import numpy as np
import gillespy2
from gillespy2.solvers.cpp.c_decoder import IterativeSimDecoder
from gillespy2.solvers.utilities import solverutils as cutils
from gillespy2.core import GillesPySolver, Model, Event, RateRule
from gillespy2.core.gillespyError import *
from typing import Union
from typing import Set
from enum import IntEnum
from gillespy2.core import Results
from .c_solver import CSolver, SimulationReturnCode
from gillespy2.solvers.cpp.build.template_gen import SanitizedModel
[docs]class TauHybridCSolver(GillesPySolver, CSolver):
"""
This solver uses a root-finding interpretation of the direct SSA method,
along with ODE solvers to simulate ODE and Stochastic systems
interchangeably or simultaneously.
"""
name = "TauHybridCSolver"
target = "hybrid"
def __init__(self, model = None, output_directory = None, delete_directory = True, resume=None, variable = False, constant_tau_stepsize=None):
self.constant_tau_stepsize = constant_tau_stepsize
super().__init__(model=model, output_directory=output_directory,
delete_directory=delete_directory, resume=resume, variable=variable)
[docs] class ErrorStatus(IntEnum):
UNKNOWN = 1
LOOP_OVER_INTEGRATE = 2
INTEGRATOR_FAILED = 3
INVALID_AFTER_SSA = 4
NEGATIVE_STATE_NO_SSA_REACTION = 5
NEGATIVE_STATE_AT_BEGINING_OF_STEP = 6
def __create_options(self, sanitized_model: "SanitizedModel") -> "SanitizedModel":
"""
Populate the given list of species modes into a set of template macro definitions.
Generated options are specific to the Tau Hybrid solver,
and get passed as custom definitions to the build engine.
:param sanitized_model: Sanitized model containing sanitized species definitions.
The GPY_HYBRID_SPECIES_MODES option will be set as an option for the model.
:type sanitized_model: SanitizedModel
:returns: Pass-through of sanitized model object.
:rtype: SanitizedModel
"""
species_mode_map = {
"continuous": "CONTINUOUS_MODE",
"discrete": "DISCRETE_MODE",
"dynamic": "DYNAMIC_MODE",
}
boundary_condition_types = [
# When species.boundary_condition == False
"STANDARD",
# When species.boundary_condition == True
"BOUNDARY",
]
trigger_mode_types = [
# When event.use_values_from_trigger_time == False
"USE_EVAL",
# When event.use_values_from_trigger_time == True
"USE_TRIGGER",
]
persist_types = [
# When event.trigger.persistent == False
"IRREGULAR",
# When event.trigger.persistent == True
"PERSISTENT",
]
initial_value_types = [
# When event.trigger.initial_value == False
"INIT_FALSE",
# When event.trigger.initial_value == True
"INIT_TRUE",
]
species_mode_list = []
for spec_id, species in enumerate(sanitized_model.species.values()):
mode_keyword = species_mode_map.get(species.mode, species_mode_map["dynamic"])
# Casting a bool to an int evaluates: False -> 0, and True -> 1
# Explicit cast to bool for safety, in case boundary_condition is given weird values
boundary_keyword = boundary_condition_types[int(bool(species.boundary_condition))]
# Example: SPECIES_MODE(2, 10, CONTINUOUS_MODE, BOUNDARY)
entry = f"SPECIES_MODE({spec_id},{species.switch_min},{species.switch_tol},{mode_keyword},{boundary_keyword})"
species_mode_list.append(entry)
# EVENT(event_id, {targets}, trigger, delay, priority, use_trigger, use_persist)
event_list = []
# [SPECIES/VARIABLE]_ASSIGNMENT(assign_id, target_id, expr)
event_assignment_list = []
assign_id = 0
for event_id, event in enumerate(sanitized_model.model.listOfEvents.values()):
trigger = sanitized_model.expr.getexpr_cpp(event.trigger.expression)
delay = sanitized_model.expr.getexpr_cpp(event.delay) \
if event.delay is not None else "0"
priority = sanitized_model.expr.getexpr_cpp(event.priority) \
if event.priority is not None else "0"
use_trigger = trigger_mode_types[int(bool(event.use_values_from_trigger_time))]
use_persist = persist_types[int(bool(event.trigger.persistent))]
initial_value = initial_value_types[int(bool(event.trigger.value or False))]
assignments: "list[str]" = []
for assign in event.assignments:
variable = assign.variable
expression = sanitized_model.expr.getexpr_cpp(assign.expression)
if isinstance(variable, str):
if variable in sanitized_model.model.listOfSpecies:
variable = sanitized_model.model.listOfSpecies.get(variable)
elif variable in sanitized_model.model.listOfParameters:
variable = sanitized_model.model.listOfParameters.get(variable)
else:
raise ValueError(f"Error in event={event} "
f"Invalid event assignment {assign}: received name {variable} "
f"Must match the name of a valid Species or Parameter.")
