[Mesa-users] segfault when using mesa_833.net with version 10398

[Francis Timmes]

hi meridith,

> I think it has to do with the controls setting 
> default_net_name = 'mesa_833.net'
> It runs without issue when using the default basic.net.


833 isotopes in every cell will chew up memory.

the short answer is the maximum memory that mesa can use for its matrix is ~ 16GB.
depending on how many isotopes want, you may need to reduce the number of cells.

the longer answer is the size of the matrix is set by the 4-byte integer variable lwork
in star/private/star_newton.f90 in get_newton_work_sizes. the upshot, assuming i did the
substitutions correctly, is 

memory = (niso + 5) * nz * (3*(niso + 5) + 9) < 2^31 = 2,147,483,648

where niso is the number of isotopes and nz is the number of cells.
given niso, one can solve for the maximum number of cells nz before this memory limit is hit.

fxt





> On Feb 7, 2019, at 10:25 PM, Meridith Joyce via Mesa-users <mesa-users at lists.mesastar.org> wrote:
> 
> Hi All,
> 
> I receive the following error/output when I attempt to use the attached inlist in MESA version 10398:
> 
> 1549602124.13
> gfortran -fopenmp -o ../star run_star_extras.o run_star.o  run.o  -L/avatar/mjoyce/mesa-r10398/lib -lstar -lionization -latm -lcolors -lmlt -lnet -leos -lkap -lrates -lneu -lchem -linterp_2d -linterp_1d -lnum -lf2crlibm -lcrlibm -lmtx -lconst -lutils `mesasdk_lapack_link` `mesasdk_blas_link` `mesasdk_hdf5_link` `mesasdk_pgplot_link` -lz 
> DATE: 2019-02-08
> TIME: 16:02:05
>                                          version_number       10398
>  read 2M_pulses.inlist
> 
> 
>  The terminal output contains the following information
> 
>       'step' is the number of steps since the start of the run,
>       'lg_dt' is log10 timestep in years,
>       'age_yr' is the simulated years since the start run,
>       'lg_Tcntr' is log10 center temperature (K),
>       'lg_Dcntr' is log10 center density (g/cm^3),
>       'lg_Pcntr' is log10 center pressure (ergs/cm^3),
>       'Teff' is the surface temperature (K),
>       'lg_R' is log10 surface radius (Rsun),
>       'lg_L' is log10 surface luminosity (Lsun),
>       'lg_LH' is log10 total PP and CNO hydrogen burning power (Lsun),
>       'lg_L3a' is log10 total triple-alpha helium burning power (Lsun),
>       'lg_LZ' is log10 total burning power excluding LH and L3a and photodisintegrations (Lsun),
>       'lg_LNuc' is log10 nuclear power excluding photodisintegration (Lsun),
>       'lg_LNeu' is log10 total neutrino power (Lsun),
>       'lg_Psurf' is log10 surface pressure (gas + radiation),
>       'Mass' is the total stellar baryonic mass (Msun),
>       'lg_Mdot' is log10 magnitude of rate of change of mass (Msun/year),
>       'lg_Dsurf' is log10 surface density (g/cm^3),
>       'H_env' is the amount of mass where H is the most abundant iso,
>       'He_core' is the largest mass where He is most abundant iso.
>       'C_core' is the largest mass where C is most abundant iso.
>       'H_cntr' is the center H1 mass fraction,
>       'He_cntr' is the center He4 mass fraction,
>       'C_cntr' is the center C12 mass fraction,
>       'N_cntr' is the center N14 mass fraction,
>       'O_cntr' is the center O16 mass fraction,
>       'Ne_cntr' is the center Ne20 mass fraction,
>       'X_avg' is the star average hydrogen mass fraction,
>       'Y_avg' is the star average helium mass fraction,
>       'Z_avg' is the star average metallicity,
>       'gam_cntr' is the center plasma interaction parameter,
>       'eta_cntr' is the center electron degeneracy parameter,
>       'zones' is the number of zones in the current model,
>       'iters' is the number of newton iterations for the current step,
>       'retry' is the number of step retries required during the run,
>       'bckup' is the number of step backups required during the run,
>       'dt_limit' is an indication of what limited the timestep.
> 
>  All this and more are saved in the LOGS directory during the run.
>  create pre-main-sequence model
>  use_eosPTEH_for_low_density T
>  use_eosPTEH_for_high_Z T
>                                           species  mass         833    2.0000000000000000D+00
> 
>                                         relax_num_steps
> 
> 
> Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
> 
> ... snip
> 
> 
> #5  0xffffffffffffffff in ???
> 
> I think it has to do with the controls setting 
> default_net_name = 'mesa_833.net'
> It runs without issue when using the default basic.net.
> 
> On a previous attempt with the same inlist, it seems MESA correctly generated the *.bin files associated with the larger nuclear network in mesa-r10398/data/rates_data/cache/
> 
> Could anyone point me in the right direction?
> 
> Thank you,
> 
> -Meridith
> <2M_pulses.inlist>_______________________________________________
> mesa-users at lists.mesastar.org
> https://lists.mesastar.org/mailman/listinfo/mesa-users