[CIG-LONG] Gale 2.0.0

Mon Apr 2 16:49:33 PDT 2012

Hi Rainer,

Cc'ing the list.

I just did some experiments with input/cookbook/thermal.json, and it
seems to work fine.  Could you send your input file?  That should make
it easier to figure out what is going on.

Cheers,
Walter Landry

Rainer Nerlich <rainer at simula.no> wrote:
> Dear Mr. Landry,
> 
> I am sorry to bother you again, but I still encounter a problem while
> running Gale. I saw your solution regarding running a thermal model in
> parallel (i.e. using the same mesh for temperatures as for
> velocities), which works fine. However, it seems that the time step
> problem, which I had already mentioned in my previous email, is not
> fixed yet.
> 
> 
> I tested various viscosities for my model by scaling the velocities
> and - if temperatures are included - the diffusivity accordingly
> (heating elements are set to zero). I expected that the velocity field
> would basically look identical (except that the length of the vectors
> differ slightly for different viscosities) and that the time steps
> vary only with respect to the exponent.
> 
> When I include velocity boundary conditions only, the code is doing
> what I expected. If I additionally include temperature boundary and
> initial conditions, I get the expected result only if I assign
> viscosities of 10^22 or higher. If I use lower viscosities, the
> velocity field looks different (wrong) and the time steps vary not
> only with respect to the exponent. Below, you will see what I get.
> 
> My guess is that there is still something wrong with the temperatures
> and maybe you already understand what is going on and how it can be
> fixed.
> 
> Thanks again for your help!
> 
> Rainer Nerlich
> 
> 
> *Only velocity boundary conditions (everything works):*
> 
> 
> *1) scaling factor = 10^19*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_BCs_only/10_19_velocity_BCs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 4.40056e-06
> 
> TimeStep = 2, Time = 8.80112e-06
> 
> TimeStep = 3, Time = 1.32017e-05
> 
> TimeStep = 4, Time = 1.76022e-05
> 
> TimeStep = 5, Time = 2.20028e-05
> 
> 
> real 2m53.061s
> 
> user 22m25.880s
> 
> sys 0m4.770s
> 
> 
> *2) scaling factor = 10^20*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_BCs_only/10_20_velocity_BCs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 4.40056e-07
> 
> TimeStep = 2, Time = 8.80112e-07
> 
> TimeStep = 3, Time = 1.32017e-06
> 
> TimeStep = 4, Time = 1.76022e-06
> 
> TimeStep = 5, Time = 2.20028e-06
> 
> 
> real 3m23.387s
> 
> user 26m28.250s
> 
> sys 0m5.390s
> 
> 
> *3) scaling factor = 10^21*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_BCs_only/10_21_velocity_BCs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 4.40056e-08
> 
> TimeStep = 2, Time = 8.80112e-08
> 
> TimeStep = 3, Time = 1.32017e-07
> 
> TimeStep = 4, Time = 1.76022e-07
> 
> TimeStep = 5, Time = 2.20028e-07
> 
> 
> real 3m36.426s
> 
> user 28m13.990s
> 
> sys 0m5.540s
> 
> 
> *4) scaling factor = 10^22*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_BCs_only/10_22_velocity_BCs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 4.40056e-09
> 
> TimeStep = 2, Time = 8.80112e-09
> 
> TimeStep = 3, Time = 1.32017e-08
> 
> TimeStep = 4, Time = 1.76022e-08
> 
> TimeStep = 5, Time = 2.20028e-08
> 
> 
> real 3m58.297s
> 
> user 30m22.950s
> 
> sys 0m6.670s
> 
> 
> *5) scaling factor = 10^23*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_BCs_only/10_23_velocity_BCs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 4.40056e-10
> 
> TimeStep = 2, Time = 8.80112e-10
