[mesa-users] variability in runs

[Richard Townsend]

Hi Andres --

As a first suggestion, could you try repeating the runs with OMP_NUM_THREADS set to 1. This will allow us to diagnose whether your findings are caused by a problem with multi-threading.

cheers,

Rich

On Apr 23, 2012, at 2:11 PM, andres suarez wrote:

> Hello all,
> 
> I noticed that running the same model in mesa more than once doesn't always give the same evolution.
> I am using the exact same inlist and inlist_project for the different runs and running in the same machine, but I get different results in the star properties and in the number of output files. The different runs sometimes give the exact same results, more often will give something similar, but sometimes the evolution will be quite distinct.
> As an example, I attach a simple inlist file used for three 73 solar mass runs (same in three runs) and a plot of the number of zones in each run (as an illustration of the variability found).
> 
> I am running mesa version 3851 on a linux platform on an x86_64 multi-processor computer, using ifort.
> Does anyone know if there's a known reason for this? I don't know if mesa has any non-deterministic part in it... or if rounding errors could lead to such varying evolution.
> 
> Regards,
> 
> a.
> 
> 
> 
> &star_job
> 
>       mesa_data_dir = '/home/andres/Arbeit/mesa/data'
> 
>       create_pre_main_sequence_model = .false.
> 
>       change_net = .true. ! switch nuclear reaction network
>       new_net_name = 'o18_and_ne22.net'
> 
> / !end of star_job
> 
> &controls
>      ! starting specifications
>          initial_mass = 73 ! in Msun units
>          initial_Z = 0.02
>          initial_Y = -1
> 
>      ! mass gain or loss
>          RGB_wind_scheme = 'Reimers'
>          Reimers_wind_eta = 0.4  
> 
>      ! when to stop
>          xa_central_lower_limit_species(1) = 'h1'
>          xa_central_lower_limit(1) = 1d-10
>             ! stop when the center abundance by mass of h1 drops below this limit.
> 
> /
> <compara.ps>------------------------------------------------------------------------------
> For Developers, A Lot Can Happen In A Second.
> Boundary is the first to Know...and Tell You.
> Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
> http://p.sf.net/sfu/Boundary-d2dvs2
> _______________________________________________
> mesa-users mailing list
> mesa-users at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/mesa-users