Bump pinned black version

.github/workflows/lint.yml

@@ -28,7 +28,7 @@ jobs:
       - name: Black Code Formatter
         uses: psf/black@stable
         with:
-          version: "21.6b0"
+          version: "22.3"
 
       - name: isort Import Sorter
         uses: isort/[email protected]

dev_requirements.txt

@@ -2,6 +2,6 @@ pylint
 pytest
 pytest-runner
 pre-commit
-black>=21.6b0,<22
+black>=22.3,<23
 isort
 autoflake

examples/scripts/deconv_ppp_bm3d_pgm.py

@@ -43,7 +43,7 @@
 n = 5  # convolution kernel size
 σ = 20.0 / 255  # noise level
 
-psf = snp.ones((n, n)) / n ** 2
+psf = snp.ones((n, n)) / n**2
 A = linop.Convolve(h=psf, input_shape=x_gt.shape)
 
 Ax = A(x_gt)  # blurred image

examples/scripts/deconv_ppp_bm4d_admm.py

@@ -48,7 +48,7 @@
 n = 5  # convolution kernel size
 σ = 20.0 / 255  # noise level
 
-psf = snp.ones((n, n, n)) / (n ** 3)
+psf = snp.ones((n, n, n)) / (n**3)
 A = linop.Convolve(h=psf, input_shape=x_gt.shape)
 
 Ax = A(x_gt)  # blurred image

examples/scripts/denoise_tv_iso_pgm.py

@@ -140,7 +140,7 @@ def prox(self, v: JaxArray, lam: float, **kwargs) -> JaxArray:
 solver_iso = AcceleratedPGM(
     f=f_iso,
     g=g_iso,
-    L0=16.0 * f_iso.lmbda ** 2,
+    L0=16.0 * f_iso.lmbda**2,
     x0=x0,
     maxiter=100,
     itstat_options={"display": True, "period": 10},
@@ -188,7 +188,7 @@ def prox(self, v: JaxArray, lam: float, **kwargs) -> JaxArray:
 solver = AcceleratedPGM(
     f=f,
     g=g,
-    L0=16.0 * f.lmbda ** 2,
+    L0=16.0 * f.lmbda**2,
     x0=x0,
     maxiter=100,
     itstat_options={"display": True, "period": 10},

scico/blockarray.py

@@ -922,11 +922,11 @@ def __rfloordiv__(a, b):
 
     @_block_array_binary_op_wrapper
     def __pow__(a, b):
-        return a ** b
+        return a**b
 
     @_block_array_binary_op_wrapper
     def __rpow__(a, b):
-        return b ** a
+        return b**a
 
     @_block_array_binary_op_wrapper
     def __gt__(a, b):

scico/examples.py

@@ -333,9 +333,9 @@ def create_3D_foam_phantom(
     im = snp.zeros(im_shape) + c_lo
     for c, r in zip(centers, radii):
         dist = snp.sum((x - c) ** 2, axis=-1)
-        if snp.mean(im[dist < r ** 2] - c_lo) < 0.01 * c_hi:
+        if snp.mean(im[dist < r**2] - c_lo) < 0.01 * c_hi:
             # In numpy: im[dist < r**2] = c_hi
-            im = im.at[dist < r ** 2].set(c_hi)
+            im = im.at[dist < r**2].set(c_hi)
 
     return im
 

scico/functional/_norm.py

@@ -286,7 +286,7 @@ def __init__(self, delta: float = 1.0):
             delta: Huber function parameter :math:`\delta`.
         """
         self.delta = delta
-        self._call_lt_branch = lambda xl2: 0.5 * xl2 ** 2
+        self._call_lt_branch = lambda xl2: 0.5 * xl2**2
         self._call_gt_branch = lambda xl2: self.delta * xl2 - 0.5
         super().__init__()
 

scico/linop/_linop.py

@@ -337,7 +337,7 @@ def __init__(
     __class__ = OpClass  # needed for super() to work
 
     OpClass.__doc__ = f"Linear operator version of :func:`{f_name}`."
-    OpClass.__init__.__doc__ = fr"""
+    OpClass.__init__.__doc__ = rf"""
 
         Args:
             input_shape: Shape of input array.

