Steady State Result Changes Depending on the Loading Method

[Cal Davies]

Hi all,

I have a really strange case where my model is changing the result of the steady state analysis by how I load the model into COPASI. Opening the model .cps file directly will always result in a steady state found at 1e-9 resolution. If I open COPASI and then open the .cps file I sometimes will get a steady state analysis that reports that no steady state can be found (but is found at 1e-8). I cannot reproduce this as it happens randomly. However, if I open a different model and then open the .cps file in COPASI, it will always result in a steady state not being found (but again is found at 1e-8).  This is on the latest version of COPASI and using Newton + Integration method.

Furthermore, I exported the model to every version of SBML available. Each version was randomly deciding whether to find a steady state or not, at 1e-9. after reloading the models and running again. The only explanation I can think of is COPASI is storing intermediate data between each model execution/loading which could be interfering with the steady state analysis but I'm not sure whether COPASI does that. Though it is strange that every attempt works at 1e-8.

Any advice to resolve this issue would be greatly appreciated.

Many thanks,

Callum