format by black

examples/multiple_score.py

@@ -6,7 +6,7 @@
 # Make a set of candidates, test_X
 D = 2  # The number of params (the dimension of parameter space)
 Nx = 11  # The number of candidates
-N = Nx*Nx
+N = Nx * Nx
 
 # score = "HVPI"
 score = "EHVI"

examples/simple.py

@@ -9,7 +9,7 @@
 
 
 def simulator(actions: np.ndarray) -> np.ndarray:
-    """ Objective function
+    """Objective function
 
     Quadratic function, -Σ_i x_i^2
     Receives an array of actions (indices of candidates) and returns the corresponding results as an array

examples/simple_time.py

@@ -14,7 +14,7 @@
 
 
 def simulator(actions: np.ndarray) -> np.ndarray:
-    """ Objective function
+    """Objective function
 
     Quadratic function, -Σ_i x_i^2
     Receives an array of actions (indices of candidates) and returns the corresponding results as an array

physbo/gp/core/__init__.py

@@ -1,3 +1,4 @@
 from .prior import prior
 from .model import model
-#from predictor import predictor 
+
+# from predictor import predictor

physbo/gp/core/learning.py

@@ -309,7 +309,7 @@ def get_one_update(self, params, X, t):
 
 
 class adam(online):
-    """ default """
+    """default"""
 
     def __init__(self, gp, config):
         """
@@ -351,7 +351,7 @@ def get_one_update(self, params, X, t):
         """
         grad = self.gp.get_grad_marlik(params, X, t)
         self.m = self.m * self.beta + grad * (1 - self.beta)
-        self.v = self.v * self.gamma + grad ** 2 * (1 - self.gamma)
+        self.v = self.v * self.gamma + grad**2 * (1 - self.gamma)
         hat_m = self.m / (1 - self.beta ** (self.num_iter))
         hat_v = self.v / (1 - self.gamma ** (self.num_iter))
         return -self.alpha * hat_m / (np.sqrt(hat_v) + self.epsilon)

physbo/gp/core/model.py

@@ -301,7 +301,7 @@ def post_sampling(self, X, Z, params=None, N=1, alpha=1):
 
         fmean = self.get_post_fmean(X, Z, params=None)
         fcov = self.get_post_fcov(X, Z, params=None, diag=False)
-        return np.random.multivariate_normal(fmean, fcov * alpha ** 2, N)
+        return np.random.multivariate_normal(fmean, fcov * alpha**2, N)
 
     def predict_sampling(self, X, Z, params=None, N=1):
         """

physbo/gp/core/prior.py

@@ -3,7 +3,7 @@
 
 
 class prior:
-    """ prior of gaussian process """
+    """prior of gaussian process"""
 
     def __init__(self, mean, cov):
         """

physbo/gp/cov/gauss.py

@@ -5,7 +5,7 @@
 
 
 class gauss:
-    """ gaussian kernel """
+    """gaussian kernel"""
 
     def __init__(
         self,
@@ -68,7 +68,7 @@ def print_params(self):
         print(" Parameters of Gaussian kernel \n ")
         print(" width  = ", +self.width)
         print(" scale  = ", +self.scale)
-        print(" scale2 = ", +self.scale ** 2)
+        print(" scale2 = ", +self.scale**2)
         print(" \n")
 
     def prepare(self, params=None):
@@ -122,7 +122,7 @@ def get_grad(self, X, params=None):
             grad[0 : self.num_params - 1, :, :] = grad_width64(X, width, G)
         else:
             pairwise_dists = spatial.distance.pdist(X / width, "euclidean")
-            grad[0, :, :] = G * spatial.distance.squareform(pairwise_dists ** 2)
+            grad[0, :, :] = G * spatial.distance.squareform(pairwise_dists**2)
 
         grad[-1, :, :] = 2 * G
         return grad
@@ -150,7 +150,7 @@ def get_cov(self, X, Z=None, params=None, diag=False):
             covariant matrix
         """
         params, width, scale = self.prepare(params)
-        scale2 = scale ** 2
+        scale2 = scale**2
 
