Hanno Rein

Nice! I completely agree, type hints would be great. And adding it as a new package rather than breaking backwards compatibility seems like the right way to go for the...

Hm. I'm not very confident about the hash implementation for this use case. Is there an easy way to test if the problem occurs when you don't uses hashes? (e.g....

Fair enough. Another potential non-rebound issue: Could your random number generator generate the same number twice? If you call it in your initialization routine and then later in the collision...

If for some reason the same seed gets used twice, then you get the same hash and the probability is more like 1/500.

OK. That's not it then. Is this an unexpected side-effect due to #494? If so, commenting out this line should fix it: https://github.com/hannorein/rebound/blob/main/src/integrator_mercurius.c#L366

I think that makes sense. So maybe #494 should only be an option and turned off by default.

You're probably not the only one who might run into something related to this. So I'll make this feature optional and turn it off by default. Better to be slow...

Wow. What an active discussion! I love it! Everything seems clear to me now. It is related to #494. The new hashes of the massive particles get overwritten after the...

There are some instructions on how to install it on Windows here: https://rebound.readthedocs.io/en/latest/quickstart_installation/

You have to use Windows Subsystem for Linux (WSL).