Added Pareto set drawing (#200)

Author: LebedevIlyaG

examples/MCO_Grishagin_example.py

@@ -1,3 +1,4 @@
+from iOpt.output_system.listeners.static_painters import StaticPainterParetoListener
 from problems.grishagin_mco import Grishagin_mco
 from iOpt.solver import Solver
 from iOpt.solver_parametrs import SolverParameters
@@ -18,6 +19,9 @@
     cfol = ConsoleOutputListener(mode='full')
     solver.add_listener(cfol)
 
+    sppl = StaticPainterParetoListener("Grishagin_mco_pareto.png")
+    solver.add_listener(sppl)
+
     sol = solver.solve()
 
     # output of the Pareto set (coordinates - function values)

examples/MCO_Test1_example.py

@@ -3,6 +3,7 @@
 from iOpt.solver_parametrs import SolverParameters
 
 from iOpt.output_system.listeners.console_outputers import ConsoleOutputListener
+from iOpt.output_system.listeners.static_painters import StaticPainterParetoListener
 import matplotlib.pyplot as plt
 
 if __name__ == "__main__":
@@ -24,6 +25,10 @@
     cfol = ConsoleOutputListener(mode='full')
     solver.add_listener(cfol)
 
+    # Добавляем построение графика множества Парето y[0]-y[1]
+    sppl = StaticPainterParetoListener("mco_test1_pareto.png")
+    solver.add_listener(sppl)
+
     # Решаем задачу
     sol = solver.solve()
 
@@ -33,17 +38,3 @@
     print("size pareto set: ", len(var))
     for fvar, fval in zip(var, val):
         print(fvar, fval)
-
-    # Точки для построения
-    # x1 = [trial.point.float_variables[0] for trial in sol.best_trials]
-    # x2 = [trial.point.float_variables[1] for trial in sol.best_trials]
-
-    # plt.plot(x1, x2, 'ro')
-    # plt.show()
-
-    # Точки для построения графика множества Парето y[0]-y[1]
-    fv1 = [trial.function_values[0].value for trial in sol.best_trials]
-    fv2 = [trial.function_values[1].value for trial in sol.best_trials]
-
-    plt.plot(fv1, fv2, 'ro')
-    plt.show()

examples/Machine_learning/SVC/_2D/Example_SVC_2D_MCO_Transformators_State.py

@@ -1,4 +1,4 @@
-from iOpt.output_system.listeners.static_painters import StaticPainterNDListener
+from iOpt.output_system.listeners.static_painters import StaticPainterParetoListener
 from iOpt.output_system.listeners.console_outputers import ConsoleOutputListener
 
 from iOpt.solver import Solver
@@ -12,7 +12,7 @@
 # Для построения HV индекса
 # from pymoo.util.misc import stack
 # from pymoo.indicators.hv import HV
-import numpy as np
+# import numpy as np
 
 def factory_dataset():
     x = []
@@ -34,13 +34,16 @@ def factory_dataset():
     kernel_coefficient_bound = {'low': -3, 'up': 1}
     problem = MCO_SVC_2D_Transformators_State.MCO_SVC_2D_Transformators_State(X, Y, regularization_value_bound,
                                                                       kernel_coefficient_bound)
-    method_params = SolverParameters(r=np.double(2.0), iters_limit=50, number_of_parallel_points=1,
+    method_params = SolverParameters(r=np.double(2.0), iters_limit=500, number_of_parallel_points=1,
                                      evolvent_density=12, number_of_lambdas=5)
     solver = Solver(problem=problem, parameters=method_params)
     # Добавляем вывод результатов в консоль
     cfol = ConsoleOutputListener(mode='full')
     solver.add_listener(cfol)
 
+    sppl = StaticPainterParetoListener("MCO_SVC_2D_Transformators_pareto.png")
+    solver.add_listener(sppl)
+
     # Решаем задачу
     sol = solver.solve()
 

examples/Machine_learning/SVC/_2D/Problems/MCO_SVC_2D_Transformators_State.py

@@ -42,11 +42,6 @@ def __init__(self, x_dataset: np.ndarray, y_dataset: np.ndarray,
         self.upper_bound_of_float_variables = np.array([regularization_bound['up'], kernel_coefficient_bound['up']],
                                                    dtype=np.double)
 
-        #self.cv = StratifiedKFold(shuffle=True, random_state=42)
-
-
-
-
     def calculateAllFunction(self, point: Point, function_values: np.ndarray(shape=(1), dtype=FunctionValue)) -> \
             np.ndarray(shape=(1), dtype=FunctionValue):
         """

iOpt/output_system/listeners/static_painters.py

@@ -1,7 +1,7 @@
 from iOpt.method.listener import Listener
 from iOpt.method.method import Method
 from iOpt.method.search_data import SearchData
-from iOpt.output_system.painters.static_painters import StaticPainter, StaticPainterND, DiscretePainter
+from iOpt.output_system.painters.static_painters import StaticPainter, StaticPainterND, DiscretePainter, StaticPainterPareto
 from iOpt.solution import Solution
 
 
@@ -89,7 +89,7 @@ def __init__(self, file_name: str, path_for_saves="", indx=0, is_points_at_botto
         """
         Constructor of the StaticPainterListener class
 
