How much do massless test particles increase integration overhead?

[void4]

Theoretically, increasing the number of massless particles should only increase the computational overhead linearly (and not quadratically), since they do not affect each other. Is this true for REBOUND when adding massless particles?

[hannorein]

The answer depends on various settings, so without more details I cannot provide any an answer.

[void4]

Could you mention which settings/sim.add() parameters may influence this? I'm very new to REBOUND and have not found an explanation of this in the documentation. I'm basically following these tutorials:

https://rebound.readthedocs.io/en/latest/ipython_examples/Horizons/
https://rebound.readthedocs.io/en/latest/ipython_examples/OrbitalElements/

, adding major bodies via JPL and a bunch of minor bodies by supplying orbital parameters to sim.add(), supplying no m(ass) parameter (data via https://ssd.jpl.nasa.gov/tools/sbdb_query.html tables), checking for close approaches to major bodies (non-Earth impact events do not seem to currently well monitored by anyone).

Code is still a fresh mess, but here: https://github.com/void4/impactprediction/blob/main/main.py

Naively, I'd expect adding objects with mass to increase the computation time quadratically (?), and ideally massless objects to only increase the runtime linearly (in each integration step, first computing interactions of objects with mass, and then for each massless object the gravitational pull to them)

[hannorein]

You might want to read the tutorial on test particles:
https://rebound.readthedocs.io/en/latest/ipython_examples/Testparticles/

Make sure you set N_active.

[void4]

Ah, very nice, this is exactly what I needed - thank you very much :) Closed!

Edit: I just saw the 3d web interface - so cool! Awesome work!