Re: Phantom as a N-body tool?
13 views
Skip to first unread message
Daniel Price
Apr 18, 2018, 5:41:10 AM4/18/18
to Cristian Beauge, Nicolas Cuello, Phantom users
Dear Christian & Nicolás,
(cc to phantom users list)
We’ve done this in various ways. Phantom has both collisional and collisionless N-body dynamics so depends what you’re after. Sink particles are collisional, i.e. the force is N^2 and unsoftened by default. To do collisionless dynamics the best way is to just set up dust particles with no gas (but you need to compile with DUST=yes). A second approach we have used is to set pressure equal to zero on gas particles, which can be achieved with ieos=11.
The time integration is only second order, but it is a symplectic integrator and the tolerance on the timestep is tight. You can see the relevant N-body tests in the phantom paper for details, we get similar accuracy to fourth order integrators but just with more steps. For very long term evolution you would want to check things like angular momentum and energy conservation to be sure they are preserved to good accuracy over the lifetime of the calculation. If they were drifting too much you could simply drop the C_force factor to give a more accurate time integration.
Does that help?
Daniel
> On 18 Apr 2018, at 1:33 am, Cristian Beauge <bea...@oac.unc.edu.ar> wrote:
>
> Hi,
>
> Sorry to barge into the mail, but just to add a couple of details.
> The idea was whether PHANTOM could be used as a very fast purely gravitational
> N-body code, to use mainly on two problems: ring dynamics and planetary formation.
>
> In both cases we would need to include the gravitational interactions between the bodies,
> so perhaps using the gas instead of the dust population would be better.
>
> Thanks!
>
> Cristian
>
> On Tue, Apr 17, 2018 at 12:24 PM, Nicolas Cuello <cuello...@gmail.com> wrote:
> Hi Daniel,
>
> After the short Phantom tutorial at the Observatorio Astronómico de Córdoba (Argentina), we had a chat with Cristian Beaugé (in cc) about potential applications. Dynamicists liked Phantom very much :-) One of the final question was: is it possible to use Phantom as an N-body integrator? I guess that the easiest way to do that would be to initialize both gas and dust disc turning off the drag. Then we just follow the dust without considering the gas. That would be pure N-body without dust-dust interactions, isn’t it? If I am not mistaken, Cristian is after long term evolution and resonances in discs (hence he needs self-gravity) and he wants to see if Phantom is accurate and fast for this kind of problems.
>
> To sum up, Cristian would like to compare an SPH-approach, O(N*N_neigh), to a N-body method where you sum over the entire tree, O(N^2). The idea is the following: simulate the solids as a gaseous fluid without the term of artificial viscosity and turning the self-gravity on. Does this make any sense? Do you expect any numerical issues if we set alpha_SPH and beta_SPH to 0?
>
> Thanks in advance for the feedback!
> Cheers,
> Nicolás
>
(cc to phantom users list)
We’ve done this in various ways. Phantom has both collisional and collisionless N-body dynamics so depends what you’re after. Sink particles are collisional, i.e. the force is N^2 and unsoftened by default. To do collisionless dynamics the best way is to just set up dust particles with no gas (but you need to compile with DUST=yes). A second approach we have used is to set pressure equal to zero on gas particles, which can be achieved with ieos=11.
The time integration is only second order, but it is a symplectic integrator and the tolerance on the timestep is tight. You can see the relevant N-body tests in the phantom paper for details, we get similar accuracy to fourth order integrators but just with more steps. For very long term evolution you would want to check things like angular momentum and energy conservation to be sure they are preserved to good accuracy over the lifetime of the calculation. If they were drifting too much you could simply drop the C_force factor to give a more accurate time integration.
Does that help?
Daniel
> On 18 Apr 2018, at 1:33 am, Cristian Beauge <bea...@oac.unc.edu.ar> wrote:
>
> Hi,
>
> Sorry to barge into the mail, but just to add a couple of details.
> The idea was whether PHANTOM could be used as a very fast purely gravitational
> N-body code, to use mainly on two problems: ring dynamics and planetary formation.
>
> In both cases we would need to include the gravitational interactions between the bodies,
> so perhaps using the gas instead of the dust population would be better.
>
> Thanks!
>
> Cristian
>
> On Tue, Apr 17, 2018 at 12:24 PM, Nicolas Cuello <cuello...@gmail.com> wrote:
> Hi Daniel,
>
> After the short Phantom tutorial at the Observatorio Astronómico de Córdoba (Argentina), we had a chat with Cristian Beaugé (in cc) about potential applications. Dynamicists liked Phantom very much :-) One of the final question was: is it possible to use Phantom as an N-body integrator? I guess that the easiest way to do that would be to initialize both gas and dust disc turning off the drag. Then we just follow the dust without considering the gas. That would be pure N-body without dust-dust interactions, isn’t it? If I am not mistaken, Cristian is after long term evolution and resonances in discs (hence he needs self-gravity) and he wants to see if Phantom is accurate and fast for this kind of problems.
>
> To sum up, Cristian would like to compare an SPH-approach, O(N*N_neigh), to a N-body method where you sum over the entire tree, O(N^2). The idea is the following: simulate the solids as a gaseous fluid without the term of artificial viscosity and turning the self-gravity on. Does this make any sense? Do you expect any numerical issues if we set alpha_SPH and beta_SPH to 0?
>
> Thanks in advance for the feedback!
> Cheers,
> Nicolás
>
Reply all
Reply to author
Forward
0 new messages