Controller is not reset during simulation of patients about simglucose HOT 5 CLOSED

Yes, that is a real bug.

Need to reset it here

Thanks for the report.

Let me know if you are interested in a PR.

from simglucose.

I'm interested and will open a PR soon.

from simglucose.

Hi, could you explain a bit more about what this issue is? I've been using the batch simulation regularly, and would be interested in knowing if this affects any of my experiments. What happens when the controller does not reset?

from simglucose.

In my case I trained a neural network that used a continuous chain of states and stored the internal state of the neural network (LSTM; hidden state).
Due to the absence of the reset call, the hidden state was carried over from simulation to simulation and when a new simulation was started, a chain of states from the previous simulation + new states from the new simulation was used. This has led to a significant distortion of the results from the second simulation of a simulation batch onwards.
Depending on your experiment setup this could lead to distorted results too.

from simglucose.

This issue is going to be solved when this PR is merged #47 .

from simglucose.