Pencil Code -- Tests    

Automatic test results

To ensure reproducability, the Pencil Code is tested daily for a number of sample applications. This is important for us in order to make sure certain improvements in some parts of the code do not affect the functionality of other parts. For other users who suspect that a new problem has emerged it could be useful to first see whether this problem also shows up in our own tests. The latest test results for a can be seen online:
  • opto3 (Linux on 4 x Opteron 2.2GB, ifort 9.1 compiler with MPICH, by Anders Johansen)
  • GNU Fortran (Ubuntu 4.4.1-4ubuntu9) 4.4.1 (by Philippe Bourdin)
  • Shal (Linux on 2 x Quadcore Intel Xeon E5320@1.86GHz, ifort 64 bits v10.0.023, by Boris Dintrans)
  • Nordita Weekend Test (norlx51, gfortran, openmpi, by Wolfgang/Axel)
  • Nordita PowerMac (os10, g95, ompi, by Axel) [previous]

    Note: before checking in your own changes, you should at least do the very minimal auto-test: 

      pc_auto-test --level=0 --no-pencil-check -C

    Results from tests

    samples/1d-tests

  • Sod shock tube tests (checked in under samples/1d-tests/sod_10 to sod_1000). Initial condition is a smoothed (width=0.03) isothermal pressure jump ranging from 10:1 to 1000:1.
    pressure jump 10:1 pressure jump 100:1 pressure jump 1000:1
    ν=.02, χ=.0005, t=2.7: ν=.04, χ=.0005, t=1.9: ν=.08, χ=.0005, t=1.5:
    The values of viscosity are chosen rather conservatively; for weak shocks one can get away with less viscosity (ν=.014 for 10:1 and ν=.028 for 100:1). For strong shocks (pressure jumps of 1000:1 and above) the discrepancy compared with the inviscid analytic solution becomes quite noticeable.

  • Rarefaction shocks (checked in under samples/1d-tests/expans, expans_bfield and riemann_bfield).
    no B-field with B-field shock with B-field
    cf. Fig. 1 of Falle (2002) cf. Fig. 2 of Falle (2002) cf. Fig. 6 of Falle (2002)
  • Conditions for non-magnetic rarefaction shock. Left state: ρ=1, p=10, ux=-3. Right state: ρ=.87469, p=8, ux=-2.46537. This corresponds to s/cp=1.68805 on both sides. 800 points. ν=0.05, χ=0.0002.
  • Conditions for magnetic rarefaction shock. Left state: ρ=1, p=.2327, ux=-4.6985, uy=-1.085146, Bx=-0.7, By=1.9680. Right state: ρ=.7270, p=.1368, ux=-4.0577, uy=-0.8349, Bx=-0.7, By=1.355. This corresponds to s/cp=-0.5682 on both sides. 800 points. ν=χ=η=0.07.
  • Conditions for magnetic Riemann problem. Left state: ρ=0.5, p=10, ux=0, uy=2, Bx=2, By=2.5. Right state: ρ=0.1, p=0.1, ux=-10, uy=0, Bx=2, By=2. This corresponds to s/cp=2.38119 on the left and 1.22753 on the right. 800 points. ν=χ=η=2.

  • Note: in the tests above, uniform viscosity is used. The viscosity has to be chosen such as to cope with the strongest compression in the domain. By using a nonuniform (artificial) viscosity, both compression and expansion shocks can be made as sharp as possible.

    samples/conv-slab-2d

    Vertical cross-section at t=920:
    K=0.008, ν=0.004

  • 2-D convection (checked in under samples/2d-tests/conv-slab-2d and conv-slab-2d2).
    no B-field with B-field
    (t=920) (t=320)

  • 2-D convection (checked in under samples/2d-tests/A3+chi11+Ra1e5).
    aspect ratio 3, density ratio 11, Ra=105
    (t=530, K=ν=0.0011, 150x51 points)

    samples/turbulence/helical-MHDturb32-4procs

    Low resolution helical MHD turbulence run, 2x2 processors, initial field: random
    t=200-1700 t=200-800 initial field: Beltrami
    full sequence(7.3Mb) every 2nd frame (1.4Mb) hires run: 5123 (1.1Mb)
    rot512_Om0a

    samples/turbulence/hydro512f

    A non-helical hydro turbulence run, 5123 meshpoints, 128 processors
    z-comp of velocity (4.7Mb)
    t=500-565    		 z-comp of velocity (6.4Mb)
    t=566-640    		 log of density: lnrho (4.1Mb)
    t=566-640

    samples/shearing-box/BH256_3D_mean_Bz=0b1

    Shearing box simulation, 2563 meshpoints, 32 processors
    z-comp of velocity (2.8Mb)
    t=296-393

    samples/interlocked-fluxrings

    Time evolution up to t=1.0:

    Isosurface of |B| at t=1.0:

    $Date: 2010-04-24 03:42:16 $