Cleanup of trajectory class

l2l/optimizers/crossentropy/optimizer.py

@@ -171,7 +171,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             traj.f_add_result('$set.$.individual', self.eval_pop[ind_index])
             traj.f_add_result('$set.$.fitness', fitness)

l2l/optimizers/evolution/optimizer.py

@@ -128,7 +128,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             traj.f_add_result('$set.$.individual', self.eval_pop[ind_index])
             traj.f_add_result('$set.$.fitness', fitness)

l2l/optimizers/evolutionstrategies/optimizer.py

@@ -128,7 +128,7 @@ def __init__(self,
         noise_std_shape = np.array(parameters.noise_std).shape
         assert noise_std_shape == () or noise_std_shape == self.current_individual_arr.shape
 
-        traj.f_add_derived_parameter(
+        traj.f_add_parameter(
             'dimension',
             self.current_individual_arr.shape,
             comment='The dimension of the parameter space of the optimizee')
@@ -189,7 +189,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             traj.f_add_result('$set.$.individual', self.eval_pop[ind_index])
             traj.f_add_result('$set.$.fitness', fitness)

l2l/optimizers/face/optimizer.py

@@ -107,10 +107,10 @@ def __init__(self, traj, optimizee_create_individual, optimizee_fitness_weights,
         traj.f_add_parameter('seed', np.uint32(parameters.seed),
                              comment='Random seed used by optimizer')
 
-        self.random_state = np.random.RandomState(seed=traj.par.seed)
+        self.random_state = np.random.RandomState(seed=traj.parameters.seed)
         temp_indiv, self.optimizee_individual_dict_spec = dict_to_list(self.optimizee_create_individual(),
                                                                        get_dict_spec=True)
-        traj.f_add_derived_parameter('dimension', len(temp_indiv),
+        traj.f_add_parameter('dimension', len(temp_indiv),
                                      comment='The dimension of the parameter space of the optimizee')
 
         # Added a generation-wise parameter logging
@@ -167,7 +167,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             traj.f_add_result('$set.$.individual', self.eval_pop[ind_index])
             traj.f_add_result('$set.$.fitness', fitness)

l2l/optimizers/gradientdescent/optimizer.py

@@ -119,7 +119,7 @@ def __init__(self, traj,
         traj.f_add_parameter('seed', np.uint32(parameters.seed), comment='Optimizer random seed')
 
         _, self.optimizee_individual_dict_spec = dict_to_list(self.optimizee_create_individual(), get_dict_spec=True)
-        self.random_state = np.random.RandomState(seed=traj.par.seed)
+        self.random_state = np.random.RandomState(seed=traj.parameters.seed)
 
         # Note that this array stores individuals as an np.array of floats as opposed to Individual-Dicts
         # This is because this array is used within the context of the gradient descent algorithm and
@@ -185,7 +185,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             individual = old_eval_pop[ind_index]
 

l2l/optimizers/naturalevolutionstrategies/optimizer.py

@@ -210,7 +210,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
 
             traj.f_add_result('$set.$.individual', self.eval_pop[ind_index])
             traj.f_add_result('$set.$.fitness', fitness)

l2l/optimizers/paralleltempering/optimizer.py

@@ -307,7 +307,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
             individual = old_eval_pop[ind_index]
 
             # Accept or reject the new solution

l2l/optimizers/simulatedannealing/optimizer.py

@@ -225,7 +225,7 @@ def post_process(self, traj, fitnesses_results):
             # We need to convert the current run index into an ind_idx
             # (index of individual within one generation)
             traj.v_idx = run_index
-            ind_index = traj.par.ind_idx
+            ind_index = traj.parameters.ind_idx
             individual = old_eval_pop[ind_index]
 
