[FLASH-USERS] Possible Restart Bug in 3.0

[Anshu Dubey]

James,

We will definitely need all the Simulation_ and Grid_ files to have any
chance of finding out
the problem. However, without having all the files it is not clear that we
can solve your problem, because
we run many problems with restart in our nightly test suite, and all of them
start transparently. It seems
very likely that your restart issue is specific to your problem.

If you are really uncomfortable about sharing the files, then you can try
the following steps, at each step
making sure that you have transparent restart :
1. Start with a simple problem like Sedov.
2. Add self gravity to it.
3. Add your custom gravity routines.
4. Start your own problem with gravity turned off, and default
mark_refine_derefine
5. Add each custom component one by one.

That is basically the strategy I would use for identifying the problem.

Let me also assure you that if you were to give the files to us, only one
person will have access to them,
and he/she will delete them as soon as they are done with them.

Anshu

On Tue, May 13, 2008 at 4:01 PM, James Guillochon <jfg at ucolick.org> wrote:

> I just checked and the problem occurs even if the max refinement level is
> the same before and after the restart.
>
> I've attached two flash.par files, the first is from the initial run, the
> second is for the restarted run. I've modified the following built-in files:
>
> Gravity_accelAtCoords.F90
> Gravity_accelListOfBlocks.F90
> Gravity_accelOneRow.F90
> Gravity_data.F90
> Gravity_init.F90
> Gravity_potentialListOfBlocks.F90
> Grid_data.F90
> Grid_init.F90
> Grid_markRefineDerefine.F90
> Simulation_data.F90
> Simulation_initBlock.F90
> Simulation_init.F90
>
> Which in particular would you like me to attach? (I don't want to attach
> all of them)
>
> --
> James Guillochon
> Department of Astronomy & Astrophysics
> University of California, Santa Cruz
> jfg at ucolick.org
>
>
> Anshu Dubey wrote:
>
>> Hi James,
>>
>> Can you share your setup, flash.par and any other customized file with us
>> ? We have never encountered the problem
>> that you are seeing, and we'd really like to be able to reproduce it
>> ourselves to figure out what is going on.
>>
>> Anshu
>>
>> On Tue, May 13, 2008 at 2:35 PM, James Guillochon <jfg at ucolick.org<mailto:
>> jfg at ucolick.org>> wrote:
>>
>>    Hey guys,
>>
>>    It seems that the velz variable is getting shifted relative to the
>>    other variables when I restart a simulation. This shift appears to
>>    be equal to the nzb variable, or 8 blocks, at the highest refinement
>>    level. I have attached two screenshots: The first is prior to the
>>    restart, the second is immediately after the restart.
>>
>>    The only variable that changes between restarts is the max
>>    refinement level; however the shift in velz is visible before the
>>    grid is refined. No other variables are shifted. This is a real
>>    problem for me since my problem assumes symmetry about the z axis!
>>
>>    Any ideas guys?
>>
>>    --    James Guillochon
>>    Department of Astronomy & Astrophysics
>>    University of California, Santa Cruz
>>    jfg at ucolick.org <mailto:jfg at ucolick.org>
>>
>>
>> !DSPAM:10135,4829ef7c2382245126300!
>>
>
> lrefine_min             = 1
> lrefine_max             = 10
>
>
> basenm                  = "tidal_"
> restart                 = .true.
> checkpointFileIntervalTime = 200.0
> checkpointFileNumber = 9
> plotFileIntervalTime    = 0.5
>
> xmax                    = 6.e11
> ymax                    = 6.e11
> zmax                    = 6.e11
>
> ptmass                  = 2.e39
> ptxpos                  = 3.e11
> ptypos                  = 3.e11
