Different integrators obtain different orbits

[ntsimon]

I try to use two integrators (whfast and ias15), to simulate a solar system with several planets and one comet for their orbits. However, the result shows that for whfast, the comet leaves the system while for ias15, it moves around the Sun. why? The other parameters such as time step remain unchanged
Any comments are welcome.
thanks.

Simon

[hannorein]

The two integrators are very different and are expected to give different results, especially if your system is chaotic and sensitive to initial conditions. A few things:

• Note that IAS15 is an adaptive method. You say you set the timestep. But the timestep will change unless you make it especially a non-adaptive method (which is of limited use).
• I assume your comet had orbit crossings with other planets. If you, you should either use IAS15, or WHFast with democratic heliocentric coordinates. The default WHFast setting is using Jacobi coordinates which is nor ideal for your setup.
• Regardless of what integrator you use, you should make sure your results are converged and do not depend on numerical fudge factors.

[ntsimon]

Thanks,

If all interested solar planets with a comet àre added onto the system, and just do the following two codes for changing to the center-of-mass frame and plotting the orbits without selecting integrators.

sim. move_to_com()
op=rebound. OrbitPlot(sim).

then the status shows IAS15 integrator and the orbits are displayed.

How do I know then whether the resulting orbits are accurate?

Is there any parameter setup in order to simulate much more periods?

Can the resulting raw data be extracted for references? If yes, how to get them?

Further advice are welcome, thank.

Simon.

[hannorein]

As mentioned above, regardless of what integrator you use, you should make sure your results are converged and do not depend on numerical fudge factors.