Remove astype.

deepxde/data/function_spaces.py

@@ -85,13 +85,13 @@ def __init__(self, N=100, M=1):
         self.M = M
 
     def random(self, size):
-        return 2 * self.M * np.random.rand(size, self.N).astype(config.real(np)) - self.M
+        return 2 * self.M * np.random.rand(size, self.N) - self.M
 
     def eval_one(self, feature, x):
-        return np.dot(feature, x ** np.arange(self.N, dtype=config.real(np)))
+        return np.dot(feature, x ** np.arange(self.N))
 
     def eval_batch(self, features, xs):
-        mat = np.ones((self.N, len(xs)), dtype=config.real(np))
+        mat = np.ones((self.N, len(xs)))
         for i in range(1, self.N):
             mat[i] = np.ravel(xs**i)
         return np.dot(features, mat)
@@ -114,7 +114,7 @@ def __init__(self, N=100, M=1):
         self.M = M
 
     def random(self, size):
-        return 2 * self.M * np.random.rand(size, self.N).astype(config.real(np)) - self.M
+        return 2 * self.M * np.random.rand(size, self.N) - self.M
 
     def eval_one(self, feature, x):
         return np.polynomial.chebyshev.chebval(2 * x - 1, feature)
@@ -218,15 +218,15 @@ def bases(self, sensors):
         return np.array([np.ravel(f(sensors)) for f in self.eigfun])
 
     def random(self, size):
-        return np.random.randn(size, self.num_eig).astype(config.real(np))
+        return np.random.randn(size, self.num_eig)
 
     def eval_one(self, feature, x):
         eigfun = [f(x) for f in self.eigfun]
-        return np.sum(eigfun * feature).astype(config.real(np))
+        return np.sum(eigfun * feature)
 
     def eval_batch(self, features, xs):
         eigfun = np.array([np.ravel(f(xs)) for f in self.eigfun])
-        return np.dot(features, eigfun).astype(config.real(np))
+        return np.dot(features, eigfun)
 
 
 class GRF2D(FunctionSpace):