Why do I get an "Integration tolerance Not Met" in a SimBiology "Fit Data" task?

I am running a Fit Data task in Simbiology for a two step model:

L + A <---> LA

LA + BG <---> C

The reaction rate for the two reactions are respectively:

kon1*A*L - koff1*LA

kon2*LA*BG - koff2*C

I am using the concentration of C as my independent / response variable and time as my dependent / predictor variable to estimate the parameter "kon2"; the other three parameters are known.

Every time I run the simulation, I get the following error:

Integration Tolerance Not Met

I have tried changing i) the initial value of "kon2", ii) the tolerance and number of iterations, and iii) the solver used to estimate the parameter, to no avail.

What could be causing this error and how can I resolve this?

Answer 1

[MathWorks Support Team]

This error occurs because when solving an ODE numerically, the expected error at the next timestep is estimated, but the solver's tolerances are not sufficient to predict it.

For example, this error may occur because a sudden change occurs at some point during the simulation and the current solver's tolerances do not allow it to fall within the error range.

It may also occur when determining the time change of variables that have very large and small values in the same differential equation. This can be caused by the model's formula being stiff, which tends to diverge during numerical operations.

Possible countermeasures include reviewing the solver, loosening the tolerance, or turning off "AbsoluteToleranceScaling" in Simulation Settings.

For more information, please check the help documentation below.

・Troubleshooting Simulation Problems

https://www.mathworks.com/help/releases/R2023b/simbio/ug/troubleshooting-simulation-errors.html