black trial

PUQ/designmethods/gen_funcs/acquisition_funcs.py

@@ -57,7 +57,7 @@ def maxvar(
     obsvar3d = obsvar.reshape(1, d, d)
     diags = np.diag(obsvar[real_x, real_x.T])
     d_real = real_x.shape[0]
-    coef = (2**d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
+    coef = (2 ** d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
 
     # Create a candidate list.
     n_clist = 100 * n

PUQ/designmethods/gen_funcs/acquisition_funcs_support.py

@@ -39,7 +39,7 @@ def compute_postvar(obs, emumean, covmat1, covmat2, coef):
     part1 = part1 * (1 / coef)
 
     # part2 = compute_likelihood(emumean, emuvar, obs, obsvar, is_cov)
-    part2 = part2**2
+    part2 = part2 ** 2
 
     return part1 - part2
 
@@ -73,7 +73,7 @@ def compute_eivar(summatrix, emuphi, emumean, emuvar, obs, is_cov, prioreval):
     denum = multiple_determinants(covmat2)
     part2 = rndpdf / np.sqrt(denum)
     # print(part2.shape)
-    return -np.sum(part2 * (prioreval**2))
+    return -np.sum(part2 * (prioreval ** 2))
 
 
 def multiple_pdfs(x, means, covs):

PUQ/designmethods/utils.py

@@ -3,8 +3,12 @@
 
 def parse_arguments():
     parser = argparse.ArgumentParser("Parameters for calibration")
-    parser.add_argument("-nworkers", metavar="N2", default=2, type=int, help="Number of workers.")
-    parser.add_argument("-minibatch", metavar="N2", default=1, type=int, help="Minibatch size.")
+    parser.add_argument(
+        "-nworkers", metavar="N2", default=2, type=int, help="Number of workers."
+    )
+    parser.add_argument(
+        "-minibatch", metavar="N2", default=1, type=int, help="Minibatch size."
+    )
     parser.add_argument(
         "-max_eval",
         metavar="N2",

PUQ/posterior.py

@@ -23,7 +23,7 @@ def predict(self, theta):
         obsvar3d = obsvar.reshape(1, d, d)
         diags = np.diag(obsvar[real_x, real_x.T])
         d_real = real_x.shape[0]
-        coef = (2**d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
+        coef = (2 ** d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
 
         emupred_test = emu.predict(x=data_cls.x, theta=theta)
         emumean = emupred_test.mean()
@@ -33,7 +33,7 @@ def predict(self, theta):
         var_obsvar1 = emuvarT + obsvar3d
         var_obsvar2 = emuvarT + 0.5 * obsvar3d
         diags = np.diag(obsvar[real_x, real_x.T])
-        coef = (2**d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
+        coef = (2 ** d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
 
         postmean = multiple_pdfs(
             obs, emumeanT[:, real_x.flatten()], var_obsvar1[:, real_x, real_x.T]
@@ -76,7 +76,7 @@ def compute_postvar(obs, emumean, covmat1, covmat2, coef):
     part1 = part1 * (1 / coef)
 
     # part2 = compute_likelihood(emumean, emuvar, obs, obsvar, is_cov)
-    part2 = part2**2
+    part2 = part2 ** 2
 
     return part1 - part2
 

PUQ/surrogatemethods/PCGP.py

@@ -295,7 +295,7 @@ def predict(predinfo, fitinfo, x, theta, **kwargs):
                 predvars_gradtheta[:, k, :] = -(infos[k]["sig2"] * 2) * np.einsum(
                     "ij,ijk->ik", rVh2, dr
                 )
-        predvars[:, k] = infos[k]["sig2"] * np.abs(1 - np.sum(rVh**2, 1))
+        predvars[:, k] = infos[k]["sig2"] * np.abs(1 - np.sum(rVh ** 2, 1))
 
     # calculate predictive mean and variance
     predinfo["mean"] = np.full((x.shape[0], theta.shape[0]), np.nan)
@@ -427,7 +427,7 @@ def acquisition(fitinfo, x, theta1, theta2, **kwargs):
         )
         # print('predcov:', predcov.shape)
 
