Added sample class

This new file provides sample classes for agent and environment

sample/new_modules_example.py

@@ -0,0 +1,79 @@
+'''
+Created on 9 de fev de 2017
+
+@author: goramos
+'''
+
+from environment import Environment
+from learner import Learner
+import numpy as np
+
+class SimpleEnv(Environment):
+
+    def __init__(self):
+        super(SimpleEnv, self).__init__()
+        self.actions = ['actA', 'actB']
+        self.calls_actB = 0
+
+    def get_state_actions(self, state=None):
+        return self.actions
+
+    def run_episode(self):#, max_steps=-1):
+        while True:
+            self.run_step()
+
+    def run_step(self):
+        learner = self._learners.values()[0]
+        s, a = learner.act_last()
+        print 'Agent chosen %s' % a
+        r = 1.0
+        if a == 'actB':
+            self.calls_actB += 1
+            r = 1.0 / self.calls_actB
+        learner.feedback_last(r, s)
+
+    #IGNORE
+    def has_episode_ended(self):
+        return
+
+class SimpleAg(Learner):
+
+    def __init__(self, name, env):
+        super(SimpleAg, self).__init__(name, env, self)
+        #self.avgs[] = dict({a:0 for a in self._env.get_state_actions(self._starting_state)})
+
+    def act_last(self, state=None, available_actions=None):
+        actions = self._env.get_state_actions()
+        rand = np.random.randint(len(actions))
+        return state, actions[rand]
+
+    def feedback_last(self, reward, new_state, prev_state=None, prev_action=None):
+        print 'Received reward of %f' % reward
+
+    #IGNORE
+    def reset_all(self):
+        pass
+    def reset_episodic(self):
+        pass
+    def act1(self, state=None, available_actions=None):
+        pass
+    def act2(self, state=None, available_actions=None):
+        pass
+    def act3(self, state=None, available_actions=None):
+        pass
+    def act4(self, state=None, available_actions=None):
+        pass
+    def feedback1(self, reward, new_state, prev_state=None, prev_action=None):
+        pass
+    def feedback2(self, reward, new_state, prev_state=None, prev_action=None):
+        pass
+    def feedback3(self, reward, new_state, prev_state=None, prev_action=None):
+        pass
+
+
+if __name__ == '__main__':
+
+    env = SimpleEnv()
+    learner = SimpleAg('A1', env)
+
+    env.run_episode()