         if Z is None:
             if diag:
@@ -332,7 +332,7 @@ def rand_expans(self, num_basis, params=None):
         tupple (W, b, amp)
         """
         params, width, scale = self.prepare(params)
-        scale2 = scale ** 2
+        scale2 = scale**2
         amp = np.sqrt((2 * scale2) / num_basis)
         W = np.random.randn(num_basis, self.num_dim) / width
         b = np.random.rand(num_basis) * 2 * np.pi

physbo/gp/lik/gauss.py

@@ -3,7 +3,7 @@
 
 
 class gauss:
-    """ Gaussian likelihood function """
+    """Gaussian likelihood function"""
 
     def __init__(self, std=1, max_params=1e6, min_params=1e-6):
         """
@@ -99,7 +99,7 @@ def get_cov(self, num_data, params=None):
             Diagonal element matrix of exp(2.0*params)
         """
         std = self.trans_params(params)
-        var = std ** 2
+        var = std**2
         return var * np.identity(num_data)
 
     def get_grad(self, num_data, params=None):
@@ -119,7 +119,7 @@ def get_grad(self, num_data, params=None):
             Diagonal element matrix of 2.0 * exp(2.0*params)
         """
         std = self.trans_params(params)
-        var = std ** 2
+        var = std**2
         return var * np.identity(num_data) * 2
 
     def set_params(self, params):

physbo/gp/mean/const.py

@@ -2,7 +2,7 @@
 
 
 class const:
-    """ constant """
+    """constant"""
 
     def __init__(self, params=None, max_params=1e12, min_params=-1e12):
         """

physbo/gp/mean/zero.py

@@ -2,7 +2,7 @@
 
 
 class zero:
-    """ zero """
+    """zero"""
 
     def __init__(self):
         self.num_params = 0

physbo/gp/predictor.py

@@ -2,7 +2,7 @@
 
 
 class predictor(physbo.predictor.base_predictor):
-    """ predictor """
+    """predictor"""
 
     def __init__(self, config, model=None):
         """

physbo/misc/__init__.py

@@ -1,4 +1,4 @@
-from .centering  import centering
+from .centering import centering
 from .gauss_elim import gauss_elim
 from .set_config import set_config
 from ._src.diagAB import diagAB_64 as diagAB

physbo/opt/adam.py

@@ -72,7 +72,7 @@ def update(self, params, *args, **kwargs):
         """
         g = self.grad(params, *args, **kwargs)
         self.m = self.m * self.beta + g * (1 - self.beta)
-        self.v = self.v * self.gamma + g ** 2 * (1 - self.gamma)
+        self.v = self.v * self.gamma + g**2 * (1 - self.gamma)
         hat_m = self.m / (1 - self.beta ** (self.epoch + 1))
         hat_v = self.v / (1 - self.gamma ** (self.epoch + 1))
         self.epoch += 1

physbo/search/discrete/policy.py

@@ -60,7 +60,9 @@ def __init__(self, test_X, config=None, initial_data=None, comm=None):
             self.mpisize = comm.size
             self.mpirank = comm.rank
             self.actions = np.array_split(self.actions, self.mpisize)[self.mpirank]
-            self.config.learning.is_disp = (self.config.learning.is_disp and self.mpirank == 0)
+            self.config.learning.is_disp = (
+                self.config.learning.is_disp and self.mpirank == 0
+            )
 
     def set_seed(self, seed):
         """
@@ -130,7 +132,9 @@ def write(
         )
         self.training.add(X=X, t=t, Z=Z)
         local_index = np.searchsorted(self.actions, action)
-        local_index = local_index[np.take(self.actions, local_index, mode='clip') == action]
+        local_index = local_index[
+            np.take(self.actions, local_index, mode="clip") == action
+        ]
         self.actions = self._delete_actions(local_index)
         if self.new_data is None:
             self.new_data = variable(X=X, t=t, Z=Z)
@@ -168,7 +172,6 @@ def random_search(
             utility.show_interactive_mode(simulator, self.history)
 
         for n in range(0, max_num_probes):
-
             time_total = time.time()
             if is_disp and N > 1:
                 utility.show_start_message_multi_search(self.history.num_runs)
@@ -479,7 +482,9 @@ def _get_marginal_score(self, mode, chosen_actions, K, alpha):
 
