dev: StdSOMonitor: get_min_fitness -> get_best_fitness

examples/cmaes_classic_control.py

@@ -66,11 +66,11 @@ def __call__(self, x):
 state = workflow.init(workflow_key)
 # run the workflow for 100 steps
 for i in range(100):
-    print(monitor.get_min_fitness())
+    print(monitor.get_best_fitness())
     state = workflow.step(state)
 
 sample_pop, state = workflow.sample(state)
 # problem._render(state.get_child_state("problem"), adapter.to_tree(sample_pop[0]))
 
-min_fitness = monitor.get_min_fitness()
+min_fitness = monitor.get_best_fitness()
 print(min_fitness)

examples/pgpe_pong.py

@@ -70,11 +70,11 @@ def __call__(self, img):
 state = workflow.init(workflow_key)
 # run the workflow for 100 steps
 for i in range(10):
-    print(monitor.get_min_fitness())
+    print(monitor.get_best_fitness())
     state = workflow.step(state)
 
 sample_pop, state = workflow.sample(state)
 # problem._render(adapter.to_tree(sample_pop[0]), ale_render_mode="human")
 # the result should be close to 0
-min_fitness = monitor.get_min_fitness()
+min_fitness = monitor.get_best_fitness()
 print(min_fitness)

src/evox/monitors/std_so_monitor.py

@@ -74,7 +74,7 @@ def get_topk_fitness(self):
     def get_topk_solutions(self):
         return self.topk_solutions
 
-    def get_min_fitness(self):
+    def get_best_fitness(self):
         if self.topk_fitness is None:
             warnings.warn("trying to get info from a monitor with no recorded data")
             return None

tests/test_containers.py

@@ -29,7 +29,7 @@ def test_clustered_cma_es():
     for i in range(200):
         state = workflow.step(state)
 
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     assert min_fitness < 2
 
 
@@ -91,7 +91,7 @@ def test_vectorized_coevolution(random_subpop):
     for i in range(200):
         state = workflow.step(state)
 
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     assert min_fitness < 1
 
 
@@ -124,7 +124,7 @@ def test_coevolution(random_subpop, num_subpop_iter):
     for i in range(4 * 200):
         state = workflow.step(state)
 
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     assert min_fitness < 2
 
 
@@ -156,6 +156,6 @@ def test_random_mask_cso():
     for i in range(10):
         state = workflow.step(state)
 
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     print(min_fitness)
     assert abs(min_fitness - 19.6) < 0.1

tests/test_distributed_pipeline.py

@@ -29,7 +29,7 @@ def test_distributed_cso():
         state = workflow.step(state)
 
     # the result should be close to 0
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     print(min_fitness)
     assert min_fitness < 1e-4
     workflow.health_check(state)

tests/test_gym.py

@@ -61,6 +61,6 @@ def __call__(self, x):
 
     monitor.close()
     # the result should be close to 0
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     # gym is deterministic, so the result should always be the same
     assert min_fitness == -16.0

tests/test_monitors.py

@@ -12,7 +12,7 @@ def test_std_so_monitor_top1():
     fitness1 = jnp.arange(3)
     monitor.record_pop(pop1)
     monitor.record_fit(fitness1)
-    assert monitor.get_min_fitness() == 0
+    assert monitor.get_best_fitness() == 0
     assert monitor.get_topk_fitness() == 0
     assert (monitor.get_best_solution() == pop1[0]).all()
     assert (monitor.get_topk_solutions() == pop1[0:1]).all()
@@ -21,7 +21,7 @@ def test_std_so_monitor_top1():
     fitness2 = -jnp.arange(3)
     monitor.record_pop(pop2)
     monitor.record_fit(fitness2)
-    assert monitor.get_min_fitness() == -2
+    assert monitor.get_best_fitness() == -2
     assert monitor.get_topk_fitness() == -2
     assert (monitor.get_best_solution() == pop2[2]).all()
     assert (monitor.get_topk_solutions() == pop2[2:3]).all()
@@ -34,7 +34,7 @@ def test_std_so_monitor_top2():
     fitness1 = jnp.arange(3)
     monitor.record_pop(pop1)
     monitor.record_fit(fitness1)
-    assert monitor.get_min_fitness() == 0
+    assert monitor.get_best_fitness() == 0
     assert (monitor.get_topk_fitness() == jnp.array([0, 1])).all()
     assert (monitor.get_best_solution() == pop1[0]).all()
     assert (monitor.get_topk_solutions() == pop1[0:2]).all()
@@ -43,7 +43,7 @@ def test_std_so_monitor_top2():
     fitness2 = -jnp.arange(3)
     monitor.record_pop(pop2)
     monitor.record_fit(fitness2)
-    assert monitor.get_min_fitness() == -2
+    assert monitor.get_best_fitness() == -2
     assert (monitor.get_topk_fitness() == jnp.array([-2, -1])).all()
     assert (monitor.get_best_solution() == pop2[2]).all()
     assert (monitor.get_topk_solutions() == pop2[2:0:-1]).all()

tests/test_neuroevolution.py

@@ -77,7 +77,7 @@ def test_neuroevolution_treemap():
         state = workflow.step(state)
 
     # the result should be close to 0
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     print(f"Treemap loss: {min_fitness}  time: {time.perf_counter() - start}")
 
 
@@ -111,5 +111,5 @@ def test_neuroevolution_adapter():
         state = workflow.step(state)
 
     # the result should be close to 0
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     print(f"Adapter loss: {min_fitness}  time: {time.perf_counter() - start}")

tests/test_single_objective_algorithms.py

@@ -29,7 +29,7 @@ def run_single_objective_algorithm(
     for i in range(num_iter):
         state = workflow.step(state)
 
-    return monitor.get_min_fitness()
+    return monitor.get_best_fitness()
 
 
 def test_cso():

tests/test_uni_workflow.py

@@ -26,7 +26,7 @@ def run_uni_workflow_with_jit_problem():
         state = workflow.step(state)
 
     monitor.close()
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     return min_fitness
 
 
@@ -52,7 +52,7 @@ def run_uni_workflow_with_non_jit_problem():
         state = workflow.step(state)
 
     monitor.close()
-    min_fitness = monitor.get_min_fitness()
+    min_fitness = monitor.get_best_fitness()
     return min_fitness