Fix formatting

pyoptmat/chunktime.py

@@ -19,8 +19,9 @@
 from pyoptmat.utility import mbmm
 
 
-def newton_raphson_chunk(fn, x0, solver, rtol=1e-6, atol=1e-10, miter=100,
-        throw_on_fail = False):
+def newton_raphson_chunk(
+    fn, x0, solver, rtol=1e-6, atol=1e-10, miter=100, throw_on_fail=False
+):
     """
     Solve a nonlinear system with Newton's method with a tensor for a
     BackwardEuler type chunking operator context manager.
@@ -55,8 +56,7 @@ def newton_raphson_chunk(fn, x0, solver, rtol=1e-6, atol=1e-10, miter=100,
     if i == miter:
         if throw_on_fail:
             raise RuntimeError("Implicit solve did not succeed.")
-        else:
-            warnings.warn("Implicit solve did not succeed.  Results may be inaccurate...")
+        warnings.warn("Implicit solve did not succeed.  Results may be inaccurate...")
 
     return x
 

pyoptmat/hardening.py

@@ -701,6 +701,7 @@ def dhistory_rate_derate(self, s, h, t, ep, T, e):
         """
         return (self.C(T) - self.g(T) * h[..., 0] * torch.sign(ep))[..., None, None]
 
+
 class FAKinematicHardeningModelNoRecovery(KinematicHardeningModel):
     # pylint: disable=line-too-long
     """
@@ -810,9 +811,9 @@ def dhistory_rate_dhistory(self, s, h, t, ep, T, e):
         Returns:
           torch.tensor:       derivative with respect to history
         """
-        return (
-            -self.g(T) * torch.ones_like(h[..., 0]) * torch.abs(ep)
-        )[..., None, None]
+        return (-self.g(T) * torch.ones_like(h[..., 0]) * torch.abs(ep))[
+            ..., None, None
+        ]
 
     def dhistory_rate_derate(self, s, h, t, ep, T, e):
         """
@@ -832,6 +833,7 @@ def dhistory_rate_derate(self, s, h, t, ep, T, e):
         """
         return (self.C(T) - self.g(T) * h[..., 0] * torch.sign(ep))[..., None, None]
 
+
 class ChabocheHardeningModel(KinematicHardeningModel):
     # pylint: disable=line-too-long
     """

pyoptmat/ode.py

@@ -299,7 +299,7 @@ def __init__(
         direct_solve_min_size=0,
         adjoint_params=None,
         guess_type="zero",
-        throw_on_fail = False,
+        throw_on_fail=False,
         **kwargs,
     ):
         # Store basic info about the system

pyoptmat/optimize.py

@@ -189,7 +189,7 @@ class StatisticalModel(PyroModule):
                                     entry i represents the noise in test type i
     """
 
-    def __init__(self, maker, names, locs, scales, eps, nan_num = False):
+    def __init__(self, maker, names, locs, scales, eps, nan_num=False):
         super().__init__()
 
         self.maker = maker
@@ -242,7 +242,7 @@ def forward(self, exp_data, exp_cycles, exp_types, exp_control, exp_results=None
 
         # Setup the full noise, which can be type specific
         if self.type_noise:
-            full_noise = torch.empty(exp_data.shape[-1], device = self.eps.device)
+            full_noise = torch.empty(exp_data.shape[-1], device=self.eps.device)
             for i in experiments.exp_map.values():
                 full_noise[exp_types == i] = self.eps[i]
         else:
@@ -470,7 +470,8 @@ def guide(exp_data, exp_cycles, exp_types, exp_control, exp_results=None):
                     # Fix this to init to the mean (or a sample I guess)
                     ll_param = pyro.param(
                         name + self.param_suffix,
-                        val.detach().clone()
+                        val.detach()
+                        .clone()
                         .unsqueeze(0)
                         .repeat((exp_data.shape[2],) + (1,) * dim),
                     )
@@ -553,7 +554,7 @@ def _make_weight_tensor(self, exp_types):
         Assemble a full tensor for the data weights, based on the self.weights
         dictionary
         """
-        weights = torch.zeros(exp_types.shape, device = exp_types.device)
+        weights = torch.zeros(exp_types.shape, device=exp_types.device)
         for tt, v in self.weights.items():
             weights[exp_types == tt] = v
 

pyoptmat/scaling.py

@@ -208,6 +208,7 @@ def scale_mean(self, loc, scale):
         """
         return self.scale(loc)
 
+
 class LogBoundedScalingFunction(ScalingFunction):
     """
     Scaling function where the unscaled parameters are

test/test_hardening.py

@@ -222,7 +222,9 @@ class TestFAKinematicHardening(unittest.TestCase, HardeningBase):
     def setUp(self):
         self.C = torch.tensor(100.0)
         self.g = torch.tensor(1.2)
-        self.model = hardening.FAKinematicHardeningModelNoRecovery(CP(self.C), CP(self.g))
+        self.model = hardening.FAKinematicHardeningModelNoRecovery(
+            CP(self.C), CP(self.g)
+        )
 
         self.nbatch = 10
         self.bdim = 1
@@ -239,7 +241,9 @@ class TestFAKinematicHardeningMultiBatch(unittest.TestCase, HardeningBase):
     def setUp(self):
         self.C = torch.tensor(100.0)
         self.g = torch.tensor(1.2)
-        self.model = hardening.FAKinematicHardeningModelNoRecovery(CP(self.C), CP(self.g))
+        self.model = hardening.FAKinematicHardeningModelNoRecovery(
+            CP(self.C), CP(self.g)
+        )
 
         self.nbatch = 10
         self.bdim = 2