         # draw K samples of the values of objective function of chosen actions
         new_test_local = self.test.get_subset(chosen_actions)
-        virtual_t_local = self.predictor.get_predict_samples(self.training, new_test_local, K)
+        virtual_t_local = self.predictor.get_predict_samples(
+            self.training, new_test_local, K
+        )
         if self.mpisize == 1:
             new_test = new_test_local
             virtual_t = virtual_t_local

physbo/search/discrete/results.py

@@ -22,19 +22,19 @@ def __init__(self):
 
     @property
     def time_total(self):
-        return copy.copy(self.time_total_[0:self.num_runs])
+        return copy.copy(self.time_total_[0 : self.num_runs])
 
     @property
     def time_update_predictor(self):
-        return copy.copy(self.time_update_predictor_[0:self.num_runs])
+        return copy.copy(self.time_update_predictor_[0 : self.num_runs])
 
     @property
     def time_get_action(self):
-        return copy.copy(self.time_get_action_[0:self.num_runs])
+        return copy.copy(self.time_get_action_[0 : self.num_runs])
 
     @property
     def time_run_simulator(self):
-        return copy.copy(self.time_run_simulator_[0:self.num_runs])
+        return copy.copy(self.time_run_simulator_[0 : self.num_runs])
 
     def write(
         self,

physbo/search/discrete_multi/policy.py

@@ -100,7 +100,9 @@ def write(
                 self.new_data_list[i].add(X=X, t=t[:, i], Z=Z)
             self.training_list[i].add(X=X, t=t[:, i], Z=Z)
         local_index = np.searchsorted(self.actions, action)
-        local_index = local_index[np.take(self.actions, local_index, mode='clip') == action]
+        local_index = local_index[
+            np.take(self.actions, local_index, mode="clip") == action
+        ]
         self.actions = self._delete_actions(local_index)
 
     def _model(self, i):
@@ -123,7 +125,6 @@ def random_search(
         is_disp=True,
         disp_pareto_set=False,
     ):
-
         if self.mpirank != 0:
             is_disp = False
 
@@ -133,7 +134,6 @@ def random_search(
             utility.show_interactive_mode(simulator, self.history)
 
         for n in range(0, max_num_probes):
-
             time_total = time.time()
             if is_disp and N > 1:
                 utility.show_start_message_multi_search(
@@ -181,7 +181,6 @@ def bayes_search(
         interval=0,
         num_rand_basis=0,
     ):
-
         if self.mpirank != 0:
             is_disp = False
 
@@ -418,14 +417,18 @@ def _get_marginal_score(self, mode, chosen_actions, K, alpha):
         virtual_t_list = [np.zeros((K, 0)) for _ in range(self.num_objectives)]
         for i in range(self.num_objectives):
             new_test_local = self.test_list[i].get_subset(chosen_actions)
-            virtual_t_local = self.predictor_list[i].get_predict_samples(self.training_list[i], new_test_local, K)
+            virtual_t_local = self.predictor_list[i].get_predict_samples(
+                self.training_list[i], new_test_local, K
+            )
             if self.mpisize == 1:
                 new_test_list[i] = new_test_local
                 virtual_t_list[i] = virtual_t_local
             else:
                 for nt in self.mpicomm.allgather(new_test_local):
                     new_test_list[i].add(X=nt.X, t=nt.t, Z=nt.Z)
-                virtual_t_list[i] = np.concatenate(self.mpicomm.allgather(virtual_t_local), axis=1)
+                virtual_t_list[i] = np.concatenate(
+                    self.mpicomm.allgather(virtual_t_local), axis=1
+                )
 
