[Mesa-users] Inquiry About MESA Convergence Issues During Convective Region Expansion

[Frank Timmes]

one could try restarting from before the h ingestion event, and reduce the timestep more than the inlists provided suggest have been tried. some controls to try include

! time step resolution on central temperature and density changes
      delta_lgT_cntr_limit   = 0.001   ! default 0.01
      delta_lgT_limit = 0.05d0         ! default 0.5

      delta_lgRho_cntr_limit = 0.005   ! default 0.05
      delta_lgRho_limit = 0.1          ! default 1.0


! when any cell goes above logT = 7.0, impose time step resolution on temperature changes
      delta_lgT_max_limit_lgT_min = 7.0d0
      delta_lgT_max_limit = 0.001     
      delta_lgT_max_hard_limit = 0.05

! time step resolution the helium burning luminosity
      delta_lgL_He_limit = 0.005 
!      delta_lgL_H_limit - 0.005 

! time step limit abundance changes in a timestep
      dX_nuc_drop_limit = 2.0d-3        ! default = 5.0d-2
      dX_nuc_drop_limit_at_high_T = -1  ! default = -1 = same as dX_nuc_drop_limit
      dX_nuc_drop_min_X_limit = 1.0d-3  ! default = 1.0d-4
      dX_nuc_drop_max_A_limit = 52    


> ... improve spatial resolution, particularly at the convective boundaries, ...

one could try the mesh control

https://docs.mesastar.org/en/latest/reference/controls.html#convective-bdy-dq-limit

alternatively one could a different convective mixing model; mesa offers several.

fxt




> On May 18, 2025, at 8:34 PM, Guo Li via Mesa-users <mesa-users at lists.mesastar.org> wrote:
> 
> Dear All,
> I am using MESA-24.08.1 to evolve a star with M=1 Msun and [Fe/H]=-3. During the second thermal pulse, a proton ingestion event and H-flash occur. After the lg_LH exceeds 8.0, the convective zone in the He intershell expands outward, and the code fails to converge at the upper boundary of the convective region. I have attached my inlist file, mod file, terminal output, and the mass fraction profile of nuclides at the termination point for reference.
> I attempted to improve spatial resolution, particularly at the convective boundaries, but the convergence issue persisted. The problem was only resolved by relaxing the convergence criteria as follows:
> tol_correction_norm = 8d-4 
> tol_max_correction = 8d-2 
> 
> use_gold2_tolerances = .false. 
> 
> convergence_ignore_equL_residuals = .true. 
> 
> use_gold_tolerances = .true. ! the following only apply when this flag is true 
> gold_tol_residual_norm1 = 1d-6 
> gold_tol_max_residual1 = 1d-4 
> gold_iter_for_resid_tol2 = 5 
> gold_tol_residual_norm2 = 1d-4 
> gold_tol_max_residual2 = 1d-2 
> gold_iter_for_resid_tol3 = 10 
> gold_tol_residual_norm3 = 1d-1 
> gold_tol_max_residual3 = 1d1 
> However, I believe this approach compromises accuracy, as it appears to cause unnaturally rapid expansion of the He intershell convective zone.
> Question:
> For this convergence issue during the expansion phase of the He intershell convective region, what methods should I try to resolve it without sacrificing accuracy? Any advice would be greatly appreciated.
> Best regards,
> Guo Li
> National Astronomical Observatories, CAS<image.png><image.png><inlist><inlist_common><inlist_TP_AGB><inlist_TP2_4535_4550_history_LH1d9><out_TP2_4535_4550_history_LH1d9><TP2_4535_4550_history_1d9.mod>_______________________________________________
> mesa-users at lists.mesastar.org
> https://lists.mesastar.org/mailman/listinfo/mesa-users
>