Move to Black

Author: wesselb

README.md

@@ -3,6 +3,7 @@
 [![CI](https://github.com/wesselb/lab/workflows/CI/badge.svg?branch=master)](https://github.com/wesselb/lab/actions?query=workflow%3ACI)
 [![Coverage Status](https://coveralls.io/repos/github/wesselb/lab/badge.svg?branch=master&service=github)](https://coveralls.io/github/wesselb/lab?branch=master)
 [![Latest Docs](https://img.shields.io/badge/docs-latest-blue.svg)](https://wesselb.github.io/lab)
+[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black)
 
 A generic interface for linear algebra backends: code it once, run it on any 
 backend

benchmark.py

@@ -44,6 +44,7 @@ def f2(x):
     z = f1(x)
 us_lab = (time() - s) / its * 1e6
 
-print('Overhead: {:.1f} us / {:.1f} %'
-      ''.format(us_lab - us_native,
-                100 * (us_lab / us_native - 1)))
+print(
+    "Overhead: {:.1f} us / {:.1f} %"
+    "".format(us_lab - us_native, 100 * (us_lab / us_native - 1))
+)

lab/autograd/custom.py

@@ -3,7 +3,7 @@
 
 from ..util import as_tuple
 
-__all__ = ['autograd_register']
+__all__ = ["autograd_register"]
 _dispatch = Dispatcher()
 
 

lab/autograd/linear_algebra.py

@@ -5,11 +5,7 @@
 
 from . import dispatch, B, Numeric
 from .custom import autograd_register
-from ..custom import (
-    toeplitz_solve, s_toeplitz_solve,
-    expm, s_expm,
-    logm, s_logm
-)
+from ..custom import toeplitz_solve, s_toeplitz_solve, expm, s_expm, logm, s_logm
 from ..linear_algebra import _default_perm
 from ..util import batch_computation
 
@@ -41,8 +37,7 @@ def transpose(a, perm=None):
 @dispatch(Numeric)
 def trace(a, axis1=0, axis2=1):
     if axis1 == axis2:
-        raise ValueError('Keyword arguments axis1 and axis2 cannot be the '
-                         'same.')
+        raise ValueError("Keyword arguments axis1 and axis2 cannot be the same.")
 
     # AutoGrad does not support the `axis1` and `axis2` arguments...
 
@@ -52,8 +47,9 @@ def trace(a, axis1=0, axis2=1):
 
     # Bring the trace axes forward.
     if (axis1, axis2) != (0, 1):
-        perm = [axis1, axis2] + \
-               [i for i in range(B.rank(a)) if i != axis1 and i != axis2]
+        perm = [axis1, axis2] + [
+            i for i in range(B.rank(a)) if i != axis1 and i != axis2
+        ]
         a = anp.transpose(a, axes=perm)
 
     return anp.trace(a)
@@ -110,10 +106,9 @@ def cholesky_solve(a, b):
 @dispatch(Numeric, Numeric)
 def triangular_solve(a, b, lower_a=True):
     def _triangular_solve(a_, b_):
-        return asla.solve_triangular(a_, b_,
-                                     trans='N',
-                                     lower=lower_a,
-                                     check_finite=False)
+        return asla.solve_triangular(
+            a_, b_, trans="N", lower=lower_a, check_finite=False
+        )
 
     return batch_computation(_triangular_solve, (a, b), (2, 2))
 

lab/autograd/shaping.py

@@ -30,7 +30,7 @@ def diag(a):
 @dispatch(Numeric)
 def vec_to_tril(a, offset=0):
     if B.rank(a) != 1:
-        raise ValueError('Input must be rank 1.')
+        raise ValueError("Input must be rank 1.")
     side, upper, perm = _vec_to_tril_shape_upper_perm(a, offset=offset)
     a = anp.concatenate((a, anp.zeros(upper, dtype=a.dtype)))
     return anp.reshape(a[perm], (side, side))
@@ -39,16 +39,16 @@ def vec_to_tril(a, offset=0):
 @dispatch(Numeric)
 def tril_to_vec(a, offset=0):
     if B.rank(a) != 2:
-        raise ValueError('Input must be rank 2.')
+        raise ValueError("Input must be rank 2.")
     n, m = B.shape(a)
     if n != m:
-        raise ValueError('Input must be square.')
+        raise ValueError("Input must be square.")
     return a[anp.tril_indices(n, k=offset)]
 
