Issue #50 resolved

Time to improve capabilities of the public version?

Christoph Bischko
created an issue

Seeing it has been more than two years since the public release of GIZMO, I would very much like for it to be extended with a little bit more physics. At least cooling and star formation would be a nice addition to the public version, since GADGET-2's public version includes both.

Sadly, small groups don't have access to a more modern SPH-based simulation code. AREPO does not and, according to V. Springel, will not have a public release any time soon and GADGET-2 suffers from problems that led to AREPO and GIZMO in the first place. I feel a slightly more capable public version would only have benefits.

I'm also interested in the reasons, why GIZMO's public is hydro- and gravity-only. Hope you can help me out there and maybe consider a more capable public version.

The documentation states:

(1) that some of these features are designed for very specific problems, and not designed to be used as a ‘black box’ (i.e. they are not designed for a public code, to be compiled in general ways for general problems, etc), and (2) many have been developed not by me, but by other collaborators and peers, and they would prefer these modules are not made public at this time. If you want to use some of these physics, please contact me directly.

1) As I understand it, the code is modular. Would it not suffice, to turn more involved modules off by default 2) The star formation recipe of GADGET-2 is widely used and public, same holds for the cooling code. So this should not be an issue 3) I can see how support requests and sub-par might be a problem for a more useful public version. Is this another reason for not including SF, cooling and other widely tested, unproblematic modules?

Comments (6)

  1. Philip Hopkins repo owner

    Thank you, Christoph.

    Actually, neither the cooling nor star formation routines in GADGET are public, nor have they ever been. Unfortunately they have been widely distributed, in many cases against Volker's wishes (or without his approval).

    The way they are currently written, there is no way to make the cooling and/or star formation modules public and actually useable without making a large number of additional code modules and physics public as well, almost all of which are proprietary and developed by students and postdocs actively working on projects. Of course people are welcome to develop on the public version, and the private version can be used on a per-project discretionary basis with permissions of the relevant module authors (many of whom are not me).

    The public version of the GIZMO code includes multiple hydrodynamics solvers, arbitrary external and self-gravity with adaptive gravitational softening, magnetohydrodynamics (also with multiple solvers), support for multiple fluids, shearing boxes, integration of aerodynamic particles and direct (particle-in-cell-type) Lorentz force integration, as well as a large number of numerical optimizations such as hybrid openmp+mpi parallelization. Many of these have been added in releases since the first public version of the code. Updates planned for the near future (next ~few months) include anisotropic conduction (with and without Spitzer conductivities), anisotropic viscosity (with and without Braginskii viscosities), subgrid-scale eddy diffusion models for energy/momentum/passive scalars, and turbulent driving routines. Updates planned for the ~year timescale include full support for multiple radiation-hydrodynamics solvers. The intention is that the public code support a wide variety of fundamental physics as a gravity + fluid/plasma physics platform.

  2. Christoph Bischko reporter

    Thank you for taking the time to answer. I did not expect such a good and comprehensive response. I'm glad to hear the public version will be extended in capabilities.

    The only question I have left is, what does it take to be granted access to the star formation and cooling modules of GIZMO? What sort of requirements do you impose on possible users? Our group has worked with Volker and used GADGET-2 with these modules in the past. We are currently working on a comparison study of ram-pressure stripping with observers from Italy, where GADGET-2 as proven outdated. We would very much like to make the switch.

    I'm only asking publicly here, because this information could proof useful to students or other small groups. Maybe you could elaborate a bit on which modules are rather unproblematic to share, which fall under more rigid care by their respective authors and what your minimum requirements are to use them in publications.

  3. Log in to comment