0

I would like to use JiTCDDE to solve system of DDE which has a past derivative ẏ(t−τ) on the RHS.

Should I use the get_state function or is there a more suitable way?

Wrzlprmft
  • 4,234
  • 1
  • 28
  • 54
Nikita
  • 1
  • 1

1 Answers1

2

When a DDE depends on a past derivative, it becomes a so-called neutral DDE. In such DDEs discontinuities in the initial condition do not loose their degree over time (the process is neutral with respect to the degree of the discontinuity). Such neutral DDEs usually require special solvers, which JiTCDDE doesn’t support – since they require a lot of extra work and are not the kind of application JiTCDDE is aimed at.

For this reason, there currently is no way to access the past derivative in a dynamic equation. (I am the author of the package, and I would know if I had implemented this.)

There is now an experimental feature that allows you to access the past derivative, using dy. Take a look at this example on how to use it.

Wrzlprmft
  • 4,234
  • 1
  • 28
  • 54
  • That being said, it is actually not so difficult to implement this, and it can work with the current solver if you can avoid initial discontinuities altogether, i.e., you can design your initial past such that the derivative at the start of integration matches the derivative of your past at that point. If you think you can, feel free to [file a feature request](https://github.com/neurophysik/jitcdde/issues) with your actual equations and the expected outcome so I have something to work with. It may take me a few weeks until I can have a look at this though. – Wrzlprmft Oct 13 '19 at 07:42
  • Thank you for an answer. I will think it over. – Nikita Oct 13 '19 at 07:45