[Mesa-users] Identical Issue in MESA 15140 and 9793: "max decrease"

Thu Apr 29 08:16:54 UTC 2021

Hi Zafar,

Could you attach your inlists or are you really just using the default work folder?  I ask because I just ran a default work folder in r15140 (SDK 20.3.2 on Fedora 32) and, though the output is similar to yours, it isn't identical, which it should be.  But it also sounds like you may have changed something about the timesteps.

Also, the output you've attached doesn't actually contain an error message.  The "flags" you refer to are just terminal output giving information about what is currently limiting the timestep.  That doesn't automatically mean there's a problem.

I did notice is that the second or third timestep is when the default PGSTAR windows open.  Could you try removing or commenting the line

     pgstar_flag = .true.

in `&star_job` (at the star of `inlist_project`) or setting it to `.false.`?  I wonder if PGSTAR is somehow quietly failing and that's stopping the run.

I'll try SDK 21.4.1 and reply again if I find something (but I doubt it).

Cheers,
Warrick

PS:  It's a *good* thing that you get the same behaviour with the same MESA version across different computers and SDKs.  The devs make an effort to make sure this is the case.

___________

Warrick Ball
Postdoc, School of Physics and Astronomy
University of Birmingham, Edgbaston, Birmingham B15 2TT
W.H.Ball at bham.ac.uk
+44 (0)121 414 4552

On Thu, 29 Apr 2021, mesa-users at lists.mesastar.org wrote:

> Hello MESA users,
> 
> I've successfully installed MESA version 9793, along with the contemporary version of MESASDK (mesasdk-x86_64-linux-20170802.tar.gz) on my linux machine. I tried following the generic MESA tutorial to generate a 15 solar-mass pre-main-sequence star,
> but was met with a convergence issue that resulted in the model stopping after 3 timesteps with the flags "retry", "max decrease", "max increase", and "varcontrol" often appearing. The output is pasted below. 
> 
> Fascinatingly, I was able to reproduce this *exact error* with MESA version 15140 (and the corresponding version of MESASDK--mesasdk-x86_64-linux-21.4.1.tar.gz) on the linux machine. I was also able to reproduce this *exact same error* on my laptop
> (MBP, Mojave 10.14.5) with both version 15140 and 9793. In each case, the export paths were triple checked, the MESASDK versions were verified to be correct, and the code was installed correctly. They all just crash after 3 loops, regardless of my
> perturbations to the input variables and the time step settings in "controls.defaults". I attempted to evolve the star after main sequence, and was met with the same error. I also attempted running Jonathan Fortney's hot Jupiter evolution problems
> from the 2017 summer school, and saw the same thing. 
> 
> Below is a typical output of a run. Please note that this error is reproducible even with large timesteps of 10^5 years (unlike the ones below, which look quite small).
> 
> Any advice would be appreciated!
> 
> Thank you very much,
> Zafar Rustamkulov
> 
>                                          version_number        9793
> 
>  read inlist_project
> 
> 
> 
>  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
> 
>                                           species  mass           8    1.5000000000000000D+01
> 
> 
>                                         relax_num_steps
> 
> 
>        retry log10(dt/yr), log10(dt), retry_factor     1     1       -5.602060        1.897052        0.500000
> 
>  first model is slow to converge: num tries          70
> 
>  first model is slow to converge: num tries          80
> 
>  first model is slow to converge: num tries          90
> 
>  first model is slow to converge: num tries         100
> 
>  first model is slow to converge: num tries         110
> 
>  first model is slow to converge: num tries         120
> 
>  first model is slow to converge: num tries         130
> 
>           1   5.479738   3329.689 -21.749184 -21.749184  15.000000  15.000000   0.700000   0.001008   0.280000   0.700000 -12.925080   1077      1
> 
>   -5.301030  -5.414823   2.521546 -99.000000 -23.147052 -99.000000   0.000000   0.280000   0.009360   0.020000   0.280000   0.013156    137      0
> 
>  5.0000E-06   8.245847   4.086189 -99.000000   3.081810  -8.232014   0.000000   0.003442   0.002099  2.000E-02  2.000E-02  0.291E+04         retry
> 
> 
>           2   5.478931   3363.207 -21.749408 -21.749408  15.000000  15.000000   0.700000   0.001008   0.280000   0.700000 -12.925698    611      1
> 
>   -5.397940  -5.416320   2.521572 -99.000000 -23.147276 -99.000000   0.000000   0.280000   0.009360   0.020000   0.280000   0.013165      6      0
> 
>  9.0000E-06   8.243432   4.103640 -37.327719   3.077837  -8.251254   0.000000   0.003442   0.002099  2.000E-02  2.000E-02  0.196E+02  max decrease
> 
> 
>           3   5.479032   3380.736 -21.749698 -21.749698  15.000000  15.000000   0.700000   0.001008   0.280000   0.700000 -12.925451    439      1
> 
>   -5.243709  -5.416057   2.521615 -99.000000 -23.147566 -99.000000   0.000000   0.280000   0.009360   0.020000   0.280000   0.013165      6      0
> 
>  1.4705E-05   8.243801   4.112758 -99.000000   3.069478  -8.262098   0.000000   0.003442   0.002099  2.000E-02  2.000E-02  0.233E+02    varcontrol
> 
> 
> 
> 
> 
>