whfast safe mode with external precession?

[nickchoksi]

Quick question. I've got an external precession force in my setup, implemented via modify-orbits-direct. I'm not otherwise accessing the particle properties or using them in any way, except for writing output files. Can I turn off WHFast's safe mode?

Thanks!
Nick

[hannorein]

That would formally lower the order of the scheme (see the REBOUNDx paper: https://arxiv.org/abs/1908.05634). Does it matter in practice? Probably not. But I would keep the safe mode turned on. I just don't think the <2x speedup is worth the trouble.

cc @dtamayo

[nickchoksi]

Thanks. The integrations I'm trying to do are long (~1e8 orbits). So order-unity speedups are still pretty helpful if possible...

[dtamayo]

I agree it would basically look like a first-order mapping, except your outputs would be at the wrong phase along the orbit (shifted by the half Kepler steps on the end). If you're only interested in the other orbital elements that might be OK, but I would certainly want to run some test to make sure I trust the result.

I completely agree with Hanno that the extra human effort in terms of running tests to make sure things are working as expected, and explaining to someone else what it is you're doing in the paper (which is complicated!) is not worth the extra CPU time of doing the simple thing and waiting twice as long.

[hannorein]

Only occurred to me now, but if you calculate orbital elements, then you need positions and velocities which will be offset if the safemode is off. This could be a systematic error.