Parcourir la source

Update of A tour of Macop

Jérôme BUISINE il y a 3 ans
Parent
commit
476b4fed26

+ 2 - 2
docs/source/api.rst

@@ -134,8 +134,8 @@ macop.evaluators
 .. autosummary::
    :toctree: macop
    
-   macop.evaluators.discrete.mono
-   macop.evaluators.discrete.multi
+   macop.evaluators.continuous.mono
+   macop.evaluators.continuous.multi
 
 macop.operators
 -------------------

+ 32 - 2
docs/source/documentations.rst

@@ -1363,9 +1363,12 @@ If we want to exploit this functionality, then we will need to exploit them with
 All the features of **Macop** were presented. The next section will aim to quickly present the few implementations proposed within **Macop** to highlight the modulality of the package.
 
 
-Implementation examples
+Advanced usages
 =======================
 
+Multi-objective discrete optimisation
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
 Within the API of **Macop**, you can find an implementation of The Multi-objective evolutionary algorithm based on decomposition (MOEA/D) is a general-purpose algorithm for approximating the Pareto set of multi-objective optimization problems. 
 It decomposes the original multi-objective problem into a number of single-objective optimization sub-problems and then uses an evolutionary process to optimize these sub-problems simultaneously and cooperatively. 
 MOEA/D is a state-of-art algorithm in aggregation-based approaches for multi-objective optimization.
@@ -1384,7 +1387,22 @@ An example with MOEAD for knapsack problem is available in knapsackMultiExample.
 
 .. _knapsackMultiExample.py: https://github.com/jbuisine/macop/blob/master/examples/knapsackMultiExample.py
 
+Continuous Zdt problems
+~~~~~~~~~~~~~~~~~~~~~~~
+
+Even if the package is not primarily intended for continuous optimisation, it allows for adaptation to continuous optimisation. 
+
+Based on the Zdt_ benchmarks function, it offers an implementation of Solution, Operator and Evaluator to enable the optimisation of this kind of problem.
+
+.. _Zdt: https://en.wikipedia.org/wiki/Test_functions_for_optimization
+
+- macop.solutions.continuous.ContinuousSolution_: manage float array solution in order to represent continuous solution;
+- macop.operators.continuous.mutators.PolynomialMutation_: update solution using polynomial mutation over solution's data;
+- macop.operators.continuous.crossovers.BasicDifferentialEvolutionCrossover_: use of new generated solutions in order to obtain new offspring solution;
+- macop.evaluators.continous.mono.ZdtEvaluator_: continuous evaluator for `Zdt` problem instance. Take into its ``data``, the ``f`` Zdt function;
+- macop.callbacks.classicals.ContinuousCallback_: manage callback and backup of continuous solution.
 
+A complete implementation example with the Rosenbrock_ function is available.
 
 .. _macop.algorithms.base: macop/macop.algorithms.base.html#module-macop.algorithms.base
 .. _macop.algorithms.mono: macop/macop.algorithms.mono.html#module-macop.algorithms.mono
@@ -1406,4 +1424,16 @@ An example with MOEAD for knapsack problem is available in knapsackMultiExample.
 .. _macop.callbacks.base.Callback: macop/macop.callbacks.base.html#macop.callbacks.base.Callback
 
 .. _macop.algorithms.multi.MOSubProblem: macop/macop.algorithms.multi.html#macop.algorithms.multi.MOSubProblem
-.. _macop.algorithms.multi.MOEAD: macop/macop.algorithms.multi.html#macop.algorithms.multi.MOEAD
+.. _macop.algorithms.multi.MOEAD: macop/macop.algorithms.multi.html#macop.algorithms.multi.MOEAD
+
+.. _macop.solutions.continuous.ContinuousSolution: macop/macop.solutions.continuous.html#macop.solutions.continuous.ContinuousSolution
+
+.. _macop.operators.continuous.mutators.PolynomialMutation: macop/macop.operators.continuous.mutators.html#macop.operators.continuous.mutators.PolynomialMutation
+.. _macop.operators.continuous.crossovers.BasicDifferentialEvolutionCrossover: macop/macop.operators.continuous.crossovers.html#macop.operators.continuous.crossovers.BasicDifferentialEvolutionCrossover
+
+.. _macop.evaluators.continous.mono.ZdtEvaluator: macop/macop.evaluators.continuous.mono.html#macop.evaluators.continous.mono.ZdtEvaluator
+
+.. _macop.callbacks.classicals.ContinuousCallback: macop/macop.callbacks.classicals.html#macop.callbacks.classicals.ContinuousCallback
+
+
+.. _Rosenbrock: https://github.com/jbuisine/macop/blob/master/examples/ZdtExample.py

