I tried to solve the three body problem numerically. While after 1000 time steps the system looks stable (See Fig.1) it is completely unstable after 10000 time steps (See Fig.2). Is that an intrinsic property of the three body problem or a mistake in my code or initial conditions? N=1000 N=10000

  • 1
    $\begingroup$ The IT-support question "did you try to switch it off and on again?" can be adjusted here for astrophysics, namely: "Did you try to make your timestep smaller?" $\endgroup$ Commented Jun 6, 2019 at 17:39
  • $\begingroup$ Agree w/ @AtmosphericPrisonEscape -- I've done similar things trying to match NASA's calculations, and, the smaller the timestamp, the closer the match. Using a timestep even a little too large results in chaos. $\endgroup$
    – user21
    Commented Jun 6, 2019 at 17:44
  • 1
    $\begingroup$ @StephenG Please link to abstracts instead of PDFs, e.g. Liao 2013. $\endgroup$
    – Mike G
    Commented Jun 6, 2019 at 19:52
  • 1
    $\begingroup$ @Mike Sorry about that. I had thought I'd pasted a link to the abstract, but clearly a mistake on my part. $\endgroup$ Commented Jun 6, 2019 at 20:00
  • 1
    $\begingroup$ If you provide some more detail about your simulation, it will be easier to help, but my guess is that you've used a too small gravity softening length (or maybe even none?). Then if particles come too close to each other, forces may blow up due to numerical inaccuracies. Note though that in principle there may be no error, since the three-body problem is not necessarily stable; one particle may be slingshot away, also in real life. $\endgroup$
    – pela
    Commented Jun 6, 2019 at 20:23


You must log in to answer this question.

Browse other questions tagged .