Hanno Rein

I haven't investigated this yet. Including the interruptible pool in REBOUND was always a bit of a hack. Did you try the multiprocess (without ing) module?

PSA: The bisection method used in the REBOUND example is a bit crude. I'm currently working with a summer student to hopefully come up with a slightly more elegant way...

This is tricky if there are properties. See https://stackoverflow.com/questions/15750522/class-properties-and-setattr

Have a look at the post at the very end of this thread #401. I think that should answer all your questions!

I see. Well, MERCURIUS is not automatically faster than IAS15. It depends on the system you're integrating, what you're interested in, what kind of accuracy you're aiming for, etc. There...

Feel free to reopen the issue again later. Especially if you can post a short example that I can run myself, then we can brainstorm on how to make it...

- Mercurius doesn't use Jacobi coordinates (they don't make much sense if particles have close encounters) but democratic heliocentric coordinates. - You do not need to recalculate coordinates manually when...

Glad to hear it's faster. Yes, regarding ias15.min_dt. You could estimate it by dividing the planet radius by its escape speed. But it depends what you want to do. You...

Hi Nick! The merge function doesn't do anything to dcrit. It's just the dcrit of one of the original particles (which one of the two is random, on purpose). That's...

I think you're correct with everything (slowdown is primarily due to the synchronization, ~25% sounds reasonable). But it's tricky. I'm not sure you really want to change dcrit every timestep....