Date: Mon Aug 25 2003 - 13:34:53 CDT
Hi,
I have a problem (maybe two) with FLASH2.3, and I wonder whether someone of
you found and solved something similar.
I attach two files to this e-mail: they are taken from the output of my
simulation, made including the following modules:
INCLUDE database/amr/paramesh2.0
INCLUDE driver/time_dep
INCLUDE gravity/poisson/multipole
INCLUDE hydro/explicit/split/ppm
INCLUDE io/amr/hdf4
INCLUDE materials/composition/iso7
INCLUDE materials/eos/helmholtz
INCLUDE mesh/amr/paramesh2.0/quadratic_cartesian
INCLUDE solvers/linalg/ma28
INCLUDE solvers/poisson/multipole
INCLUDE source_terms/burn/iso7
INCLUDE util/initialization/1d
INCLUDE util/interp/polynomial
INCLUDE util/tools
INCLUDE util/wrapping
I start mapping in 2D a white dwarf, put in hydrostatic equilibrium with the
routine provided by FLASH.
The file supernova_hdf_chk_tot_vel0001.png shows the velocity plot of the
inner part of my WD, with the velocity vectors, after a few timesteps. As you
can see, there is the part closer to x=0 and y=0 (exactly two zones along the
axis) with a strange, larger velocity.
These are the features of this problem:
1) it deals only with the first two zones at x=0, y=0, without any dependence
on the resolution of my grid (I switched the AMR off);
2) it doesn't depend on the procedure for putting the initial model in
hydrostatic equilibrium;
3) it doesn't depend on boundaries (with a static gravity module that we
wrote, I tried to simulate the whole star not only a quadrant, and this
feature is still there at the center);
4) it doesn't depend on geometry (I tried cartesian, too, with the static
gravity);
5) it is still there after a lot of timesteps, when I run with
self-gravity;
6) it doesn't depend on the choice on xmin and ymin.
I don't know if it causes the crash of the simulation, because a more serious
problem comes after this. In supernova_hdf_chk_tot_vel0003.png I show you the
velocity after some other timesteps; as you can see, along the y-axis, there
are very large, spurious velocities. In that location, temperature grows
unphysically and the simulation fails.
It has, too, the features 2), 3), 6), and:
7) it appears only with cylindrical geometry, even with static gravity;
8) any calculation with cartesian geometry only shows the previous problem,
not this one.
I tried a lot of different solutions on runtime parameters, without any
result... Did you have some similar problem, that you solved? What could be a
possible solution?
Looking forward to your comments,
Regards
Luigi
-- --------------------------------------------------------------- Luigi Iapichino Max Planck Institut fuer Astrophysik Karl-Schwarzschild-Str. 1 D-85741, Garching bei Muenchen, Germany phone: (49) -89 300002231 fax:(49) -89 300002235 e-mail: iapichin@mpa-garching.mpg.de