> 
> TimeStep = 3, Time = 1.32017e-09
> 
> TimeStep = 4, Time = 1.76022e-09
> 
> TimeStep = 5, Time = 2.20028e-09
> 
> 
> real 4m57.561s
> 
> user 36m5.310s
> 
> sys 0m9.610s
> 
> 
> *Velocity boundary conditions + temperature boundary and initial
> *conditions (error 1: varying time steps for viscosities < 10^22; error
> *2: inconsistencies in calculation times):*
> 
> 
> *1) scaling factor = 10^19*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_19_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 1.98605e-08
> 
> TimeStep = 2, Time = 4.3173e-08
> 
> TimeStep = 3, Time = 6.35794e-08
> 
> TimeStep = 4, Time = 8.92363e-08
> 
> TimeStep = 5, Time = 1.02624e-07
> 
> 
> real 3m1.425s
> 
> user 23m12.920s
> 
> sys 0m6.180s
> 
> 
> *2) scaling factor = 10^20*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_20_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 2.53938e-08
> 
> TimeStep = 2, Time = 4.64073e-08
> 
> TimeStep = 3, Time = 7.02431e-08
> 
> TimeStep = 4, Time = 9.56053e-08
> 
> TimeStep = 5, Time = 1.10074e-07
> 
> 
> real 3m15.835s
> 
> user 24m45.900s
> 
> sys 0m6.400s
> 
> 
> *3) scaling factor = 10^21*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_21_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 1.67951e-08
> 
> TimeStep = 2, Time = 3.95265e-08
> 
> TimeStep = 3, Time = 6.22898e-08
> 
> TimeStep = 4, Time = 8.483e-08
> 
> TimeStep = 5, Time = 9.72328e-08
> 
> 
> real 3m28.267s
> 
> user 26m25.040s
> 
> sys 0m6.480s
> 
> 
> *4) scaling factor = 10^22*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_22_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 2.66209e-09
> 
> TimeStep = 2, Time = 5.32418e-09
> 
> TimeStep = 3, Time = 7.98626e-09
> 
> TimeStep = 4, Time = 1.06484e-08
> 
> TimeStep = 5, Time = 1.33104e-08
> 
> 
> real 3m16.467s
> 
> user 24m55.460s
> 
> sys 0m5.170s
> 
> 
> *5) scaling factor = 10^23*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_23_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 2.66209e-10
> 
> TimeStep = 2, Time = 5.32418e-10
> 
> TimeStep = 3, Time = 7.98626e-10
> 
> TimeStep = 4, Time = 1.06484e-09
> 
> TimeStep = 5, Time = 1.33104e-09
> 
> 
> real 3m4.166s
> 
> user 24m1.080s
> 
> sys 0m5.000s
> 
> 
> *6) scaling factor = 10^24*
> 
> rainer at rainer-laptop:~/Gale2.0.0/new_source_code/gale$ time mpirun -np
> 8 build/bin/Gale
> input/my_model/velocity_temp_BCs/10_24_velocity_temp_BCs_temp_ICs.json
> 
> TimeStep = 0, Time = 0
> 
> TimeStep = 1, Time = 2.66209e-11
> 
> TimeStep = 2, Time = 5.32418e-11
> 
> TimeStep = 3, Time = 7.98627e-11
> 
> TimeStep = 4, Time = 1.06484e-10
> 
> TimeStep = 5, Time = 1.33104e-10
> 
> 
> real 4m0.920s
> 
> user 30m30.580s
> 
> sys 0m5.770s
> 
> 
> 
> 
> On 03/08/2012 08:26 PM, Walter Landry wrote:
>> Rainer Nerlich<rainer at simula.no>  wrote:
>>>   More likely it seems to me that there is a more general problem with
>>> temperatures, perhaps only on my computer.
>> It is not just you.  It is the same problem that other people are
>> seeing.  I am working on it.
>>
>>>   Finally, I was also wondering about how the timesteps are actually
>>> calculated. I noticed that if I run my input file with velocity
>>> boundary
>>> conditions only, the timesteps are the same as if I run my input file
>>> including temperature boundary/initial conditions (on one
>>> processor). Does
>>> that mean that the temperature condition are actually ignored when the
>>> timesteps are determined?
>> The time step is supposed to depend on thermal parameters.  There
>> seems to be something wrong there too.
>>
>> Cheers,
>> Walter Landry
> 