+ 22 - 0
docs/source/macop/macop.evaluators.continuous.mono.rst

@@ -0,0 +1,22 @@
+macop.evaluators.continuous.mono
+================================
+
+.. automodule:: macop.evaluators.continuous.mono
+
+   
+   
+   
+
+   
+   
+   .. rubric:: Classes
+
+   .. autosummary::
+   
+      ZdtEvaluator
+   
+   
+
+   
+   
+   

+ 16 - 0
docs/source/macop/macop.evaluators.continuous.multi.rst

@@ -0,0 +1,16 @@
+macop.evaluators.continuous.multi
+=================================
+
+.. automodule:: macop.evaluators.continuous.multi
+
+   
+   
+   
+
+   
+   
+   
+
+   
+   
+   

+ 28 - 32
examples/ZdtExample.py

@@ -3,19 +3,20 @@ import logging
 import os
 import random
 import numpy as np
+import math
 
 # module imports
-from macop.solutions.discrete import BinarySolution
-from macop.evaluators.discrete.mono import UBQPEvaluator
+from macop.solutions.continuous import ContinuousSolution
+from macop.evaluators.continuous.mono import ZdtEvaluator
 
-from macop.operators.continuous.mutators import SimpleMutation
-from macop.operators.discrete.mutators import SimpleBinaryMutation
+from macop.operators.continuous.mutators import PolynomialMutation
+from macop.operators.continuous.crossovers import BasicDifferentialEvolutionCrossover
 
 from macop.policies.classicals import RandomPolicy
 
 from macop.algorithms.mono import IteratedLocalSearch as ILS
 from macop.algorithms.mono import HillClimberFirstImprovment
-from macop.callbacks.classicals import BasicCheckpoint
+from macop.callbacks.classicals import ContinuousCheckpoint
 
 if not os.path.exists('data'):
     os.makedirs('data')
@@ -26,52 +27,47 @@ logging.basicConfig(format='%(asctime)s %(message)s', filename='data/example.log
 random.seed(42)
 
 # usefull instance data
-n = 100
-ubqp_instance_file = 'instances/ubqp/ubqp_instance.txt'
-filepath = "data/checkpoints_ubqp.csv"
+n = 10
+filepath = "data/checkpoints_zdt_Rosenbrock.csv"
+problem_interval = -10, 10 # fixed value interval (avoid infinite)
 
 
-# default validator
+# check each value in order to validate
 def validator(solution):
+
+    mini, maxi = problem_interval
+
+    for x in solution.data:
+        if x < mini or x > maxi:
+            return False
+
     return True
 
 # define init random solution
 def init():
-    return BinarySolution.random(n, validator)
-
-
-filepath = "data/checkpoints.csv"
+    return ContinuousSolution.random(n, problem_interval, validator)
 
 def main():
 
-    # load UBQP instance
-    with open(ubqp_instance_file, 'r') as f:
-
-        lines = f.readlines()
-
-        # get all string floating point values of matrix
-        Q_data = ''.join([ line.replace('\n', '') for line in lines[8:] ])
-
-        # load the concatenate obtained string
-        Q_matrix = np.fromstring(Q_data, dtype=float, sep=' ').reshape(n, n)
-
-    print(f'Q_matrix shape: {Q_matrix.shape}')
+    # Rosenbrock function with a=1 and b=100 (see https://en.wikipedia.org/wiki/Rosenbrock_function)
+    Rosenbrock_function = lambda s: sum([ 100 * math.pow(s.data[i + 1] - (math.pow(s.data[i], 2)), 2) + math.pow((1 - s.data[i]), 2) for i in range(len(s.data) - 1) ])
 
