add option to disable Newton fallback via set_options method (#1702)

This commit is contained in:
quant9 2020-10-08 14:13:21 -04:00 committed by GitHub
parent 77da5b8c06
commit e41ed25646
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -21,6 +21,15 @@ class IStanBackend(ABC):
def __init__(self):
self.model = self.load_model()
self.stan_fit = None
self.newton_fallback = True
def set_options(self, **kwargs):
"""
Specify model options as kwargs.
* newton_fallback [bool]: whether to fallback to Newton if L-BFGS fails
"""
if 'newton_fallback' in kwargs:
self.newton_fallback = kwargs['newton_fallback']
@staticmethod
@abstractmethod
@ -58,7 +67,8 @@ class CmdStanPyBackend(IStanBackend):
model_name = 'prophet.stan'
target_name = 'prophet_model.bin'
sm = cmdstanpy.CmdStanModel(stan_file=os.path.join(model_dir, model_name))
sm = cmdstanpy.CmdStanModel(
stan_file=os.path.join(model_dir, model_name))
sm.compile()
copy(sm.exe_file, os.path.join(target_dir, target_name))
@ -77,24 +87,25 @@ class CmdStanPyBackend(IStanBackend):
iterations = int(1e4)
try:
self.stan_fit = self.model.optimize(data=stan_data,
inits=stan_init,
iter=iterations,
**kwargs)
inits=stan_init,
iter=iterations,
**kwargs)
except RuntimeError as e:
# Fall back on Newton
if kwargs['algorithm'] != 'Newton':
if self.newton_fallback and kwargs['algorithm'] != 'Newton':
logger.warning(
'Optimization terminated abnormally. Falling back to Newton.'
)
kwargs['algorithm'] = 'Newton'
self.stan_fit = self.model.optimize(data=stan_data,
inits=stan_init,
iter=iterations,
**kwargs)
inits=stan_init,
iter=iterations,
**kwargs)
else:
raise e
params = self.stan_to_dict_numpy(self.stan_fit.column_names, self.stan_fit.optimized_params_np)
params = self.stan_to_dict_numpy(
self.stan_fit.column_names, self.stan_fit.optimized_params_np)
for par in params:
params[par] = params[par].reshape((1, -1))
return params
@ -109,9 +120,9 @@ class CmdStanPyBackend(IStanBackend):
kwargs['iter_warmup'] = iter_half
self.stan_fit = self.model.sample(data=stan_data,
inits=stan_init,
iter_sampling=iter_half,
**kwargs)
inits=stan_init,
iter_sampling=iter_half,
**kwargs)
res = self.stan_fit.sample
(samples, c, columns) = res.shape
res = res.reshape((samples * c, columns))
@ -243,13 +254,16 @@ class PyStanBackend(IStanBackend):
args.update(kwargs)
try:
self.stan_fit = self.model.optimizing(**args)
except RuntimeError:
except RuntimeError as e:
# Fall back on Newton
logger.warning(
'Optimization terminated abnormally. Falling back to Newton.'
)
args['algorithm'] = 'Newton'
self.stan_fit = self.model.optimizing(**args)
if self.newton_fallback and args['algorithm'] != 'Newton':
logger.warning(
'Optimization terminated abnormally. Falling back to Newton.'
)
args['algorithm'] = 'Newton'
self.stan_fit = self.model.optimizing(**args)
else:
raise e
params = dict()
@ -277,4 +291,4 @@ class StanBackendEnum(Enum):
try:
return StanBackendEnum[name].value
except KeyError as e:
raise ValueError("Unknown stan backend: {}".format(name)) from e
raise ValueError("Unknown stan backend: {}".format(name)) from e