             # Accept or reject the new solution

l2l/tests/test_all.py

@@ -1,30 +1,27 @@
 import unittest
 
-
-from . import test_ce_optimizer
-from . import test_ga_optimizer
-from . import test_sa_optimizer
-from . import test_gd_optimizer
-from . import test_innerloop
-from . import test_outerloop
-from . import test_setup
-
-
-def suite():
-
-    suite = unittest.TestSuite()
-    suite.addTest(test_setup.suite())
-    suite.addTest(test_outerloop.suite())
-    suite.addTest(test_innerloop.suite())
-    suite.addTest(test_ce_optimizer.suite())
-    suite.addTest(test_sa_optimizer.suite())
-    suite.addTest(test_gd_optimizer.suite())
-    suite.addTest(test_ga_optimizer.suite())
-
-    return suite
-
-
-if __name__ == "__main__":
-
-    runner = unittest.TextTestRunner(verbosity=2)
-    runner.run(suite())
+import l2l.tests.test_ga_optimizer as test_ga_optimizer
+import l2l.tests.test_sa_optimizer as test_sa_optimizer
+import l2l.tests.test_gd_optimizer as test_gd_optimizer
+import l2l.tests.test_gs_optimizer as test_gs_optimizer
+import l2l.tests.test_pt_optimizer as test_pt_optimizer
+import l2l.tests.test_face_optimizer as test_face_optimizer
+import l2l.tests.test_es_optimizer as test_es_optimizer
+import l2l.tests.test_setup as test_setup
+
+
+suite  = unittest.TestSuite()
+loader = unittest.TestLoader()
+
+suite.addTests(test_setup.suite())
+suite.addTests(test_es_optimizer.suite())
+suite.addTests(test_sa_optimizer.suite())
+suite.addTests(test_gd_optimizer.suite())
+suite.addTests(test_ga_optimizer.suite())
+suite.addTests(test_gs_optimizer.suite())
+suite.addTests(test_face_optimizer.suite())
+suite.addTests(test_pt_optimizer.suite())
+suite.addTests(test_es_optimizer.suite())
+
+runner = unittest.TextTestRunner(verbosity=3)
+result = runner.run(suite)

l2l/tests/test_ce_optimizer.py

@@ -1,11 +1,12 @@
 import unittest
 
 import numpy as np
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 from l2l.optimizers.crossentropy.distribution import NoisyGaussian
 from l2l.optimizers.crossentropy import CrossEntropyOptimizer, CrossEntropyParameters
 
-class CEOptimizerTestCase(OptimizerTestCase):
+
+class CEOptimizerTestCase(TestCaseOptimizer):
 
     def test_setup(self):
 

l2l/tests/test_es_optimizer.py

@@ -1,10 +1,11 @@
 import unittest
 
 import numpy as np
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 from l2l.optimizers.evolutionstrategies import EvolutionStrategiesParameters, EvolutionStrategiesOptimizer
 
-class ESOptimizerTestCase(OptimizerTestCase):
+
+class TestCaseESOptimizer(TestCaseOptimizer):
 
     def test_setup(self):
 
@@ -41,7 +42,7 @@ def test_setup(self):
 
 
 def suite():
-    suite = unittest.makeSuite(CEOptimizerTestCase, 'test')
+    suite = unittest.makeSuite(TestCaseESOptimizer, 'test')
     return suite
 
 

l2l/tests/test_face_optimizer.py

@@ -1,12 +1,12 @@
 import unittest
 
 import numpy as np
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 from l2l.optimizers.crossentropy.distribution import Gaussian
 from l2l.optimizers.face import FACEOptimizer, FACEParameters
 
 
-class FACEOptimizerTestCase(OptimizerTestCase):
+class FACEOptimizerTestCase(TestCaseOptimizer):
 
     def test_setup(self):
 

l2l/tests/test_ga_optimizer.py

@@ -1,10 +1,10 @@
 import unittest
 
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 from l2l.optimizers.evolution import GeneticAlgorithmOptimizer, GeneticAlgorithmParameters
 
 
-class GAOptimizerTestCase(OptimizerTestCase):
+class GAOptimizerTestCase(TestCaseOptimizer):
 
     def test_setup(self):
 

l2l/tests/test_gd_optimizer.py

@@ -3,12 +3,12 @@
 import numpy as np
 from l2l.optimizers.gradientdescent.optimizer import GradientDescentOptimizer
 from l2l.optimizers.gradientdescent.optimizer import RMSPropParameters
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 
 from l2l import list_to_dict
 
 
-class GDOptimizerTestCase(OptimizerTestCase):
+class GDOptimizerTestCase(TestCaseOptimizer):
 
     def test_gd(self):
         optimizer_parameters = RMSPropParameters(learning_rate=0.01, exploration_step_size=0.01,

l2l/tests/test_gs_optimizer.py

@@ -1,13 +1,13 @@
 import unittest
 
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 
 from l2l.optimizers.gridsearch import GridSearchOptimizer, GridSearchParameters
 
 from l2l import list_to_dict
 
 
-class GSOptimizerTestCase(OptimizerTestCase):
+class GSOptimizerTestCase(TestCaseOptimizer):
 
     def test_gd(self):
         n_grid_divs_per_axis = 2

l2l/tests/test_optimizer.py

@@ -7,7 +7,7 @@
 from collections import namedtuple
 
 
-class OptimizerTestCase(unittest.TestCase):
+class TestCaseOptimizer(unittest.TestCase):
 