if isinstance(variable, gillespy2.Species):
assign_str = f"SPECIES_ASSIGNMENT(" \
f"{assign_id},{sanitized_model.species_id.get(variable.name)},{expression})"
elif isinstance(variable, gillespy2.Parameter):
assign_str = f"VARIABLE_ASSIGNMENT(" \
f"{assign_id},{sanitized_model.parameter_id.get(variable.name)},{expression})"
else:
raise ValueError(f"Invalid event assignment {assign}: received variable of type {type(variable)} "
f"Must be of type str, Species, or Parameter")
assignments.append(str(assign_id))
event_assignment_list.append(assign_str)
assign_id += 1
# Check for "None"s
for a in assignments:
if a is None: raise Exception(f"assignment={a} is None in event={event}")
if event_id is None: raise Exception(f"event_id is None in event={event}")
if trigger is None: raise Exception(f"trigger is None in event={event}")
if delay is None: raise Exception(f"delay is None in event={event}")
if priority is None: raise Exception(f"priority is None in event={event}")
if use_trigger is None: raise Exception(f"use_trigger is None in event={event}")
if use_persist is None: raise Exception(f"use_persist is None in event={event}")
if initial_value is None: raise Exception(f"initial_value is None in event={event}")
assignments: "str" = " AND ".join(assignments)
event_list.append(
f"EVENT("
f"{event_id},"
f"{{{assignments}}},"
f"{trigger},"
f"{delay},"
f"{priority},"
f"{use_trigger},"
f"{use_persist},"
f"{initial_value}"
f")"
)
sanitized_model.options["GPY_HYBRID_SPECIES_MODES"] = " ".join(species_mode_list)
sanitized_model.options["GPY_HYBRID_EVENTS"] = " ".join(event_list)
sanitized_model.options["GPY_HYBRID_NUM_EVENTS"] = str(len(event_list))
sanitized_model.options["GPY_HYBRID_EVENT_ASSIGNMENTS"] = " ".join(event_assignment_list)
sanitized_model.options["GPY_HYBRID_NUM_EVENT_ASSIGNMENTS"] = str(len(event_assignment_list))
return sanitized_model
[docs] @classmethod
def get_supported_features(cls):
return {
Event,
RateRule,
}
[docs] @classmethod
def get_supported_integrator_options(cls) -> "Set[str]":
return { "rtol", "atol", "max_step" }
def _build(self, model: "Union[Model, SanitizedModel]", simulation_name: str, variable: bool, debug: bool = False,
custom_definitions=None) -> str:
variable = variable or len(model.listOfEvents) > 0
sanitized_model = self.__create_options(SanitizedModel(model, variable=variable))
for rate_rule in model.listOfRateRules.values():
sanitized_model.use_rate_rule(rate_rule)
# determine if a constant stepsize has been requested
if self.constant_tau_stepsize is not None:
sanitized_model.options['GPY_CONSTANT_TAU_STEPSIZE'] = str(float(self.constant_tau_stepsize))
else:
sanitized_model.options['GPY_CONSTANT_TAU_STEPSIZE'] = '0'
return super()._build(sanitized_model, simulation_name, variable, debug)
def _handle_return_code(self, return_code: "int") -> "int":
if return_code == TauHybridCSolver.ErrorStatus.UNKNOWN:
raise SimulationError("C++ solver failed (no error code given).")
if return_code == TauHybridCSolver.ErrorStatus.LOOP_OVER_INTEGRATE:
raise SimulationError("Loop over integrate exceeded, problem space is too stiff")
if return_code == TauHybridCSolver.ErrorStatus.INTEGRATOR_FAILED:
raise SimulationError("Sundials ODE solver failed with 'BAD_STEP_SIZE'")
if return_code == TauHybridCSolver.ErrorStatus.INVALID_AFTER_SSA:
raise SimulationError("Invalid state after single SSA step")
if return_code == TauHybridCSolver.ErrorStatus.NEGATIVE_STATE_NO_SSA_REACTION:
raise SimulationError("Negative State detected in step, and no reaction found to fire.")
if return_code == TauHybridCSolver.ErrorStatus.NEGATIVE_STATE_AT_BEGINING_OF_STEP:
raise SimulationError("Negative State detected at beginning of step. Species involved in reactions can not be negative.")
return super()._handle_return_code(return_code)
[docs] @classmethod
def get_solver_settings(cls):
"""
Returns a list of arguments supported by tau_hybrid_c_solver.run.
:return: Tuple of strings, denoting all keyword argument for this solvers run() method.