         for k in range(K):
             predictor_list = [copy.deepcopy(p) for p in self.predictor_list]
@@ -438,12 +441,17 @@ def _get_marginal_score(self, mode, chosen_actions, K, alpha):
                 if virtual_train.Z is None:
                     training_list[i].add(virtual_train.X, virtual_train.t)
                 else:
-                    training_list[i].add(virtual_train.X, virtual_train.t, virtual_train.Z)
+                    training_list[i].add(
+                        virtual_train.X, virtual_train.t, virtual_train.Z
+                    )
 
                 predictor_list[i].update(training_list[i], virtual_train)
 
             f[k, :] = self.get_score(
-                mode, predictor_list=predictor_list, training_list=training_list, parallel=False
+                mode,
+                predictor_list=predictor_list,
+                training_list=training_list,
+                parallel=False,
             )
         return np.mean(f, axis=0)
 

physbo/search/discrete_multi/results.py

@@ -25,19 +25,19 @@ def __init__(self, num_objectives):
 
     @property
     def time_total(self):
-        return copy.copy(self._time_total[0:self.num_runs])
+        return copy.copy(self._time_total[0 : self.num_runs])
 
     @property
     def time_update_predictor(self):
-        return copy.copy(self._time_update_predictor[0:self.num_runs])
+        return copy.copy(self._time_update_predictor[0 : self.num_runs])
 
     @property
     def time_get_action(self):
-        return copy.copy(self._time_get_action[0:self.num_runs])
+        return copy.copy(self._time_get_action[0 : self.num_runs])
 
     @property
     def time_run_simulator(self):
-        return copy.copy(self._time_run_simulator[0:self.num_runs])
+        return copy.copy(self._time_run_simulator[0 : self.num_runs])
 
     def write(
         self,

physbo/search/score_multi.py

@@ -30,6 +30,7 @@ def score(mode, predictor_list, test, training_list, **kwargs):
         raise NotImplementedError("mode must be EHVI, HVPI or TS.")
     return f
 
+
 def HVPI(fmean, fstd, pareto):
     """
     Calculate Hypervolume-based Probability of Improvement (HVPI).

tests/integrated/test_gp.py

@@ -17,7 +17,7 @@ def __init__(self, noise=1.0e-5):
         np.random.seed(12345)
         self.nslice = 11
         self.dim = 2
-        self.N = self.nslice ** self.dim
+        self.N = self.nslice**self.dim
         self.X = np.zeros((self.N, self.dim))
         self.t = np.zeros(self.N)
         for i, x in enumerate(

tests/integrated/test_interactive_example.py

@@ -16,9 +16,11 @@ class simulator:
     def __init__(self):
         self.nslice = 11
         self.dim = 2
-        self.N = self.nslice ** self.dim
+        self.N = self.nslice**self.dim
         self.X = np.zeros((self.N, self.dim))
-        for i, x in enumerate(product(np.linspace(0.0, 1.0, self.nslice), repeat=self.dim)):
+        for i, x in enumerate(
+            product(np.linspace(0.0, 1.0, self.nslice), repeat=self.dim)
+        ):
             self.X[i, :] = list(x)
 
     def __call__(self, actions):
@@ -32,14 +34,18 @@ def test_interactive():
     policy = physbo.search.discrete.policy(test_X=sim.X)
     policy.set_seed(12345)
 
-    actions = policy.random_search(max_num_probes=1, num_search_each_probe=nrand, simulator=None)
+    actions = policy.random_search(
+        max_num_probes=1, num_search_each_probe=nrand, simulator=None
+    )
     targets = sim(actions)
     print(actions)
     print(targets)
     policy.write(actions, targets)
     physbo.search.utility.show_search_results(policy.history, nrand)
 
-    actions = policy.bayes_search(max_num_probes=1, num_search_each_probe=nsearch, simulator=None, score='TS')
+    actions = policy.bayes_search(
+        max_num_probes=1, num_search_each_probe=nsearch, simulator=None, score="TS"
+    )
     targets = sim(actions)
     print(actions)
     print(targets)