User-defined close encounter/colision/ejection condition. about rebound HOT 3 CLOSED

Hi Gabriel,

I would not use a collision detection routine for this. I would just periodically check if the particles have swapped their position. You can do that while the simulation is running using a loop, or afterwards using the snapshots saved in the simulationarchive. Unless you check for this condition very often, it should not be a limiting factor for the runtime. You only get the time right approximately, but maybe that's good enough. You can always go back and check more often.

As a side note, when you calculate the semi-major axes of all particles, you probably want to use heliocentric coordinates rather than Jacobi coordinates.

I hope this helps.
Hanno

from rebound.

Hi, Hanno,

Thanks for the suggestions! I think that checking for around 50.000 output points (worst case scenario, since collisions can be detected earlier) won't be much a of a problem.

For larger integration times, the resolution at the lower end might be a little poor (on the order of 1e3 years, which is around 1e4 orbits of my innermost planet), but it's not an absolute problem for statistical purposes, I guess.

I will definitely pay more attention and make more use of the SimulationArchive, it's much better and easier than having to filter out which variables and analysis to make beforehand.

I will also keep in mind the change of coordinates and make some experiment with it! Thanks.

I will close it for now!

Kind regards,
Gabriel.

from rebound.

"Manually" check for collision condition after each timestep seems to be simpler for now. CLosing it!

from rebound.