Source code for brian2.core.core_preferences
"""
Definitions, documentation, default values and validation functions for core
Brian preferences.
"""
from numpy import float32, float64, int32
from brian2.core.preferences import BrianPreference, prefs
__all__ = []
[docs]
def dtype_repr(dtype):
return dtype.__name__
[docs]
def default_float_dtype_validator(dtype):
return dtype in [float32, float64]
[docs]
def outdated_dependency_error_validator(value):
if value is not None:
import warnings
warnings.warn(
"The 'core.outdated_dependency_error' preference is no longer "
"used and will be removed in a future version. Brian2 now "
"relies on package managers for dependency management.",
UserWarning,
stacklevel=2,
)
return True # We always accept the value user provides, but warn them
prefs.register_preferences(
"core",
"Core Brian preferences",
default_float_dtype=BrianPreference(
default=float64,
docs="""
Default dtype for all arrays of scalars (state variables, weights, etc.).
""",
representor=dtype_repr,
validator=default_float_dtype_validator,
),
default_integer_dtype=BrianPreference(
default=int32,
docs="""
Default dtype for all arrays of integer scalars.
""",
representor=dtype_repr,
),
outdated_dependency_error=BrianPreference(
default=None,
docs="""
**DEPRECATED**: This preference is no longer used. Brian2 now relies on
package managers (pip, conda) for dependency management instead of runtime
version checking. Setting this preference has no effect and it will be
removed in a future version.
Previously controlled whether to raise an error for outdated dependencies
(``True``) or just a warning (``False``).
""",
validator=outdated_dependency_error_validator,
),
stop_on_keyboard_interrupt=BrianPreference(
default=True,
docs="""
Whether to "gracefully" stop a simulation after pressing Ctrl+C (defaults to
``True``). Note that pressing Ctrl+C a second time will force the usual
interruption mechanism.
""",
),
)
prefs.register_preferences(
"legacy",
"Preferences to enable legacy behaviour",
refractory_timing=BrianPreference(
default=False,
docs="""
Whether to use the semantics for checking the refractoriness condition
that were in place up until (including) version 2.1.2. In that
implementation, refractory periods that were multiples of dt could lead
to a varying number of refractory timesteps due to the nature of
floating point comparisons). This preference is only provided for exact
reproducibility of previously obtained results, new simulations should
use the improved mechanism which uses a more robust mechanism to
convert refractoriness into timesteps. Defaults to ``False``.
""",
),
)