     def setUp(self):
         # Test function
@@ -23,3 +23,16 @@ def setUp(self):
                                                                               jube_parameter=jube_params)
         self.optimizee_parameters = namedtuple('OptimizeeParameters', [])
         self.optimizee = FunctionGeneratorOptimizee(self.trajectory,benchmark_function, seed=1)
+
+def suite():
+    suite = unittest.makeSuite(TestCaseOptimizer, 'test')
+    return suite
+
+
+def run():
+    runner = unittest.TextTestRunner(verbosity=2)
+    runner.run(suite())
+
+
+if __name__ == "__main__":
+    run()

l2l/tests/test_pt_optimizer.py

@@ -1,12 +1,12 @@
 import unittest
 
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 import numpy as np
 from l2l.optimizers.paralleltempering.optimizer import AvailableCoolingSchedules
 from l2l.optimizers.paralleltempering.optimizer import ParallelTemperingParameters, ParallelTemperingOptimizer
 
 
-class PTOptimizerTestCase(OptimizerTestCase):
+class PTOptimizerTestCase(TestCaseOptimizer):
 
     def test_sa(self):
         cooling_schedules = [AvailableCoolingSchedules.EXPONENTIAL_ADDAPTIVE,

l2l/tests/test_sa_optimizer.py

@@ -1,11 +1,11 @@
 import unittest
 
-from l2l.tests.test_optimizer import OptimizerTestCase
+from l2l.tests.test_optimizer import TestCaseOptimizer
 import numpy as np
 from l2l.optimizers.simulatedannealing.optimizer import SimulatedAnnealingParameters, SimulatedAnnealingOptimizer, AvailableCoolingSchedules
 
 
-class SAOptimizerTestCase(OptimizerTestCase):
+class SAOptimizerTestCase(TestCaseOptimizer):
 
     def test_sa(self):
         optimizer_parameters = SimulatedAnnealingParameters(n_parallel_runs=1, noisy_step=.03, temp_decay=.99, n_iteration=1,

l2l/tests/test_setup.py

@@ -11,16 +11,11 @@
 import os
 
 
-class SetupTestCase(unittest.TestCase):
+class TestCaseSetup(unittest.TestCase):
 
     def setUp(self):
         self.name = "test_trajectory"
-        try:
-            with open('../../bin/path.conf') as f:
-                root_dir_path = f.read().strip()
-        except FileNotFoundError:
-            self.fail("L2L is not well configured. Missing path file.")
-        self.paths = Paths(self.name, dict(run_num='test'), root_dir_path=root_dir_path, suffix="-" + self.name)
+        self.paths = Paths(self.name, dict(run_num='test'), root_dir_path=".", suffix="-" + self.name)
 
     def test_paths(self):
         self.assertIsNotNone(self.paths)
@@ -105,7 +100,7 @@ def test_juberunner_setup(self):
 
 
 def suite():
-    suite = unittest.makeSuite(SetupTestCase, 'test')
+    suite = unittest.makeSuite(TestCaseSetup, 'test')
     return suite
 
 

l2l/utils/JUBE_runner.py

@@ -24,7 +24,7 @@ def __init__(self, trajectory):
         """
         self.trajectory = trajectory
         self.done = False
-        if 'JUBE_params' not in self.trajectory.par.keys():
+        if 'JUBE_params' not in self.trajectory.parameters.keys():
             raise KeyError("The trajectory must contain the parameter group JUBE_params")
         args = self.trajectory.parameters["JUBE_params"].params
 

l2l/utils/environment.py

@@ -38,10 +38,10 @@ def run(self, runfunc):
         :return: the results of running a whole generation. Dictionary indexed by generation id.
         """
         result = {}
-        for it in range(self.trajectory.par['n_iteration']):
+        for it in range(self.trajectory.parameters['n_iteration']):
             if self.multiprocessing:
                 # Multiprocessing is done through JUBE, either with or without scheduler
-                logging.info("Environment run starting JUBERunner for n iterations: " + str(self.trajectory.par['n_iteration']))
+                logging.info("Environment run starting JUBERunner for n iterations: " + str(self.trajectory.parameters['n_iteration']))
                 jube = JUBERunner(self.trajectory)
                 result[it] = []
                 # Initialize new JUBE run and execute it
@@ -67,7 +67,7 @@ def run(self, runfunc):
                         logger.exception("Error during serial execution of individuals")
                     raise
             # Add results to the trajectory
-            self.trajectory.results.f_add_result_to_group("all_results", it, result[it])
+            self.trajectory.results.f_add_result(it, result[it])
             self.trajectory.current_results = result[it]
             # Perform the postprocessing step in order to generate the new parameter set
             self.postprocessing(self.trajectory, result[it])