IsoMemoApp Local version crashes with large datasets

[CarloCocozza]

For the attached dataset (more than 26k rows), using the local version of IsoMemoApp via Docker (both Beta and normal versions), the app crashes when running an AverageR model. This does not happen with the online app.

I have used the following options (excluding default settings): Dependent variable: O18_Ann_Val Uncertainty(optional) of dep. var.: O18_Ann_Err Long and Lat

Number of basis functions: 150 Extrapolation behaviour: Constant Bayesian Model Number of MCMC iterations: 3000 Number of burnin iterations: 500

@arunge

North america.xlsx

[arunge]

I could reproduce the issue, but unfortunately there was no error message:

So currently, I have no idea what could be the reason here. I will make some test runs and return to you if I have something new.

[arunge]

@CarloCocozza I tried to run the model again, but this time I ensured that the screen was not locked after some time of absence. The app did not crash, neither when running in docker nor when running with my local R installation. I tried this several times, but even with the default setup the model finished running.

However, the reasons are unclear right now:

• maybe resources are reduced when the screen got's locked (this depends on system configuration)
• maybe the calculation is such demanding that resources are near the limit and it is a bit of luck, if the process finishes
• maybe time-triggered updates were running on the laptop that led to restarts of required processes which led the container crash (Time-triggered updates are quite common for laptops. This is one reason, why long calculations should run on a server.)

To investigate the reason, we need to go deeper and check if we can find some error message in a log file (e.g. some out-of-memory error).

I will return to you if I know more.