cleaned up

k-bandits.py

@@ -14,10 +14,8 @@ def bandit(bias):
 
 if __name__ == '__main__':
 
-	kv = 1.0
-	epsilon = 0.1
+	epsilon = 0.0		# Anti-greediness
 	rho = 0.01			# reduce epsilon with age
-	alpha = 1.0
 
 	qu = np.zeros(K)
 	nu = np.zeros(K)
@@ -43,7 +41,7 @@ if __name__ == '__main__':
 			# get reward from bandit
 			r = bandit(ql_star[a])
 			nu[a] = nu[a] + 1
-			qu[a] = qu[a] + alpha*(r - qu[a])/nu[a]
+			qu[a] = qu[a] + (r - qu[a])/nu[a]
 			r_sum += r
 		r_vec.append(r_sum/N)
 		# Reduce tendency to explore with number of steps (or with age for humans)