Re: cooling, heating and temperature oscillation

From: David Riethmiller <rieth_at_phy.ohiou.edu>
Date: Mon, 20 Aug 2012 14:25:46 -0400

Dear Gadget list,

I'm wondering if anyone has had success implementing the Exact
Integration scheme described in Townsend 2009 ApJS 181:391 into
Gadget. I know that Gadget cooling can be unstable unless an implicit
cooling scheme is used, but I wonder if this newer algorithm is easier
to use, and might avoid the iterative scheme necessary for implicit
time-integration?

Thanks,
Dave


-------------------------------------------------
David A. Riethmiller
Ph.D. Candidate, Astrophysical Institute
Ohio University



> From: Volker Springel <volker_at_MPA-Garching.MPG.DE>
> Date: Fri, 30 Mar 2007 12:13:44 +0200
>
> Hi Yves,
>
> You need to solve the cooling part in the energy equation with an
> implicit time-integration scheme to cure this problem.
>
> For example, the new energy (or entropy) at the end of the timestep
> can
> be computed as
>
> u' = u + (du/dt)_visc * dt + (du/dt)'_cool * dt
>
> Here (du/dt)'_cool is the cooling rate evaluated for the *new*
> temperature u'. It's a bit cumbersome to solve this equation (will
> require some iterative scheme), but then your oscillations should be
> gone.
>
> Volker
Received on 2012-08-20 20:25:52