continue refactoring

Author: SimonBlanke

gradient_free_optimizers/optimizers/core_optimizer/core_optimizer.py

@@ -21,6 +21,8 @@ def __init__(
     ):
         super().__init__()
 
+        self.random_seed = set_random_seed(nth_process, random_state)
+
         self.conv = Converter(search_space)
         self.init = Initializer(self.conv, initialize)
 
@@ -30,8 +32,6 @@ def __init__(
         self.nth_process = nth_process
         self.debug_log = debug_log
 
-        self.random_seed = set_random_seed(nth_process, random_state)
-
     def random_iteration(func):
         def wrapper(self, *args, **kwargs):
             if self.rand_rest_p > random.uniform(0, 1):
@@ -62,10 +62,7 @@ def move_random(self):
 
     @SearchTracker.track_new_pos
     def init_pos(self):
-        print("\n 11111 self.init.init_positions_l", self.init.init_positions_l)
-        print(" self.nth_trial", self.nth_trial)
-
-        init_pos = self.init.init_positions_l[self.nth_trial]
+        init_pos = self.init.init_positions_l[self.n_init_total]
         return init_pos
 
     def finish_initialization(self):

gradient_free_optimizers/optimizers/pop_opt/base_population_optimizer.py

@@ -35,7 +35,12 @@ def sort_pop_best_score(self):
         self.pop_sorted = [self.optimizers[i] for i in idx_sorted_ind]
 
     def _create_population(self, Optimizer):
-        diff_init = self.population - self.init.n_inits
+        if isinstance(self.population, int):
+            pop_size = self.population
+        else:
+            pop_size = len(self.population)
+        diff_init = pop_size - self.init.n_inits
+
         if diff_init > 0:
             self.init.add_n_random_init_pos(diff_init)
 
@@ -55,8 +60,6 @@ def _create_population(self, Optimizer):
         else:
             population = self.population
 
-        print("\n init_positions_l \n", self.init.init_positions_l, "\n")
-
         return population
 
     @CoreOptimizer.track_new_score

gradient_free_optimizers/results_manager.py

@@ -6,9 +6,9 @@
 
 
 class ResultsManager:
-    def __init__(self, conv):
+    def __init__(self):
         super().__init__()
-        self.conv = conv
+        self.conv = None
 
         self.results_list = []
 
@@ -33,14 +33,11 @@ def _wrapper(pos):
             results_dict = self._obj_func_results(objective_function, para)
 
             self.results_list.append({**results_dict, **para})
-            print("score self.results_list", self.results_list)
 
             return results_dict["score"]
 
         return _wrapper
 
     @property
     def search_data(self):
-        print("self.results_list", self.results_list)
-
         return pd.DataFrame(self.results_list)

gradient_free_optimizers/search.py

@@ -8,14 +8,15 @@
 
 from .progress_bar import ProgressBarLVL0, ProgressBarLVL1
 from .times_tracker import TimesTracker
+from .search_statistics import SearchStatistics
 from .memory import Memory
 from .print_info import print_info
 from .stop_run import StopRun
 
 from .results_manager import ResultsManager
 
 
-class Search(TimesTracker):
+class Search(TimesTracker, SearchStatistics):
     def __init__(self):
         super().__init__()
 
@@ -25,12 +26,11 @@ def __init__(self):
 
         self.score_l = []
         self.pos_l = []
-        self.nth_iter = 0
         self.random_seed = None
 
         self.search_state = "init"
-        self.n_init_total = 0
-        self.n_iter_total = 0
+
+        self.results_mang = ResultsManager()
 
     @TimesTracker.eval_time
     def _score(self, pos):
@@ -42,11 +42,9 @@ def _initialization(self, nth_iter):
         self.best_score = self.p_bar.score_best
 
         init_pos = self.init_pos()
-        print("\n init_pos ", init_pos)
 
         score_new = self._score(init_pos)
         self.evaluate_init(score_new)
-        print("\n score_new ", score_new)
 
         self.pos_l.append(init_pos)
         self.score_l.append(score_new)
@@ -84,6 +82,7 @@ def _init_search(self):
                 self.nth_process, self.n_iter, self.objective_function
             )
 
+    @SearchStatistics.init_stats
     def search(
         self,
         objective_function,
@@ -97,10 +96,7 @@ def search(
     ):
         self.start_time = time.time()
 
-        self.results_mang = ResultsManager(self.conv)
-
-        self.n_init_search = 0
-        self.n_iter_search = 0
+        self.results_mang.conv = self.conv
 
         if verbosity is False:
             verbosity = []
@@ -127,34 +123,19 @@ def search(
         else:
             self.score = self.results_mang.score(objective_function)
 
-        print(
-            "\n search init_positions_l \n",
-            self.init.init_positions_l,
-            "\n",
-        )
-
-        n_inits_norm = min(self.init.n_inits, n_iter)
-        print("\n n_inits_norm", n_inits_norm)
+        n_inits_norm = min((self.init.n_inits - self.n_init_total), n_iter)
 
         # if self.search_state == "init":
         # loop to initialize N positions
         for nth_iter in range(n_inits_norm):
-            print("\n init!")
             if self.stop.check(self.start_time, self.p_bar.score_best, self.score_l):
                 break
             self._initialization(nth_iter)
 
-            print("pos_new_list", self.pos_new_list)
-
         self.finish_initialization()
 
-        print("\n self.n_init_search", self.n_init_search)
-        print("\n self.n_init_total", self.n_init_total)
-
         # loop to do the iterations
         for nth_iter in range(self.n_init_search, n_iter):
-            print("\n iter!")
-
             if self.stop.check(self.start_time, self.p_bar.score_best, self.score_l):
                 break
             self._iteration(nth_iter)

gradient_free_optimizers/search_statistics.py

@@ -0,0 +1,23 @@
+# Author: Simon Blanke
+# Email: simon.blanke@yahoo.com
+# License: MIT License
+
+
+class SearchStatistics:
+    def __init__(self):
+        super().__init__()
+
+        self.nth_iter = 0
+
+        self.n_init_total = 0
+        self.n_iter_total = 0
+
+    def init_stats(func):
+        def wrapper(self, *args, **kwargs):
+            self.n_init_search = 0
+            self.n_iter_search = 0
+
+            res = func(self, *args, **kwargs)
+            return res
+
+        return wrapper