-        predvars_2[:, k] = infos[k]["sig2"] * np.abs(1 - np.sum(rVh_2**2, 1))
+        predvars_2[:, k] = infos[k]["sig2"] * np.abs(1 - np.sum(rVh_2 ** 2, 1))
         predvars_2[:, k] += infos[k]["nug"]
 
         # print('predvars_2:', predvars_2.shape)
@@ -466,7 +466,7 @@ def predictlpdf(predinfo, f, return_grad=False, addvar=0, **kwargs):
     Gf = predinfo["phi"].T * (1 / np.sqrt(totvar))
     Gfrf = Gf @ rf
     Gf2 = Gf @ Gf.T
-    likv = np.sum(rf**2, 0)
+    likv = np.sum(rf ** 2, 0)
     if return_grad:
         rf2 = -(
             predinfo["mean_gradtheta"].transpose(2, 1, 0) * (1 / np.sqrt(totvar))
@@ -525,10 +525,10 @@ def __standardizef(fitinfo, offset=None, scale=None):
 
     # Assigning new values to the dictionary
     U, S, _ = np.linalg.svd(fs.T, full_matrices=False)
-    Sp = S**2 - epsilonPC
+    Sp = S ** 2 - epsilonPC
     Up = U[:, Sp > 0]
 
-    extravar = np.nanmean((fs - fs @ Up @ Up.T) ** 2, 0) * (scale**2)
+    extravar = np.nanmean((fs - fs @ Up @ Up.T) ** 2, 0) * (scale ** 2)
 
     standardpcinfo = {
         "offset": offset,
@@ -555,7 +555,7 @@ def __PCs(fitinfo):
         S = fitinfo["standardpcinfo"]["S"]
     else:
         U, S, _ = np.linalg.svd(fs.T, full_matrices=False)
-    Sp = S**2 - epsilonPC
+    Sp = S ** 2 - epsilonPC
     pct = U[:, Sp > 0]
     pcw = np.sqrt(Sp[Sp > 0])
 
@@ -747,7 +747,7 @@ def scaledlikgrad(hypv):
         fcenter = Vh.T @ g
         subinfo["Vh"] = Vh
         n = subinfo["R"].shape[0]
-        subinfo["sig2"] = (np.mean(fcenter**2) * n + sig2ofconst) / (n + sig2ofconst)
+        subinfo["sig2"] = (np.mean(fcenter ** 2) * n + sig2ofconst) / (n + sig2ofconst)
         subinfo["Rinv"] = V @ np.diag(1 / W) @ V.T
     else:
         subinfo["hyp"] = hypn
@@ -764,7 +764,7 @@ def scaledlikgrad(hypv):
         W, V = np.linalg.eigh(subinfo["R"])
         Vh = V / np.sqrt(np.abs(W))
         fcenter = Vh.T @ g
-        subinfo["sig2"] = (np.mean(fcenter**2) * n + sig2ofconst) / (n + sig2ofconst)
+        subinfo["sig2"] = (np.mean(fcenter ** 2) * n + sig2ofconst) / (n + sig2ofconst)
         subinfo["Rinv"] = Vh @ Vh.T
         subinfo["Vh"] = Vh
     subinfo["pw"] = subinfo["Rinv"] @ g
@@ -785,7 +785,7 @@ def __negloglik(hyp, info):
     n = info["g"].shape[0]
 
     sig2ofconst = info["sig2ofconst"]
-    sig2hat = (n * np.mean(fcenter**2) + sig2ofconst) / (n + sig2ofconst)
+    sig2hat = (n * np.mean(fcenter ** 2) + sig2ofconst) / (n + sig2ofconst)
     negloglik = 1 / 2 * np.sum(np.log(np.abs(W))) + 1 / 2 * n * np.log(sig2hat)
     negloglik += 0.5 * np.sum(
         ((10 ** (-8) + hyp - info["hypregmean"]) / (info["hypregstd"])) ** 2
@@ -816,7 +816,7 @@ def __negloglikgrad(hyp, info):
     n = info["g"].shape[0]
 
