Added unit test for adaptive random search.

Author: skytreader

python/2_Stochastic_Algorithms/adaptive_random_search_tests.py

@@ -0,0 +1,81 @@
+#! usr/bin/env python3
+
+import adaptive_random_search as ars
+
+import unittest
+
+class AdaptiveRandomSearchTests(unittest.TestCase):
+	
+	def test_objective_function(self):
+		#integer
+		self.assertEqual(99**2, ars.objective_function([99]))
+		#float
+		self.assertEqual(0.1**2.0, ars.objective_function([0.1]))
+		#vector
+		self.assertEqual((1**2) + (2**2) + (3**2), ars.objective_function([1,2,3]))
+	
+	def test_rand_in_bounds(self):
+		x = ars.rand_in_bounds(0, 20)
+		self.assertTrue(x >= 0)
+		self.assertTrue(x < 20)
+		
+		x = ars.rand_in_bounds(-20, -1)
+		self.assertTrue(x >= -20)
+		self.assertTrue(x < -1)
+		
+		x = ars.rand_in_bounds(-10, 20)
+		self.assertTrue(x >= -10)
+		self.assertTrue(x < 20)
+	
+	def test_random_vector(self):
+		bounds, trials, size = [-3, 3], 300, 20
+		minmax = [bounds for i in range(size)]
+		
+		for i in range(trials):
+			vector, total = ars.random_vector(minmax), 0.0
+			self.assertEqual(size, len(vector))
+			
+			for v in vector:
+				self.assertTrue(v >= bounds[0])
+				self.assertTrue(v < bounds[1])
+				total += v
+			
+			#TODO: test with total
+	
+	def test_take_step(self):
+		p = ars.take_step([[0, 100]], [50], 3.3)
+		self.assertTrue(p[0] >= 50 - 3.3)
+		self.assertTrue(p[0] <= 50 + 3.3)
+		
+		p = ars.take_step([[0, 1]], [0], 3.3)
+		self.assertTrue(p[0] >= 0)
+		self.assertTrue(p[0] < 1)
+	
+	def test_large_step_size(self):
+		# Test using small factor
+		s = ars.large_step_size(0, 1, 2, 3, 100)
+		self.assertEqual(1*2, s)
+		
+		# Test using large factor
+		s = ars.large_step_size(100, 1, 2, 3, 100)
+		self.assertEqual(1*3, s)
+	
+	def test_take_steps(self):
+		vector_content = 5
+		for i in range(20):
+			step1, step2 = ars.take_steps([[0,10]], {"vector":[vector_content]}, 1, 3)
+			self.assertTrue(step1["vector"] is not None)
+			self.assertTrue(step1["cost"] is not None)
+			self.assertTrue(step1["vector"][0] >= vector_content - 1)
+			self.assertTrue(step1["vector"][0] < vector_content + 1)
+			
+			self.assertTrue(step2["vector"] is not None)
+			self.assertTrue(step2["cost"] is not None)
+			self.assertTrue(step2["vector"][0] >= vector_content - 3)
+			self.assertTrue(step2["vector"][0] < vector_content + 3)
+	
+	#TODO : Test search
+	
+if __name__ == "__main__":
+	tests = unittest.TestLoader().loadTestsFromTestCase(AdaptiveRandomSearchTests)
+	unittest.TextTestRunner(verbosity=2).run(tests)