Different results from FGONG or AMDL version of model file

[warrick]

Hi,

I seem to be getting different oscillation frequencies for a stellar model depending on whether I use an FGONG file or an AMDL file, converted using either ADIPLS' internal fgong converter (if you have MESA, it's at $MESA_DIR/adipls/adipack.c/adiajobs/fgong-amdl.d.f) or my own Python-based converter (using TOMSO). I've attached example files for Model S, with the AMDL file gzipped because otherwise it was rejected as an attachment.

For the summary using the FGONG model, I get

Code: Select all

                                                                                                                                                                                                                                                                



                        1                        2                        3                        4                        5                        6
                        l                     n_pg                      n_p                      n_g                 Re(freq)                 Im(freq)
                        0                       21                       21                        0  0.3038081409866325E+004  0.0000000000000000E+000
                        0                       22                       22                        0  0.3174111887772760E+004  0.0000000000000000E+000
                        0                       23                       23                        0  0.3310180199798498E+004  0.0000000000000000E+000
                        0                       24                       24                        0  0.3446893236132409E+004  0.0000000000000000E+000
                        0                       25                       25                        0  0.3583918263283139E+004  0.0000000000000000E+000
                        0                       26                       26                        0  0.3721163831988589E+004  0.0000000000000000E+000
                        0                       27                       27                        0  0.3858790528140903E+004  0.0000000000000000E+000
                        0                       28                       28                        0  0.3996454584650650E+004  0.0000000000000000E+000
                        0                       29                       29                        0  0.4134355457053003E+004  0.0000000000000000E+000
                        0                       30                       30                        0  0.4272334189292133E+004  0.0000000000000000E+000
                        0                       31                       31                        0  0.4410314077774772E+004  0.0000000000000000E+000
                        0                       32                       32                        0  0.4548466828006668E+004  0.0000000000000000E+000
                        0                       33                       33                        0  0.4686358934199711E+004  0.0000000000000000E+000
                        0                       34                       34                        0  0.4824085736439642E+004  0.0000000000000000E+000
                        0                       35                       35                        0  0.4961514516938695E+004  0.0000000000000000E+000

whereas with the AMDL file I get

Code: Select all

                                                                                                                                                                                                                                                                



                        1                        2                        3                        4                        5                        6
                        l                     n_pg                      n_p                      n_g                 Re(freq)                 Im(freq)
                        0                       15                       15                        0  0.3102764484173943E+004  0.0000000000000000E+000
                        0                       16                       16                        0  0.3290559811239307E+004  0.0000000000000000E+000
                        0                       17                       17                        0  0.3479404763661229E+004  0.0000000000000000E+000
                        0                       18                       18                        0  0.3669127390608365E+004  0.0000000000000000E+000
                        0                       19                       19                        0  0.3859486183176797E+004  0.0000000000000000E+000
                        0                       20                       20                        0  0.4050645670093928E+004  0.0000000000000000E+000
                        0                       21                       21                        0  0.4242285556617217E+004  0.0000000000000000E+000
                        0                       22                       22                        0  0.4434049354298812E+004  0.0000000000000000E+000
                        0                       23                       23                        0  0.4625786127087709E+004  0.0000000000000000E+000
                        0                       24                       24                        0  0.4817157007065860E+004  0.0000000000000000E+000

The FGONG output is what I expect.

I started looking into the GYRE's source (and comparing it with the source for ADIPLS's FGONG to AMDL script) but nothing has jumped out at me as being incorrect. I'll keep digging, though.

Cheers,
Warrick

[warrick]

As usual, shortly after answering the question, I find the answer.

This is almost certainly a bug in the calculation of V_2. e.g. in gyre_amdl_file.fpp, V_2 is set (for x > 0) by

Code: Select all

       V_2 = V_g*Gamma_1

whereas I've inferred from the other (working) model readers that it should be

Code: Select all

       V_2 = V_g*Gamma_1/x**2

The same bug affects the FAMDL reader. I'll open a pull request.

Cheers,
Warrick

[warrick]

Finally, to close this off for anywhere else who finds this thread, Rich already fixed this in the development version. I just cleaned up my input file to use Model S's value of G (6.67232e-8) and I get the same radial mode frequencies for the FGONG and the AMDL at the machine precision level.

W

[rhtownsend]

Finally, to close this off for anywhere else who finds this thread, Rich already fixed this in the development version. I just cleaned up my input file to use Model S's value of G (6.67232e-8) and I get the same radial mode frequencies for the FGONG and the AMDL at the machine precision level.

W

Machine precision is nice! Thanks for following through on this.