-    operators = [SimpleBinaryMutation(), SimpleMutation()]
+    operators = [PolynomialMutation(interval=problem_interval), BasicDifferentialEvolutionCrossover(interval=problem_interval)]
     policy = RandomPolicy(operators)
-    callback = BasicCheckpoint(every=5, filepath=filepath)
-    evaluator = UBQPEvaluator(data={'Q': Q_matrix})
+    callback = ContinuousCheckpoint(every=5, filepath=filepath)
+    evaluator = ZdtEvaluator(data={'f': Rosenbrock_function})
 
     # passing global evaluation param from ILS
-    hcfi = HillClimberFirstImprovment(init, evaluator, operators, policy, validator, maximise=True, verbose=True)
-    algo = ILS(init, evaluator, operators, policy, validator, localSearch=hcfi, maximise=True, verbose=True)
+    hcfi = HillClimberFirstImprovment(init, evaluator, operators, policy, validator, maximise=False, verbose=True)
+    algo = ILS(init, evaluator, operators, policy, validator, localSearch=hcfi, maximise=False, verbose=True)
     
     # add callback into callback list
     algo.addCallback(callback)
 
-    bestSol = algo.run(10000, ls_evaluations=100)
+    bestSol = algo.run(100000, ls_evaluations=100)
+    print(bestSol.data)
 
-    print('Solution for UBQP instance score is {}'.format(evaluator.compute(bestSol)))
+    print('Solution for Rosenbrock Zdt instance score is {}'.format(evaluator.compute(bestSol)))
 
 if __name__ == "__main__":
     main()

+ 2 - 0
macop/__init__.py

@@ -9,6 +9,8 @@ from macop.callbacks import multi
 from macop.evaluators import base
 from macop.evaluators.discrete import mono
 from macop.evaluators.discrete import multi
+from macop.evaluators.continuous import mono
+from macop.evaluators.continuous import multi
 
 from macop.operators import base
 from macop.operators.discrete import mutators

+ 4 - 4
macop/algorithms/mono.py

@@ -343,16 +343,16 @@ class IteratedLocalSearch(Algorithm):
         # by default use of mother method to initialise variables
         super().run(evaluations)
 
+        # add same callbacks
+        for callback in self._callbacks:
+            self._localSearch.addCallback(callback)
+
         # enable resuming for ILS
         self.resume()
 
         # initialise current solution
         self.initRun()
 
-        # add same callbacks
-        for callback in self._callbacks:
-            self._localSearch.addCallback(callback)
-
         # local search algorithm implementation
         while not self.stop():
 

+ 99 - 7
macop/callbacks/classicals.py

@@ -34,16 +34,16 @@ class BasicCheckpoint(Callback):
 
             logging.info("Checkpoint is done into " + self._filepath)
 
-            solution.data = ""
+            solution_data = ""
             solutionSize = len(solution.data)
 
             for index, val in enumerate(solution.data):
-                solution.data += str(val)
+                solution_data += str(val)
 
                 if index < solutionSize - 1:
-                    solution.data += ' '
+                    solution_data += ' '
 
-            line = str(currentEvaluation) + ';' + solution.data + ';' + str(
+            line = str(currentEvaluation) + ';' + solution_data + ';' + str(
                 solution.fitness) + ';\n'
 
             # check if file exists
@@ -76,12 +76,104 @@ class BasicCheckpoint(Callback):
                     self._algo.setEvaluation(globalEvaluation)
 
                 # get best solution data information
-                solution.data = list(map(int, data[1].split(' ')))
+                solution_data = list(map(int, data[1].split(' ')))
 
                 if self._algo.result is None:
-                    self._algo.result(self._algo.initialiser())
+                    self._algo.result = self._algo.initialiser()
 
