Question about intent to truncate simulation result

[htakagishi]

Hi @himoto
I would like to know the intention of saving the simulation results in the following code, where values less than sys.float_info.epsilon are considered as 0.
In my model, the values of molecular concentration tend to be very small, with some molecules appearing with values around 1e-10. Among them, especially the smaller values are considered zero by _preprocessing. However, if I remove _preprocessing from the program and keep the values as they are, I get ideal results for simulation dynamics.
What is the problem with deleting this _preprocessing?

biomass/biomass/dynamics/signaling_systems.py

Lines 105 to 116 in 2ad16db

	def _save_simulations(self, viz_type: str, simulated_values: np.ndarray) -> None:
	"""
	Save time course simulated values to simulation_data/.
	"""
	np.save(
	os.path.join(
	self.model.path,
	"simulation_data",
	"simulations_{}.npy".format(viz_type if simulated_values.ndim < 4 else "all"),
	),
	self._preprocessing(simulated_values),
	)

[himoto]

When solving ODE numerically, sometimes we get infinitesimal value as a solution, e.g., 1e-308, although it should be zero.
This causes a problem when normalizing simulation results and _preprocessing can help avoid such problems.
It converts values to zeros only when all time-course values are smaller than sys.float_info.epsilon. Even if you delete this process and get something "ideal", the solution might be very sensitive to small changes in parameter values, and therefore I strongly recommend that you do not delete the process.

[htakagishi]

So models that rely too much on small parameter changes should be avoided. I understand. Thank you very much.