I have a coupled ODE and the time series data that is representative of the ODE. I want to do a multi-step forecast. What I have a problem is coding or creating the logic flow in my code, since I have multiple input variables. For instance, I have A(t), B(t), C(t) which are fed into my neural network to predict C(t+1). If I do a recursive multi-step forecast, I have to update my C(t) input variable using the predicted output, but A(t) and B(t) stay the same. Can someone help give me a code function or snippet of a code script for this logic?