-                self._algo.result.data = np.array(solution.data)
+                self._algo.result.data = np.array(solution_data)
+                self._algo.result.fitness = float(data[2])
+
+            macop_line(self._algo)
+            macop_text(self._algo,
+                       f'Checkpoint found from `{self._filepath}` file.')
+            macop_text(
+                self._algo,
+                f'Restart algorithm from evaluation {self._algo.getEvaluation()}.'
+            )
+        else:
+            macop_text(
+                self._algo,
+                'No backup found... Start running algorithm from evaluation 0.'
+            )
+            logging.info(
+                "Can't load backup... Backup filepath not valid in Checkpoint")
+
+        macop_line(self._algo)
+
+
+class ContinuousCheckpoint(Callback):
+    """
+    ContinuousCheckpoint is used for loading previous computations and start again after loading checkpoint (only continuous solution)
+
+    Attributes:
+        algo: {:class:`~macop.algorithms.base.Algorithm`} -- main algorithm instance reference
+        every: {int} -- checkpoint frequency used (based on number of evaluations)
+        filepath: {str} -- file path where checkpoints will be saved
+    """
+
+    def run(self):
+        """
+        Check if necessary to do backup based on `every` variable
+        """
+        # get current best solution
+        solution = self._algo.result
+
+        currentEvaluation = self._algo.getGlobalEvaluation()
+
+        # backup if necessary
+        if currentEvaluation % self._every == 0:
+
+            logging.info("Checkpoint is done into " + self._filepath)
+
+            solution_data = ""
+            solutionSize = len(solution.data)
+
+            for index, val in enumerate(solution.data):
+                solution_data += str(val)
+
+                if index < solutionSize - 1:
+                    solution_data += ' '
+
+            line = str(currentEvaluation) + ';' + solution_data + ';' + str(
+                solution.fitness) + ';\n'
+
+            # check if file exists
+            if not os.path.exists(self._filepath):
+                with open(self._filepath, 'w') as f:
+                    f.write(line)
+            else:
+                with open(self._filepath, 'a') as f:
+                    f.write(line)
+
+    def load(self):
+        """
+        Load last backup line of solution and set algorithm state (best solution and evaluations) at this backup
+        """
+        if os.path.exists(self._filepath):
+
+            logging.info('Load best solution from last checkpoint')
+            with open(self._filepath) as f:
+
+                # get last line and read data
+                lastline = f.readlines()[-1]
+                data = lastline.split(';')
+
+                # get evaluation  information
+                globalEvaluation = int(data[0])
+
+                if self._algo.getParent() is not None:
+                    self._algo.getParent().setEvaluation(globalEvaluation)
+                else:
+                    self._algo.setEvaluation(globalEvaluation)
+
+                # get best solution data information
+                solution_data = list(map(float, data[1].split(' ')))
+
+                if self._algo.result is None:
+                    self._algo.result = self._algo.initialiser()
+
+                self._algo.result.data = np.array(solution_data)
                 self._algo.result.fitness = float(data[2])
 
             macop_line(self._algo)

+ 7 - 6
macop/operators/continuous/crossovers.py

@@ -43,23 +43,24 @@ class BasicDifferentialEvolutionCrossover(Crossover):
         self.F = F
 
 
-    def apply(self, solution):
+    def apply(self, solution1, solution2=None):
         """Create new solution based on solution passed as parameter
 
         Args:
-            solution: {:class:`~macop.solutions.base.Solution`} -- the solution to use for generating new solution
+            solution1: {:class:`~macop.solutions.base.Solution`} -- the first solution to use for generating new solution
+            solution2: {:class:`~macop.solutions.base.Solution`} -- the second solution to use for generating new solution
 
         Returns:
             {:class:`~macop.solutions.base.Solution`}: new continuous generated solution
         """
 
-        size = solution.size
+        size = solution1.size
 
-        solution1 = solution.clone()
+        solution1 = solution1.clone()
 
         # create two new random solutions using instance and its static method
-        solution2 = solution.random(size, interval=(self.mini, self.maxi))
-        solution3 = solution.random(size, interval=(self.mini, self.maxi))
+        solution2 = solution1.random(size, interval=(self.mini, self.maxi))
+        solution3 = solution1.random(size, interval=(self.mini, self.maxi))
 
         # apply crossover on the new computed solution
         for i in range(len(solution1.data)):