> ptzpos                  = 3.e11
>
> small =  1.e-30
> smalle = 1.e-30
> smallt = 1.e-30
> smallu = 1.e-30
> smallx = 1.e-30
> smally = 1.e-30
> smlrho = 1.e-30
> smallp = 1.e-30
> #
> #cfl = 0.8
> #cvisc = 0.e0
> #iplm = 1
> nriem = 1000
> #eos_maxNewton = 1000
> #convertToConsvdForMeshCalls = .true.
> dtinit = 1e-2
> #dtmin = 1e-1
> #dtmax = 1e-2
> #memory_stat_freq = 1
>
> nend                    = 1000000
> tmax                    = 2200.
>
> gamma                   = 1.6667
>
> eintSwitch              = 1.e-4
> #if beta is < 0, the code uses periDist instead.
> sim_periBeta            = 1.
> sim_periDist            = 1.e12
> sim_periTime            = 2000.
> sim_rhoAmbient          = 1.0e-5
> sim_pAmbient            = 1.0e+5
> sim_xctr                = 3.e11
> sim_yctr                = 3.e11
> sim_zctr                = 3.e11
> sim_nsubzones           = 10
>
>
> xl_boundary_type        = "outflow"
> xr_boundary_type        = "outflow"
> yl_boundary_type        = "outflow"
> yr_boundary_type        = "outflow"
> zl_boundary_type        = "outflow"
> zr_boundary_type        = "outflow"
> grav_boundary_type      = "isolated"
> mpole_lmax              = 8
> #useGravity              = .true.
>
> plot_var_1              = "dens"
> plot_var_2              = "temp"
> plot_var_3              = "velx"
> plot_var_4              = "vely"
> plot_var_5              = "velz"
> plot_var_6              = "gpot"
>
> refine_var_1            = "dens"
> #refine_var_2            = "pres"
> refine_filter_1         = 5.e-2
> #refine_filter_2         = 5.e-2
> refine_val_cutoff_1     = 1.e0
>
> lrefine_min             = 1
> lrefine_max             = 8
>
>
> basenm                  = "tidal_"
> restart                 = .false.
> checkpointFileIntervalTime = 200.0
> plotFileIntervalTime    = 10.0
>
> xmax                    = 6.e11
> ymax                    = 6.e11
> zmax                    = 6.e11
>
> ptmass                  = 2.e39
> ptxpos                  = 3.e11
> ptypos                  = 3.e11
> ptzpos                  = 3.e11
>
> small =  1.e-30
> smalle = 1.e-30
> smallt = 1.e-30
> smallu = 1.e-30
> smallx = 1.e-30
> smally = 1.e-30
> smlrho = 1.e-30
> smallp = 1.e-30
> #
> #cfl = 0.8
> #cvisc = 0.e0
> #iplm = 1
> nriem = 1000
> #eos_maxNewton = 1000
> #convertToConsvdForMeshCalls = .true.
> dtinit = 1e-2
> #dtmin = 1e-1
> #dtmax = 1e-2
> #memory_stat_freq = 1
>
> nend                    = 1000000
> tmax                    = 4000.
>
> gamma                   = 1.6667
>
> eintSwitch              = 1.e-4
> #if beta is < 0, the code uses periDist instead.
> sim_periBeta            = 1.
> sim_periDist            = 1.e12
> sim_periTime            = 2000.
> sim_rhoAmbient          = 1.0e-5
> sim_pAmbient            = 1.0e+5
> sim_xctr                = 3.e11
> sim_yctr                = 3.e11
> sim_zctr                = 3.e11
> sim_nsubzones           = 10
>
>
> xl_boundary_type        = "outflow"
> xr_boundary_type        = "outflow"
> yl_boundary_type        = "outflow"
> yr_boundary_type        = "outflow"
> zl_boundary_type        = "outflow"
> zr_boundary_type        = "outflow"
> grav_boundary_type      = "isolated"
> mpole_lmax              = 8
> #useGravity              = .true.
>
> plot_var_1              = "dens"
> plot_var_2              = "temp"
> plot_var_3              = "velx"
> plot_var_4              = "vely"
> plot_var_5              = "velz"
> plot_var_6              = "gpot"
>
> refine_var_1            = "dens"
> #refine_var_2            = "pres"
> refine_filter_1         = 5.e-2
> #refine_filter_2         = 5.e-2
> refine_val_cutoff_1     = 1.e0
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20080514/c8622d70/attachment.htm>