:rtype: tuple
"""
return ('model', 't', 'number_of_trajectories', 'timeout', 'increment', 'seed', 'debug', 'profile')
[docs] def run(self=None, model: Model = None, t: int = None, number_of_trajectories: int = 1, timeout: int = 0,
increment: int = None, seed: int = None, debug: bool = False, profile: bool = False, variables={},
resume=None, live_output: str = None, live_output_options: dict = {}, tau_step: int = .03, tau_tol=0.03, integrator_options: "dict[str, float]" = None, use_root_finding=False, **kwargs):
"""
:param model: The model on which the solver will operate. (Deprecated)
:type model: gillespy2.Model
:param t: End time of simulation.
:type t: int
:param number_of_trajectories: Number of trajectories to simulate. By default number_of_trajectories = 1.
:type number_of_trajectories: int
:param timeout: If set, if simulation takes longer than timeout, will exit.
:type timeout: int
:param increment: Time step increment for plotting.
:type increment: float
:param seed: The random seed for the simulation. Optional, defaults to None.
:type seed: int
:param variables: Dictionary of species and their data that will override existing species data.
:type variables: dict
:param resume: Result of a previously run simulation, to be resumed.
:type resume: gillespy2.Results
:param live_output: The type of output to be displayed by solver. Can be "progress", "text", or "graph".
:type live_output: str
:param live_output_options: dictionary contains options for live_output. By default {"interval":1}.
"interval" specifies seconds between displaying.
"clear_output" specifies if display should be refreshed with each display.
:type live_output_options: dict
:param tau_tol: Tolerance level for Tau leaping algorithm. Larger tolerance values will
result in larger tau steps. Default value is 0.03.
:type tau_tol: float
:returns: A result object containing the results of the simulation.
:rtype: gillespy2.Results
"""
from gillespy2 import log
if self is None:
# Post deprecation block
# raise SimulationError("TauHybridCSolver must be instantiated to run the simulation")
# Pre deprecation block
log.warning(
"""
`gillespy2.Model.run(solver=TauHybridCSolver)` is deprecated.
You should use `gillespy2.Model.run(solver=TauHybridCSolver(model=gillespy2.Model))
Future releases of GillesPy2 may not support this feature.
"""
)
self = TauHybridCSolver(model, resume=resume)
if model is not None:
log.warning('model = gillespy2.model is deprecated. Future releases '
'of GillesPy2 may not support this feature.')
if self.model is None:
if model is None:
raise SimulationError("A model is required to run the simulation.")
self._set_model(model=model)
self.model.compile_prep()
self.validate_model(self.model, model)
self.validate_sbml_features(model=self.model)
self.validate_tspan(increment=increment, t=t)
if increment is None:
increment = self.model.tspan[-1] - self.model.tspan[-2]
if t is None:
t = self.model.tspan[-1]
# Validate parameters prior to running the model.
self._validate_type(variables, dict, "'variables' argument must be a dictionary.")
self._validate_variables_in_set(variables, self.species + self.parameters)
self._validate_resume(t, resume)
self._validate_kwargs(**kwargs)
if resume is not None:
t = abs(t - int(resume["time"][-1]))
number_timesteps = int(round(t / increment + 1))
args = {
"trajectories": number_of_trajectories,
"timesteps": number_timesteps,
"tau_tol": tau_tol,
"end": t,
"interval": str(number_timesteps),
}
if self.variable:
populations = cutils.update_species_init_values(self.model.listOfSpecies, self.species, variables, resume)
parameter_values = cutils.change_param_values(self.model.listOfParameters, self.parameters, self.model.volume, variables)
args.update({
"init_pop": populations,
"parameters": parameter_values
})
if integrator_options is not None:
integrator_options = TauHybridCSolver.validate_integrator_options(integrator_options)
args.update(integrator_options)
seed = self._validate_seed(seed)
if seed is not None:
args.update({
"seed": seed
})
if live_output is not None:
live_output_options['type'] = live_output
display_args = {
"model": self.model, "number_of_trajectories": number_of_trajectories, "timeline": np.linspace(0, t, number_timesteps),
"live_output_options": live_output_options, "resume": bool(resume)
}
else:
display_args = None
args = self._make_args(args)
if debug:
args.append("--verbose")
if use_root_finding:
args.append("--use_root_finding")
decoder = IterativeSimDecoder.create_default(number_of_trajectories, number_timesteps, len(self.model.listOfSpecies))
sim_exec = self._build(self.model, self.target, self.variable, False)
sim_status = self._run(sim_exec, args, decoder, timeout, display_args)
trajectories, time_stopped = decoder.get_output()
simulation_data = self._format_output(trajectories)
if sim_status == SimulationReturnCode.PAUSED:
simulation_data = self._make_resume_data(time_stopped, simulation_data, t)
if resume is not None:
simulation_data = self._update_resume_data(resume, simulation_data, time_stopped)
self.result = simulation_data
self.rc = int(sim_status)
return Results.build_from_solver_results(self, live_output_options)