-        :param file_name: File name specifying the format for saving the image. 
+        :param file_name: File name specifying the format for saving the image.
         :param path_for_saves: The directory to save the image. If this parameter is not specified, the image is saved in the current working directory.
            is saved in the current working directory.
         :param indx: Index of the variable of the optimization problem. It is used in multivariate optimization.
@@ -133,9 +133,9 @@ class StaticPainterNDListener(Listener):
     def __init__(self, file_name: str, path_for_saves="", vars_indxs=[0, 1], mode='lines layers',
                  calc='objective function'):
         """
-        Конструктор класса StaticPainterNDListener
+        Constructor of the StaticPainterNDListener class
 
-        :param file_name: File name specifying the format for saving the image. 
+        :param file_name: File name specifying the format for saving the image.
         :param path_for_saves: The directory to save the image. If this parameter is not specified, the image is saved in the current working directory.
            is saved in the current working directory.
         :param vars_indxs: A pair of indices of the variables of the optimization problem for which the figure will be plotted.
@@ -167,3 +167,30 @@ def on_method_stop(self, search_data: SearchData,
         painter.paint_points()
         painter.paint_optimum()
         painter.save_image()
+
+
+class StaticPainterParetoListener(Listener):
+    """
+    The StaticPainterParetoListener class is an event listener. It contains a method handler that produces an image as a
+      image as a reaction to the method completion.
+      It is used for multicriteria optimization
+    """
+
+    def __init__(self, file_name: str, path_for_saves="", criteria_indxs=[0, 1]):
+        """
+        Constructor of the StaticPainterParetoListener class
+
+        :param file_name: File name specifying the format for saving the image.
+        :param path_for_saves: The directory to save the image. If this parameter is not specified, the image is saved in the current working directory.
+           is saved in the current working directory.
+        :param criteria_indxs: A pair of indices of the criteria of the optimization problem for which the figure will be plotted.
+        """
+        self.file_name = file_name
+        self.path_for_saves = path_for_saves
+        self.criteria_indxs = criteria_indxs
+
+    def on_method_stop(self, search_data: SearchData,
+                       solution: Solution, status: bool, ):
+        painter = StaticPainterPareto(solution, self.criteria_indxs, self.path_for_saves, self.file_name)
+        painter.paint_pareto()
+        painter.save_image()

iOpt/output_system/painters/plotters/plotters.py

@@ -243,7 +243,8 @@ def __init__(self, parameter_in_nd_problem, left_bound, right_bound):
 
         self.fig, self.ax = plt.subplots(1, 1)
         self.ax.tick_params(axis='both', labelsize=8)
-        self.ax.set_xlim([self.leftBound, self.rightBound])
+        if self.leftBound != None and self.rightBound != None:
+            self.ax.set_xlim([self.leftBound, self.rightBound])
 
     def plot_by_grid(self, calculate, section, points_count=100):
         x = np.arange(self.leftBound, self.rightBound, (self.rightBound - self.leftBound) / points_count)
@@ -433,4 +434,12 @@ def plot_points(self, points, values, clr='blue', mrkr='o', mrkrs=4):
         self.ax.relim()
         self.ax.autoscale_view()
         self.fig.canvas.draw()
-        self.fig.canvas.flush_events()
+        self.fig.canvas.flush_events()
+
+
+class PlotterPareto(Plotter2D):
+    def __init__(self):
+        super().__init__(None, None, None)
+
+    def plot_pareto(self, first_criteria_values, second_criteria_values, clr='blue', mrkr='o', mrkrs=8):
+        self.plot_points(first_criteria_values, second_criteria_values, clr, mrkr, mrkrs)

iOpt/output_system/painters/static_painters.py

@@ -3,7 +3,7 @@
 from iOpt.method.search_data import SearchData, SearchDataItem
 from iOpt.trial import Point, FunctionValue
 from iOpt.solution import Solution
-from iOpt.output_system.painters.plotters.plotters import Plotter2D, Plotter3D, DisretePlotter
+from iOpt.output_system.painters.plotters.plotters import Plotter2D, Plotter3D, DisretePlotter, PlotterPareto
 from iOpt.output_system.painters.painter import Painter
 
 import matplotlib.pyplot as plt
@@ -282,4 +282,36 @@ def calculate_func(self, x):
         point = Point(x, [])
         fv = FunctionValue()
         fv = self.objFunc(point, fv)
-        return fv.value
+        return fv.value
+
+
+class StaticPainterPareto:
+    def __init__(self,
+                 solution: Solution,
+                 criteria_indxs,
+                 path_for_saves,
+                 file_name
+                 ):
+        self.path_for_saves = path_for_saves
+        self.file_name = file_name
+
+        # values of Pareto-efficient criteria with input indices
+        self.first_criteria_values = [trial.function_values[criteria_indxs[0]].value for trial in solution.best_trials]
+        self.second_criteria_values = [trial.function_values[criteria_indxs[1]].value for trial in solution.best_trials]
+
+        # definition of plotter
+        self.plotter = PlotterPareto()
+
+    def paint_pareto(self):
+        self.plotter.plot_pareto(self.first_criteria_values, self.second_criteria_values)
+
+    def save_image(self):
+        if not os.path.isdir(self.path_for_saves):
+            if self.path_for_saves == "":
+                plt.savefig(self.file_name)
+            else:
+                os.mkdir(self.path_for_saves)
+                plt.savefig(self.path_for_saves + "/" + self.file_name)
+        else:
+            plt.savefig(self.path_for_saves + "/" + self.file_name)
+        plt.show()