 
 @dispatch([Numeric])
 def stack(*elements, **kw_args):
-    return anp.stack(elements, axis=kw_args.get('axis', 0))
+    return anp.stack(elements, axis=kw_args.get("axis", 0))
 
 
 @dispatch(Numeric)
@@ -64,7 +64,7 @@ def reshape(a, *shape):
 
 @dispatch([Numeric])
 def concat(*elements, **kw_args):
-    return anp.concatenate(elements, axis=kw_args.get('axis', 0))
+    return anp.concatenate(elements, axis=kw_args.get("axis", 0))
 
 
 @dispatch(Numeric, [Int])
@@ -75,7 +75,7 @@ def tile(a, *repeats):
 @dispatch(Numeric, object)
 def take(a, indices_or_mask, axis=0):
     if B.rank(indices_or_mask) != 1:
-        raise ValueError('Indices or mask must be rank 1.')
+        raise ValueError("Indices or mask must be rank 1.")
 
     # Put axis `axis` first.
     if axis > 0:

lab/bvn_cdf/bvn_cdf.pyx

@@ -7,12 +7,12 @@ import numpy as np
 cimport cython
 from cython.parallel import prange
 
-cdef extern from 'math.h' nogil:
+cdef extern from "math.h" nogil:
     double log(double x)
     double exp(double x)
     double sqrt(double x)
 
-cdef extern from './tvpack.h' nogil:
+cdef extern from "./tvpack.h" nogil:
     double phid_(double* x)
     double bvnd_(double* x, double* y, double* rho)
 

lab/custom.py

@@ -6,10 +6,16 @@
 # noinspection PyUnresolvedReferences
 from .bvn_cdf import bvn_cdf as bvn_cdf_, s_bvn_cdf
 
-__all__ = ['toeplitz_solve', 's_toeplitz_solve',
-           'bvn_cdf', 's_bvn_cdf',
-           'expm', 's_expm',
-           'logm', 's_logm']
+__all__ = [
+    "toeplitz_solve",
+    "s_toeplitz_solve",
+    "bvn_cdf",
+    "s_bvn_cdf",
+    "expm",
+    "s_expm",
+    "logm",
+    "s_logm",
+]
 
 log = logging.getLogger(__name__)
 
@@ -103,5 +109,6 @@ def logm(a):
 
 
 def s_logm(a):  # pragma: no cover
-    raise NotImplementedError('The derivative for the matrix logarithm is '
-                              'current not implemented.')
+    raise NotImplementedError(
+        "The derivative for the matrix logarithm is current not implemented."
+    )

lab/generic.py

@@ -11,62 +11,64 @@
     AGNumeric,
     TFNumeric,
     TorchNumeric,
-    JaxNumeric
+    JaxNumeric,
 )
 from .util import abstract
 
