Added observer pattern to allow callbacks

Author: aprams

bayes_opt/__init__.py

@@ -1,4 +1,5 @@
-from .bayesian_optimization import BayesianOptimization
+from .bayesian_optimization import BayesianOptimization, Events
 from .helpers import UtilityFunction
+from .observer import Observer
 
-__all__ = ["BayesianOptimization", "UtilityFunction"]
+__all__ = ["BayesianOptimization", "UtilityFunction", "Events", "Observer"]

bayes_opt/bayesian_optimization.py

@@ -7,9 +7,10 @@
 from sklearn.gaussian_process.kernels import Matern
 from .helpers import (UtilityFunction, PrintLog, acq_max, ensure_rng)
 from .target_space import TargetSpace
+from .observer import Observable
 
 
-class BayesianOptimization(object):
+class BayesianOptimization(Observable):
 
     def __init__(self, f, pbounds, random_state=None, verbose=1):
         """
@@ -71,6 +72,10 @@ def __init__(self, f, pbounds, random_state=None, verbose=1):
         # Verbose
         self.verbose = verbose
 
+        # Event initialization
+        events = [Events.INIT_DONE, Events.FIT_STEP_DONE, Events.FIT_DONE]
+        super(BayesianOptimization, self).__init__(events)
+
     def init(self, init_points):
         """
         Initialization method to kick start the optimization process. It is a
@@ -100,6 +105,9 @@ def init(self, init_points):
         # Updates the flag
         self.initialized = True
 
+        # Notify about finished init method
+        self.dispatch(Events.INIT_DONE)
+
     def _observe_point(self, x):
         y = self.space.observe_point(x)
         if self.verbose:
@@ -303,10 +311,16 @@ def maximize(self,
             # Keep track of total number of iterations
             self.i += 1
 
+            # Notify about finished iteration
+            self.dispatch(Events.FIT_STEP_DONE)
+
         # Print a final report if verbose active.
         if self.verbose:
             self.plog.print_summary()
 
+        # Notify about finished optimization
+        self.dispatch(Events.FIT_DONE)
+
     def points_to_csv(self, file_name):
         """
         After training all points for which we know target variable
@@ -353,3 +367,9 @@ def bounds(self):
     def dim(self):
         warnings.warn("use self.space.dim instead", DeprecationWarning)
         return self.space.dim
+
+
+class Events(object):
+    INIT_DONE = 'initialized'
+    FIT_STEP_DONE = 'fit_step_done'
+    FIT_DONE = 'fit_done'

bayes_opt/observer.py

@@ -0,0 +1,36 @@
+# Inspired/Taken from https://www.protechtraining.com/blog/post/879#simple-observer
+
+
+class Observer:
+    def update(self, event, instance):
+        # Avoid circular import
+        from .bayesian_optimization import Events
+        if event is Events.INIT_DONE:
+            print("Initialization completed")
+        elif event is Events.FIT_STEP_DONE:
+            print("Optimization step finished, current max: ", instance.res['max'])
+        elif event is Events.FIT_DONE:
+            print("Optimization finished, maximum value at: ", instance.res['max'])
+
+
+class Observable(object):
+    def __init__(self, events):
+        # maps event names to subscribers
+        # str -> dict
+        self.events = {event: dict()
+                       for event in events}
+
+    def get_subscribers(self, event):
+        return self.events[event]
+
+    def register(self, event, who, callback=None):
+        if callback == None:
+            callback = getattr(who, 'update')
+        self.get_subscribers(event)[who] = callback
+
+    def unregister(self, event, who):
+        del self.get_subscribers(event)[who]
+
+    def dispatch(self, event):
+        for subscriber, callback in self.get_subscribers(event).items():
+            callback(event, self)

examples/exploitation vs exploration.ipynb

@@ -6,7 +6,6 @@
 from sklearn.gaussian_process.kernels import Matern
 from bayes_opt.helpers import UtilityFunction, acq_max, ensure_rng
 
-
 def get_globals():
     X = np.array([
         [0.00, 0.00],

tests/test_helper_functions.py

@@ -0,0 +1,49 @@
+import unittest
+from bayes_opt.observer import Observable
+
+
+class TestObserver():
+    def __init__(self):
+        self.counter = 0
+
+    def update(self, event, instance):
+        self.counter += 1
+
+class TestObserverPattern(unittest.TestCase):
+    def setUp(self):
+        events = ['a', 'b']
+        self.observable = Observable(events)
+        self.observer = TestObserver()
+
+    def test_get_subscribers(self):
+        self.observable.register('a', self.observer)
+        self.assertTrue(self.observer in self.observable.get_subscribers('a'))
+        self.assertTrue(len(self.observable.get_subscribers('a').keys()) == 1)
+        self.assertTrue(len(self.observable.get_subscribers('b').keys()) == 0)
+
+    def test_register(self):
+        self.observable.register('a', self.observer)
+        self.assertTrue(self.observer in self.observable.get_subscribers('a'))
+
+    def test_unregister(self):
+        self.observable.register('a', self.observer)
+        self.observable.unregister('a', self.observer)
+        self.assertTrue(self.observer not in self.observable.get_subscribers('a'))
+
+    def test_dispatch(self):
+        test_observer = TestObserver()
+        self.observable.register('b', test_observer)
+        self.observable.dispatch('b')
+        self.observable.dispatch('b')
+
+        self.assertTrue(test_observer.counter == 2)
+
+
+if __name__ == '__main__':
+    r"""
+    CommandLine:
+        python tests/test_observer.py
+    """
+    # unittest.main()
+    import pytest
+    pytest.main([__file__])