search

neurophysik / jitcdde

Just-in-time compilation for delay differential equations
Other
57 stars 15 forks source link

Distributed delays? #2

Closed uluzair closed 7 years ago

uluzair commented 7 years ago

Great project, I am excited to see the final release!

By the way, do you plan to implement also (continuously) distributed delays?

Wrzlprmft commented 7 years ago

By the way, do you plan to implement also (continuously) distributed delays?

Depending on what exactly you mean by this, this is already implemented.

For example, if you want a network where the delays between nodes are drawn from some distribution, but do not change over time, you can do this analogously to this example for JiTCODE. Instead of using ω as a regular control parameter, you use it as the delay.

In general, if you can write down the DDE on paper using delay terms of the form yi(t-τ), you can integrate the system with JiTCDDE.

uluzair commented 7 years ago

All right, so the delays need to be discretized first. I guess this is straight-forward with systems as for example

where f is a positive function vanishing at infinity.

(just edited the example so it makes more sense)

Wrzlprmft commented 7 years ago

All right, so the delays need to be discretized first. I guess this is straight-forward with systems as for example

where f is a positive function vanishing at infinity.

Yes, if you want to integrate such a system, you would have to discretise the integral beforehand. A numerical integrator would have to do something similar anyway, unless you want to integrate symbolically at each step – which would cost a lot of time and is probably very nasty to implement.

Wrzlprmft commented 6 years ago

FYI: I now implemented a utility function for integrals. More on Issue #11.