[mesa-users] negative age and rotation values in "OSC" file

RICHARD H D TOWNSEND townsend at astro.wisc.edu
Sat Feb 11 17:17:56 EST 2017


Hi Juan Carlos —

In your email below, I can spot quite a few different problems you’re running into. In each case I’ll try to explain what’s going on

> On Feb 10, 2017, at 12:26 PM, Juan Carlos Suárez <jcsuarez at iaa.es> wrote:
> 
> Dear Richard,
> 
> We have modified a little bit the previous “inlist” in order to include some atmosphere (the default simple atmosphere)
> since we are interested in working with the photosphere radius. So we decided to include the following options in the
>  inlist (attached)
> ! if true, write atmosphere to pulse files
> 
> add_atmosphere_to_pulse_info = .TRUE.

Although it’s not the main cause of your problems below, I should mention that In MESA 8845 (and previous releases) there is a bug in the atmosphere writing routine. The code should output the same atmosphere as selected by the ‘which_atm_option’ parameter; however, in many cases (including the default ‘simple_photosphere’ atmosphere), the atmosphere written out is actually the ‘Paczynski_grey’ atmosphere. This bug is fixed in the development version of MESA, but only for a restricted set of atmosphere choices:

Eddington_grey
Paczynski_grey
solar_Hopf_grey
Krishna_Swamy
grey_and_kap

For other atmosphere options the code now halts — because there is insufficient information to construct the atmosphere structure.

> ! if true, add point for r=0 to pulse files
> 
> add_center_point_to_pulse_info = .FALSE.

In fact, it’s setting this flag to .FALSE. that ‘fixed’ the problem with age and rotation rate being overwritten (as discussed in the previous email) — the central second derivative data only gets calculated when add_center_point_to_pulse_info = .TRUE. (which itself is a bug — again, fixed in the development version)

> ! if true, add k=1 cell to pulse files
> 
> keep_surface_point_for_pulse_info = .false.
> ! add double points at discontinuities
> 
> add_double_points_to_pulse_info = .TRUE.

This is what’s causing the (approximate) doubling of the number of points in the OSC files. This flag causes MESA to write out a double points when it detects a density discontinuity. The threshold for recognizing density discontinuities is set by the threshold_dlnrho_for_double_point flag. Since this flag defaults to zero, *every* pair of adjacent cells in the model is producing a double point!

The fix is to set add_double_points_to_pulse_info = .FALSE., or to specify a non-zero value for threshold_dlnrho_for_double_point — e.g., 10.

> 
> Curiously, it seems that now the problem of the negative ages and rotation values is solved, but we
> do not know why. 
> 

See above!

> On the other hand unfortunately the computation halted with an error 
> 
> failed to open LOGS/profile686.data
>  star_finish_step ierr          28
>  after_step_loop ierr          28
> 
> That is, it halted at model 686, while without the above options MESA reached the 905 models with no errors.
> 

I imagine the program halted because it ran out of disk space. You are writing out a profile *and* an OSC file after each timestep. Each profile file is ~15MB, and each OSC file is ~5MB. After 686 models, you have used up well over 10 GB of disk space.

To get around this issue, you should consider the following approaches:

*) Don’t set mesh_delta_coeff so small — you use 0.15, which results in very high resolution models (some might say over-resolved). 0.5 might be a better choice for this parameter.

*) Only write out every N’th profile/OSC file, rather than all of them. To do this, set profile_interval = N

*) Don’t bother writing out the pre-MS models, if you don’t need them. This can be achieved by doing two separate runs — an initial run with no profile/OSC file output, which terminates at the ZAMS. Then, do a restart and write out profiles/OSC files as necessary.

> In addition to this, with the above options it seems that MESA has doubled the number
> of mesh points (from 4K to 8K). Is that what we should expect? The 4K additional 
> points are located at the atmosphere?
> 

No, as I mentioned above, every single model point is being written as a double point — not what you want!

So, to sum up:

*) Upgrade to the development version of MESA.

*) Set add_atmosphere_to_pulse_data = .FALSE., or set which_atm_option to one of the atmosphere choices listed above.

*) Set add_double_points_to_pulse_data = .FALSE., or specify a non-zero value for threshold_dlnrho_for_double_point.

*) Adjust mesh_delta_coeff, set_profile_interval to reduce size of files, or run the pre-MS and MS parts separately.

(Note also that some of the parameter names have changed, with the phrase ‘pulse_info’ changed to ‘pulse_data’)

Hope that helps!

cheers,

Rich


> Thanks!
> JC
> 
> 
>> On 9 Feb 2017, at 20:38, Juan Carlos Suárez <jcsuarez at iaa.es> wrote:
>> 
>> Yes here it is!
>> JC
>>> On 9 Feb 2017, at 20:05, RICHARD H D TOWNSEND <townsend at astro.wisc.edu> wrote:
>>> 
>>> Hi Juan —
>>> 
>>> Can you post the inlist file(s) that you’re using?
>>> 
>>> Many thanks,
>>> 
>>> Rich
>>> 
>>>> On Feb 9, 2017, at 12:43 PM, Juan Carlos Suárez <jcsuarez at iaa.es> wrote:
>>>> 
>>>> Dear colleagues
>>>> 
>>>> I am trying to compute a 1.5 Msun evolutionary model (with solar metallicity) from PMS to TAMS approx.
>>>> We have set the options to obtain the .OSC files to be read by our oscillation codes. 
>>>> 
>>>> We found that for all the models the age provided has negative values.
>>>> 
>>>> Moreover, although no option for rotation was included in the inlist (i.e. rotation should be zero)
>>>> we obtain some models with rotation 0 but others with negative non-zero values
>>>> 
>>>> Any idea?
>>>> 
>>>> Thanks in advance!
>>>> 
> 
> --------------------------------------------------------------------------------------------------------------------
> Dr. Juan Carlos Suárez 
>     Ramon y Cajal fellow at Física Teórica y del Cosmos Dept. University of Granada 
>     Facultad de Ciencias. 18071. Granada. Spain
>   jcsuarez at ugr.es | +34 958249056 | www.ugr.es/~fqm292/ 
>     Associate researcher at Stellar Physics Dept. IAA-CSIC
>   Glorieta de la Astronomia s/n. 18008. Granada. Spain
>   jcsuarez at iaa.es | +34 958230619 | www.iaa.es
> --------------------------------------------------------------------------------------------------------------------
> 
> 
> 
> <inlist_project>



More information about the Mesa-users mailing list