scico/linop/optics.py

@@ -235,7 +235,7 @@ def __init__(
         )
 
         self.phase = jax.device_put(
-            np.exp(1j * z * sqrt(self.k0 ** 2 - self.kp ** 2)).astype(np.complex64)
+            np.exp(1j * z * sqrt(self.k0**2 - self.kp**2)).astype(np.complex64)
         )
         self.D = Diagonal(self.phase)
         self._set_adjoint()
@@ -259,7 +259,7 @@ def adequate_sampling(self):
         """
         tmp = []
         for d, N in zip(self.dx, self.padded_shape):
-            tmp.append(d ** 2 > np.pi / (self.k0 * N) * np.sqrt(d ** 2 * N ** 2 + 4 * self.z ** 2))
+            tmp.append(d**2 > np.pi / (self.k0 * N) * np.sqrt(d**2 * N**2 + 4 * self.z**2))
         return np.all(tmp)
 
     def pinv(self, y):
@@ -317,7 +317,7 @@ def __init__(
         )
 
         self.phase = jax.device_put(
-            np.exp(1j * z * (self.k0 - self.kp ** 2 / (2 * self.k0))).astype(np.complex64)
+            np.exp(1j * z * (self.k0 - self.kp**2 / (2 * self.k0))).astype(np.complex64)
         )
         self.D = Diagonal(self.phase)
 
@@ -342,7 +342,7 @@ def adequate_sampling(self):
         """
         tmp = []
         for d, N in zip(self.dx, self.padded_shape):
-            tmp.append(d ** 2 > 2 * np.pi * self.z / (self.k0 * N))
+            tmp.append(d**2 > 2 * np.pi * self.z / (self.k0 * N))
         return np.all(tmp)
 
 
@@ -464,7 +464,7 @@ def __init__(
         elif ndim == 2:
             self.r2 = np.sqrt(x_D[0][:, None] ** 2 + x_D[1][None, :] ** 2)
 
-        phase = -1j * snp.exp(1j * k0 * z) * snp.exp(1j * 0.5 * k0 / z * self.r2 ** 2)
+        phase = -1j * snp.exp(1j * k0 * z) * snp.exp(1j * 0.5 * k0 / z * self.r2**2)
         phase *= k0 / (2 * np.pi) * np.abs(1 / z)
         phase *= np.prod(dx)  # from approximating continouous FT with DFT
         phase = phase.astype(np.complex64)
@@ -515,5 +515,5 @@ def adequate_sampling(self):
         """
         tmp = []
         for d, N in zip(self.dx, self.input_shape):
-            tmp.append(d ** 2 > 2 * np.pi * self.z / (self.k0 * N))
+            tmp.append(d**2 > 2 * np.pi * self.z / (self.k0 * N))
         return np.all(tmp)

scico/loss.py

@@ -403,12 +403,12 @@ def _dep_cubic_root(p, q):
     (see Sec. 3.C of :cite:`soulez-2016-proximity`).
     """
     q2 = q / 2
-    Δ = q2 ** 2 + (p / 3) ** 3
+    Δ = q2**2 + (p / 3) ** 3
     Δrt = snp.sqrt(Δ + 0j)
     u3, v3 = -q2 + Δrt, -q2 - Δrt
     u, v = _cbrt(u3), _cbrt(v3)
     r = (u + v).real
-    assert snp.allclose(snp.abs(r ** 3 + p * r + q), 0, atol=1e-4)
+    assert snp.allclose(snp.abs(r**3 + p * r + q), 0, atol=1e-4)
     return r
 
 

scico/metric.py

@@ -89,7 +89,7 @@ def psnr(
     if signal_range is None:
         signal_range = snp.abs(snp.max(reference) - snp.min(reference))
     with np.errstate(divide="ignore"):
-        rt = signal_range ** 2 / mse(reference, comparison)
+        rt = signal_range**2 / mse(reference, comparison)
     return 10.0 * snp.log10(rt)
 
 

scico/optimize/pgm.py

@@ -619,5 +619,5 @@ def step(self):
 
             self.fixed_point_residual = snp.linalg.norm(self.x - self.v)
             t_old = self.t
-            self.t = 0.5 * (1 + snp.sqrt(1 + 4 * t_old ** 2))
+            self.t = 0.5 * (1 + snp.sqrt(1 + 4 * t_old**2))
             self.v = self.x + ((t_old - 1) / self.t) * (self.x - x_old)

scico/test/linop/test_abel.py

@@ -15,7 +15,7 @@
 def make_im(Nx, Ny):
     x, y = snp.meshgrid(snp.linspace(-1, 1, Nx), snp.linspace(-1, 1, Ny))
 
