formatting and linting passing

pyciemss/interfaces.py

@@ -122,9 +122,9 @@ def ensemble_sample(
  check_solver(solver_method, solver_options)
 
  # Get tolerances for solver
- rtol = solver_options.pop('rtol', 1e-7) # default = 1e-7
- atol = solver_options.pop('atol', 1e-9) # default = 1e-9
- 
+ rtol = solver_options.pop("rtol", 1e-7) # default = 1e-7
+ atol = solver_options.pop("atol", 1e-9) # default = 1e-9
+
  with torch.no_grad():
  if dirichlet_alpha is None:
  dirichlet_alpha = torch.ones(len(model_paths_or_jsons))
@@ -285,10 +285,10 @@ def ensemble_calibrate(
  # Check that num_iterations is a positive integer
  if not (isinstance(num_iterations, int) and num_iterations > 0):
  raise ValueError("num_iterations must be a positive integer")
- 
+
  # Get tolerances for solver
- rtol = solver_options.pop('rtol', 1e-7) # default = 1e-7
- atol = solver_options.pop('atol', 1e-9) # default = 1e-9
+ rtol = solver_options.pop("rtol", 1e-7) # default = 1e-7
+ atol = solver_options.pop("atol", 1e-9) # default = 1e-9
 
  def autoguide(model):
  guide = pyro.infer.autoguide.AutoGuideList(model)
@@ -396,7 +396,8 @@ def sample(
  - If performance is incredibly slow, we suggest using `euler` to debug.
  If using `euler` results in faster simulation, the issue is likely that the model is stiff.
  solver_options: Dict[str, Any]
- - Options to pass to the solver (including atol and rtol). See torchdiffeq' `odeint` method for more details.
+ - Options to pass to the solver (including atol and rtol).
+ See torchdiffeq' `odeint` method for more details.
  start_time: float
  - The start time of the model. This is used to align the `start_state` from the
  AMR model with the simulation timepoints.
@@ -462,8 +463,8 @@ def sample(
  check_solver(solver_method, solver_options)
 
  # Get tolerances for solver
- rtol = solver_options.pop('rtol', 1e-7) # default = 1e-7
- atol = solver_options.pop('atol', 1e-9) # default = 1e-9
+ rtol = solver_options.pop("rtol", 1e-7) # default = 1e-7
+ atol = solver_options.pop("atol", 1e-9) # default = 1e-9
 
  with torch.no_grad():
  model = CompiledDynamics.load(model_path_or_json)
@@ -620,7 +621,8 @@ def calibrate(
  - If performance is incredibly slow, we suggest using `euler` to debug.
  If using `euler` results in faster simulation, the issue is likely that the model is stiff.
  - solver_options: Dict[str, Any]
- - Options to pass to the solver (including atol and rtol). See torchdiffeq' `odeint` method for more details.
+ - Options to pass to the solver (including atol and rtol).
+ See torchdiffeq' `odeint` method for more details.
  - start_time: float
  - The start time of the model. This is used to align the `start_state` from the
  AMR model with the simulation timepoints.
@@ -687,8 +689,8 @@ def calibrate(
  check_solver(solver_method, solver_options)
 
  # Get tolerances for solver
- rtol = solver_options.pop('rtol', 1e-7) # default = 1e-7
- atol = solver_options.pop('atol', 1e-9) # default = 1e-9
+ rtol = solver_options.pop("rtol", 1e-7) # default = 1e-7
+ atol = solver_options.pop("atol", 1e-9) # default = 1e-9
 
  pyro.clear_param_store()
 
@@ -858,7 +860,8 @@ def optimize(
  - If performance is incredibly slow, we suggest using `euler` to debug.
  If using `euler` results in faster simulation, the issue is likely that the model is stiff.
  solver_options: Dict[str, Any]
- - Options to pass to the solver (including atol and rtol). See torchdiffeq' `odeint` method for more details.
+ - Options to pass to the solver (including atol and rtol).
+ See torchdiffeq' `odeint` method for more details.
  start_time: float
  - The start time of the model. This is used to align the `start_state` from the
  AMR model with the simulation timepoints.

pyciemss/ouu/ouu.py

@@ -97,8 +97,8 @@ def __init__(
  self.u_bounds = u_bounds
  self.risk_bound = risk_bound # used for defining penalty
  warnings.simplefilter("always", UserWarning)
- self.rtol = self.solver_options.pop('rtol', 1e-7) # default = 1e-7
- self.atol = self.solver_options.pop('atol', 1e-9) # default = 1e-9
+ self.rtol = self.solver_options.pop("rtol", 1e-7) # default = 1e-7
+ self.atol = self.solver_options.pop("atol", 1e-9) # default = 1e-9
 
  def __call__(self, x):
  if np.any(x - self.u_bounds[0, :] < 0.0) or np.any(

tests/test_interfaces.py

@@ -118,7 +118,7 @@ def test_sample_no_interventions(
  logging_step_size,
  num_samples,
  start_time=start_time,
- solver_options = {'rtol':rtol, 'atol':atol}
+ solver_options={"rtol": rtol, "atol": atol},
  )["unprocessed_result"]
  with pyro.poutine.seed(rng_seed=0):
  result2 = sample_method(
@@ -127,7 +127,7 @@ def test_sample_no_interventions(
  logging_step_size,
  num_samples,
  start_time=start_time,
- solver_options = {'rtol':rtol, 'atol':atol}
+ solver_options={"rtol": rtol, "atol": atol},
  )["unprocessed_result"]
 
  result3 = sample_method(
@@ -136,7 +136,7 @@ def test_sample_no_interventions(
  logging_step_size,
  num_samples,
  start_time=start_time,
- solver_options = {'rtol':rtol, 'atol':atol}
+ solver_options={"rtol": rtol, "atol": atol},
  )["unprocessed_result"]
 
  for result in [result1, result2, result3]:
@@ -410,7 +410,7 @@ def test_calibrate_deterministic(
  "data_mapping": model_fixture.data_mapping,
  "start_time": start_time,
  "deterministic_learnable_parameters": deterministic_learnable_parameters,
- "solver_options": {'rtol':rtol, 'atol':atol},
+ "solver_options": {"rtol": rtol, "atol": atol},
  **CALIBRATE_KWARGS,
  }
 
@@ -433,7 +433,7 @@ def test_calibrate_deterministic(
  *sample_args,
  **sample_kwargs,
  inferred_parameters=inferred_parameters,
- solver_options={'rtol':rtol, 'atol':atol}
+ solver_options={"rtol": rtol, "atol": atol},
  )["unprocessed_result"]
 
  check_result_sizes(result, start_time, end_time, logging_step_size, 1)