     sig2ofconst = info["sig2ofconst"]
-    sig2hat = (n * np.mean(fcenter**2) + sig2ofconst) / (n + sig2ofconst)
+    sig2hat = (n * np.mean(fcenter ** 2) + sig2ofconst) / (n + sig2ofconst)
     dnegloglik = np.zeros(dR.shape[2])
     Rinv = Vh @ Vh.T
 

examples/figure1b.py

@@ -1,7 +1,8 @@
 import numpy as np
 import matplotlib.pyplot as plt
 import scipy.stats as sps
-#import os
+
+# import os
 from PUQ.design import designer
 from PUQ.designmethods.utils import parse_arguments
 from PUQ.prior import prior_dist
@@ -114,7 +115,7 @@ def sim(self, H, persis_info, sim_specs, libE_info):
     ax.tick_params(axis="both", labelsize=16)
     plt.savefig("Figure1b.png", bbox_inches="tight")
 
-    #if os.path.exists('ensemble.log'):
+    # if os.path.exists('ensemble.log'):
     #    os.remove('ensemble.log')
-    #if os.path.exists('libE_stats.txt'):
+    # if os.path.exists('libE_stats.txt'):
     #    os.remove('libE_stats.txt')

examples/figure3.py

@@ -79,7 +79,7 @@ def compute_eivar_full(clist, theta, theta_test, emu):
             var_obsvar1 = emuvarT + obsvar3d
             var_obsvar2 = emuvarT + 0.5 * obsvar3d
             diags = np.diag(obsvar[real_x, real_x.T])
-            coef = (2**d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
+            coef = (2 ** d_real) * (np.sqrt(np.pi) ** d_real) * np.sqrt(np.prod(diags))
 
             # Get the n_ref x d x d x n_cand phi matrix
             emuphi4d = emu.acquisition(x=x, theta1=theta_test, theta2=clist)

examples/plots.py

@@ -2,6 +2,7 @@
 import matplotlib.pyplot as plt
 import os
 
+
 def plotline(df, methods, rep_no, w=2, b=1, s="banana", ylim=[0.000001, 1], idstart=0):
     colors = ["red", "green", "blue", "cyan", "magenta"]
     markers = ["o", "+", "*", "D", "v"]
@@ -49,9 +50,7 @@ def plotline(df, methods, rep_no, w=2, b=1, s="banana", ylim=[0.000001, 1], idst
     ax.tick_params(axis="both", labelsize=ft)
     plt.savefig("Figure6_" + s + ".png", bbox_inches="tight")
 
-
-
-    #if os.path.exists('ensemble.log'):
+    # if os.path.exists('ensemble.log'):
     #    os.remove('ensemble.log')
-    #if os.path.exists('libE_stats.txt'):
+    # if os.path.exists('libE_stats.txt'):
     #    os.remove('libE_stats.txt')

examples/test_funcs.py

@@ -38,7 +38,7 @@ class banana:
     def __init__(self):
         self.data_name = "banana"
         self.thetalimits = np.array([[-20, 20], [-10, 5]])
-        self.obsvar = np.array([[10**2, 0], [0, 1]])
+        self.obsvar = np.array([[10 ** 2, 0], [0, 1]])
         self.real_data = np.array([[1, 3]], dtype="float64")
         self.out = [("f", float, (2,))]
         self.p = 2
@@ -47,7 +47,7 @@ def __init__(self):
         self.real_x = np.arange(0, self.d)[:, None]
 
     def function(self, theta1, theta2):
-        f = np.array([theta1, theta2 + 0.03 * theta1**2])
+        f = np.array([theta1, theta2 + 0.03 * theta1 ** 2])
         return f
 
     def sim(self, H, persis_info, sim_specs, libE_info):
@@ -101,7 +101,7 @@ def __init__(self):
         self.real_x = np.arange(0, self.d)[:, None]
 
     def function(self, theta1, theta2):
-        f = np.array([theta2 - theta1**2, theta2 - theta1])
+        f = np.array([theta2 - theta1 ** 2, theta2 - theta1])
         return f
 
     def sim(self, H, persis_info, sim_specs, libE_info):