Driven supersonic turbulence

The turbulent driving module implements an Ornstein–Uhlenbeck stirring pattern (Federrath; see Price & Federrath 2010, Tricco, Price & Federrath 2016). The algorithms and tests are described in the Phantom code paper (Section 2.4); this page explains how to run turbulence calculations.

Which SETUP to use

All three compile with PERIODIC=yes and DRIVING=yes (stirring compiled in).

Basic hydro turbulence run

Create a run directory outside the code tree:

mkdir -p ~/runs/turb
cd ~/runs/turb
~/phantom/scripts/writemake.sh turb > Makefile
make setup
make

Run the setup utility (interactive prompts, then edit turb.setup if needed):

./phantomsetup turb

Default geometry is a unit periodic cube with a cubic particle lattice (ilattice=1; close-packed ilattice=2 is closer to a relaxed distribution). Default units mimic a molecular cloud (3 pc box, low density, 0.2 km/s sound speed); see setup_turb.f90 for defaults.

Run the simulation:

./phantom turb.in

MHD turbulence

The turb preset does not enable MHD by default. Rebuild with:

make MHD=yes
make setup MHD=yes
make MHD=yes

Then run phantomsetup again. Set Bz_0 in turb.setup for a uniform field in \(z\). The setup prints MHD turbulence w/uniform field in z-direction when MHD is active.

Dusty turbulence

Use the pre-cooked dusty preset:

~/phantom/scripts/writemake.sh dustyturb > Makefile
make setup
make
./phantomsetup turb

This uses setup_turb.f90 with DUST=yes and the dust-as-mixture method (see Dust). Configure grain bins in turb.setup as for other dusty setups.

Stirring parameters (`.in` file)

Driving is controlled at runtime when DRIVING was enabled at compile time. Key options (see also the code paper Table of parameters, Section 2.4):

A forcing.dat file may be written/read to continue an identical forcing pattern across restarts.

Suggested starting values are written into a fresh turb.in by phantomsetup (e.g. tmax, dtmax, nfulldump for several outputs per crossing time). Tune st_energy and st_decay together: the module sets st_OUvar = sqrt(st_energy/st_decay).

Resolution and timesteps

Use hfact and kernel choice as in A brief guide to Phantom runtime options (the paper recommends quintic kernels for some dusty tests; default turb uses the setup default kernel).
C_cour and C_force default to 0.3 and 0.25 (paper test suite values).
Individual timesteps are enabled in the turb preset (IND_TIMESTEPS=yes).

Analysis and comparison

Visualise dumps with splash.
Measure \(\nabla\cdot\mathbf{v}\) and \(\nabla\times\mathbf{v}\) with phantom2divv (see List of Phantom utilities).
For standardised MHD benchmark problems from the code paper (e.g. Orszag–Tang), see the external phantom-examples repository (also mentioned in Getting started on Rusty (Flatiron cluster)).