-    im = snp.where(x ** 2 + y ** 2 < 0.3, 1.0, 0.0)
+    im = snp.where(x**2 + y**2 < 0.3, 1.0, 0.0)
 
     return im
 

scico/test/linop/test_optics.py

@@ -76,7 +76,7 @@ def test_fresnel_sampling(ndim):
     N = 128
     dx = 1
     k0 = 1
-    A = FresnelPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=N ** 2)
+    A = FresnelPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=N**2)
     assert not A.adequate_sampling()
     A = FresnelPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=1)
     assert A.adequate_sampling()
@@ -87,7 +87,7 @@ def test_fraunhofer_sampling(ndim):
     N = 128
     dx = 1
     k0 = 1
-    A = FraunhoferPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=N ** 2)
+    A = FraunhoferPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=N**2)
     assert not A.adequate_sampling()
     A = FraunhoferPropagator(input_shape=(N,) * ndim, dx=dx, k0=k0, z=1)
     assert A.adequate_sampling()

scico/test/linop/test_radon_svmbir.py

@@ -33,7 +33,7 @@
 def make_im(Nx, Ny, is_3d=True):
     x, y = snp.meshgrid(snp.linspace(-1, 1, Nx), snp.linspace(-1, 1, Ny))
 
-    im = snp.where((x - 0.25) ** 2 / 3 + y ** 2 < 0.1, 1.0, 0.0)
+    im = snp.where((x - 0.25) ** 2 / 3 + y**2 < 0.1, 1.0, 0.0)
     if is_3d:
         im = im[snp.newaxis, :, :]
     im = im.astype(snp.float32)

scico/test/optimize/test_admm.py

@@ -288,7 +288,7 @@ def setup_method(self, method):
         Nx = 8
         x = np.pad(np.ones((Nx, Nx), dtype=np.float32), Nx)
         Npsf = 3
-        psf = snp.ones((Npsf, Npsf), dtype=np.float32) / (Npsf ** 2)
+        psf = snp.ones((Npsf, Npsf), dtype=np.float32) / (Npsf**2)
         self.A = linop.CircularConvolve(
             h=psf,
             input_shape=x.shape,

scico/test/test_metric.py

@@ -17,7 +17,7 @@ def test_mae_mse(self):
         e1 = metric.mae(x, y)
         e2 = metric.mse(x, y)
         assert np.abs(e1 - xe / N) < 1e-12
-        assert np.abs(e2 - (xe ** 2) / N) < 1e-12
+        assert np.abs(e2 - (xe**2) / N) < 1e-12
 
     def test_snr_nrm(self):
         N = 16

scico/test/test_operator.py

@@ -24,12 +24,12 @@ def _eval(self, x):
 
 class SquareOperator(Operator):
     def _eval(self, x):
-        return x ** 2
+        return x**2
 
 
 class SumSquareOperator(Operator):
     def _eval(self, x):
-        return snp.sum(x ** 2)
+        return snp.sum(x**2)
 
 
 class OperatorTestObj:

scico/test/test_ray_tune.py

@@ -16,7 +16,7 @@
 
 def eval_params(config, reporter):
     x, y = config["x"], config["y"]
-    cost = x ** 2 + (y - 0.5) ** 2
+    cost = x**2 + (y - 0.5) ** 2
     reporter(cost=cost)
 
 

scico/test/test_solver.py

@@ -207,22 +207,22 @@ def test_split_join_blockarray():
 
 
 def test_bisect():
-    f = lambda x: x ** 3
+    f = lambda x: x**3
     x, info = solver.bisect(f, -snp.ones((5, 1)), snp.ones((5, 1)), full_output=True)
     assert snp.sum(snp.abs(x)) == 0.0
     assert info["iter"] == 0
     x = solver.bisect(f, -2.0 * snp.ones((5, 3)), snp.ones((5, 3)), xtol=1e-5, ftol=1e-5)
     assert snp.max(snp.abs(x)) <= 1e-5
     assert snp.max(snp.abs(f(x))) <= 1e-5
     c, key = random.randn((5, 1), dtype=np.float32)
-    f = lambda x, c: x ** 3 - c ** 3
+    f = lambda x, c: x**3 - c**3
     x = solver.bisect(f, -snp.abs(c) - 1, snp.abs(c) + 1, args=(c,), xtol=1e-5, ftol=1e-5)
     assert snp.max(snp.abs(x - c)) <= 1e-5
     assert snp.max(snp.abs(f(x, c))) <= 1e-5
 
 
 def test_golden():
-    f = lambda x: x ** 2
+    f = lambda x: x**2
     x, info = solver.golden(f, -snp.ones((5, 1)), snp.ones((5, 1)), full_output=True)
     assert snp.max(snp.abs(x)) <= 1e-7
     x = solver.golden(f, -2.0 * snp.ones((5, 3)), snp.ones((5, 3)), xtol=1e-5)