-__all__ = ['nan',
-           'pi',
-           'log_2_pi',
-           'isnan',
-           'zeros',
-           'ones',
-           'zero',
-           'one',
-           'eye',
-           'linspace',
-           'range',
-           'cast',
-           'identity',
-           'negative',
-           'abs',
-           'sign',
-           'sqrt',
-           'exp',
-           'log',
-           'sin',
-           'cos',
-           'tan',
-           'tanh',
-           'erf',
-           'sigmoid',
-           'softplus',
-           'relu',
-           'add',
-           'subtract',
-           'multiply',
-           'divide',
-           'power',
-           'minimum',
-           'maximum',
-           'leaky_relu',
-           'min',
-           'max',
-           'sum',
-           'mean',
-           'std',
-           'logsumexp',
-           'all',
-           'any',
-           'lt',
-           'le',
-           'gt',
-           'ge',
-           'bvn_cdf',
-           'scan',
-           'sort',
-           'argsort',
-           'to_numpy']
+__all__ = [
+    "nan",
+    "pi",
+    "log_2_pi",
+    "isnan",
+    "zeros",
+    "ones",
+    "zero",
+    "one",
+    "eye",
+    "linspace",
+    "range",
+    "cast",
+    "identity",
+    "negative",
+    "abs",
+    "sign",
+    "sqrt",
+    "exp",
+    "log",
+    "sin",
+    "cos",
+    "tan",
+    "tanh",
+    "erf",
+    "sigmoid",
+    "softplus",
+    "relu",
+    "add",
+    "subtract",
+    "multiply",
+    "divide",
+    "power",
+    "minimum",
+    "maximum",
+    "leaky_relu",
+    "min",
+    "max",
+    "sum",
+    "mean",
+    "std",
+    "logsumexp",
+    "all",
+    "any",
+    "lt",
+    "le",
+    "gt",
+    "ge",
+    "bvn_cdf",
+    "scan",
+    "sort",
+    "argsort",
+    "to_numpy",
+]
 
 _dispatch = Dispatcher()
 
@@ -105,8 +107,7 @@ def zeros(dtype, *shape):  # pragma: no cover
     """
 
 
-@dispatch.multi((Int,),  # Single integer is not a reference.
-                ([Int],))
+@dispatch.multi((Int,), ([Int],))  # Single integer is not a reference.
 def zeros(*shape):
     return zeros(B.default_dtype, *shape)
 
@@ -133,8 +134,7 @@ def ones(dtype, *shape):  # pragma: no cover
     """
 
 
-@dispatch.multi((Int,),  # Single integer is not a reference.
-                ([Int],))
+@dispatch.multi((Int,), ([Int],))  # Single integer is not a reference.
 def ones(*shape):
     return ones(B.default_dtype, *shape)
 
@@ -297,6 +297,7 @@ def cast(dtype, a):  # pragma: no cover
 
 # Unary functions:
 
+
 @dispatch(Numeric)
 @abstract()
 def identity(a):  # pragma: no cover
@@ -812,7 +813,7 @@ def bvn_cdf(a, b, c):
 
 @_dispatch(object)
 def _as_tuple(x):
-    return x,
+    return (x,)
 
 
 @_dispatch(tuple)
@@ -849,8 +850,10 @@ def scan(f, xs, *init_state):
 
         # Check that the state shape remained constant.
         if new_state_shape != state_shape:
-            raise RuntimeError('Shape of state changed from {} to {}.'
-                               ''.format(state_shape, new_state_shape))
+            raise RuntimeError(
+                "Shape of state changed from {} to {}."
+                "".format(state_shape, new_state_shape)
+            )
 
         # Record the state, stacked over the various elements.
         states.append(B.stack(*state, axis=0))

lab/jax/custom.py

@@ -8,7 +8,7 @@
 from . import B
 from ..util import as_tuple
 
-__all__ = ['jax_register']
+__all__ = ["jax_register"]
 _dispatch = Dispatcher()
 
 

lab/jax/linear_algebra.py

@@ -5,11 +5,7 @@
 
 from . import dispatch, B, Numeric
 from .custom import jax_register
-from ..custom import (
-    toeplitz_solve, s_toeplitz_solve,
-    expm, s_expm,
-    logm, s_logm
-)
+from ..custom import toeplitz_solve, s_toeplitz_solve, expm, s_expm, logm, s_logm
 from ..linear_algebra import _default_perm
 from ..util import batch_computation
 
@@ -94,10 +90,9 @@ def cholesky_solve(a, b):
 @dispatch(Numeric, Numeric)
 def triangular_solve(a, b, lower_a=True):
     def _triangular_solve(a_, b_):
-        return jsla.solve_triangular(a_, b_,
-                                     trans='N',
-                                     lower=lower_a,
-                                     check_finite=False)
+        return jsla.solve_triangular(
+            a_, b_, trans="N", lower=lower_a, check_finite=False
+        )
 
     return batch_computation(_triangular_solve, (a, b), (2, 2))