
+    print(f'\nBest genome:\n{winner!s}')
     print('\nOutput:')
     winner_net = neat.nn.RecurrentNetwork.create(winner, config)
     num_correct = 0
     for n in range(num_tests):
-        print('\nRun {0} output:'.format(n))
+        print(f'\nRun {n} output:')
 
         num_inputs = random.randint(1, max_inputs)
         num_ignore = random.randint(0, max_ignore)
@@ -101,12 +101,12 @@ def run():
 
         correct = True
         for i, o in zip(seq, outputs):
-            print("\texpected {0:1.5f} got {1:1.5f}".format(i, o[0]))
+            print(f"\texpected {i:1.5f} got {o[0]:1.5f}")
             correct = correct and round(o[0]) == i
         print("OK" if correct else "FAIL")
         num_correct += 1 if correct else 0
 
-    print("{0} of {1} correct {2:.2f}%".format(num_correct, num_tests, 100.0 * num_correct / num_tests))
+    print(f"{num_correct} of {num_tests} correct {100.0 * num_correct / num_tests:.2f}%")
 
     node_names = {-1: 'input', -2: 'record', -3: 'play', 0: 'output'}
     visualize.draw_net(config, winner, True, node_names=node_names)

examples/memory-variable/visualize.py

@@ -58,7 +58,7 @@ def plot_spikes(spikes, view=False, filename=None, title=None):
     if title is None:
         plt.title("Izhikevich's spiking neuron model")
     else:
-        plt.title("Izhikevich's spiking neuron model ({0!s})".format(title))
+        plt.title(f"Izhikevich's spiking neuron model ({title!s})")
 
     plt.subplot(3, 1, 2)
     plt.ylabel("Recovery (u)")

examples/neuron-demo/demo-ctrnn.py

@@ -25,7 +25,7 @@
     output = net.advance([], 0.002, 0.002)
     times.append(net.time_seconds)
     outputs.append(output)
-    print("{0:.7f} {1:.7f}".format(output[0], output[1]))
+    print(f"{output[0]:.7f} {output[1]:.7f}")
 
 outputs = np.array(outputs).T
 

examples/neuron-demo/demo-iznn.py

@@ -18,7 +18,7 @@ def plot_spikes(spikes, title):
     plt.grid()
     plt.plot(t_values, v_values, "g-")
 
-    plt.title("Izhikevich's spiking neuron model ({0!s})".format(title))
+    plt.title(f"Izhikevich's spiking neuron model ({title!s})")
 
     plt.subplot(4, 1, 2)
     plt.ylabel("Fired")
@@ -39,7 +39,7 @@ def plot_spikes(spikes, title):
     plt.plot(t_values, I_values, "r-o")
 
     fig = plt.figure()
-    plt.title("Izhikevich's spiking neuron model u/v ({0!s})".format(title))
+    plt.title(f"Izhikevich's spiking neuron model u/v ({title!s})")
     plt.xlabel("Recovery (u)")
     plt.ylabel("Potential (mv)")
     plt.grid()
@@ -55,7 +55,7 @@ def show(title, a, b, c, d):
     for i in range(1000):
         n.current = 0.0 if i < 100 or i > 800 else 10.0
         spike_train.append((1.0 * i, n.current, n.v, n.u, n.fired))
-        print('{0:d}\t{1:f}\t{2:f}\t{3:f}'.format(i, n.current, n.v, n.u))
+        print(f'{i:d}\t{n.current:f}\t{n.v:f}\t{n.u:f}')
         n.advance(0.25)
 
     plot_spikes(spike_train, title)

examples/openai-lander/evolve.py

@@ -70,7 +70,7 @@ def compute_fitness(genome, net, episodes, min_reward, max_reward):
     return reward_error
 
 
-class PooledErrorCompute(object):
+class PooledErrorCompute:
     def __init__(self, num_workers):
         self.num_workers = num_workers
         self.test_episodes = []
@@ -110,7 +110,7 @@ def simulate(self, nets):
 
             self.test_episodes.append((score, data))
 
-        print("Score range [{:.3f}, {:.3f}]".format(min(scores), max(scores)))
+        print(f"Score range [{min(scores):.3f}, {max(scores):.3f}]")
 
     def evaluate_genomes(self, genomes, config):
         self.generation += 1
@@ -120,19 +120,19 @@ def evaluate_genomes(self, genomes, config):
         for gid, g in genomes:
             nets.append((g, neat.nn.FeedForwardNetwork.create(g, config)))
 
-        print("network creation time {0}".format(time.time() - t0))
+        print(f"network creation time {time.time() - t0}")
         t0 = time.time()
 
         # Periodically generate a new set of episodes for comparison.
         if 1 == self.generation % 10:
             self.test_episodes = self.test_episodes[-300:]
             self.simulate(nets)
-            print("simulation run time {0}".format(time.time() - t0))
+            print(f"simulation run time {time.time() - t0}")
             t0 = time.time()
 
         # Assign a composite fitness to each genome; genomes can make progress either
         # by improving their total reward or by making more accurate reward estimates.
-        print("Evaluating {0} test episodes".format(len(self.test_episodes)))
+        print(f"Evaluating {len(self.test_episodes)} test episodes")
         if self.num_workers < 2:
             for genome, net in nets:
                 reward_error = compute_fitness(genome, net, self.test_episodes, self.min_reward, self.max_reward)
@@ -149,7 +149,7 @@ def evaluate_genomes(self, genomes, config):
                     reward_error = job.get(timeout=None)
                     genome.fitness = -np.sum(reward_error) / len(self.test_episodes)
 
-        print("final fitness compute time {0}\n".format(time.time() - t0))
+        print(f"final fitness compute time {time.time() - t0}\n")
 
 
 def run():
@@ -187,10 +187,10 @@ def run():
             plt.close()
 
             mfs = sum(stats.get_fitness_mean()[-5:]) / 5.0
-            print("Average mean fitness over last 5 generations: {0}".format(mfs))
+            print(f"Average mean fitness over last 5 generations: {mfs}")
 
             mfs = sum(stats.get_fitness_stat(min)[-5:]) / 5.0
-            print("Average min fitness over last 5 generations: {0}".format(mfs))
+            print(f"Average min fitness over last 5 generations: {mfs}")
 
             # Use the best genomes seen so far as an ensemble-ish control system.
             best_genomes = stats.best_unique_genomes(3)
@@ -245,7 +245,7 @@ def run():
             # Save the winners.
             if best_genomes:
                 for n, g in enumerate(best_genomes):
-                    name = 'winner-{0}'.format(n)
+                    name = f'winner-{n}'
                     with open(name + '.pickle', 'wb') as f:
                         pickle.dump(g, f)