radial order anomaly like -47 -46 -33 -42 -43 -42

[gali8292]

Hi GYRE staff,

When I check gyre output, I found an annoying bug.

l m n_pg Re(freq) Im(freq) Delta_g
1 0 -50 0.5008682947046021E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -49 0.5109359500054714E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -48 0.5214721525874243E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -47 0.5324933402526769E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -46 0.5440221467085419E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -33 0.5560894673277147E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5687164858455284E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -43 0.5819431235954060E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5958068775113836E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -41 0.6103520639155496E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -40 0.6256287702993355E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -39 0.6416841190403432E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -38 0.6585664051204567E+000 0.0000000000000000E+000 0.7464874418973579E+000

Those two bold lines show abnormal radial orders. I have no idea what happen. The MESA profile and gyre.in are attached.

Cheers

Gang

[rhtownsend]

Hi GYRE staff,

When I check gyre output, I found an annoying bug.

l m n_pg Re(freq) Im(freq) Delta_g
1 0 -50 0.5008682947046021E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -49 0.5109359500054714E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -48 0.5214721525874243E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -47 0.5324933402526769E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -46 0.5440221467085419E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -33 0.5560894673277147E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5687164858455284E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -43 0.5819431235954060E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5958068775113836E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -41 0.6103520639155496E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -40 0.6256287702993355E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -39 0.6416841190403432E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -38 0.6585664051204567E+000 0.0000000000000000E+000 0.7464874418973579E+000

Those two bold lines show abnormal radial orders. I have no idea what happen. The MESA profile and gyre.in are attached.

Cheers

Gang

Abnormal radial orders are often a sign that your stellar model harbors density discontinuities. Looking at the rho(r) profile for your model, I can see that this is indeed the case -- there's a jump in the density around r/R ~ 0.1.

To handle density discontinuities properly, GYRE needs a model containing a double point at the discontinuity (i.e., two adjacent points with the same radius values but different density values). You can instruct MESA to write out models containing double points by using the add_double_points_to_pulse_data and threshold_grad_mu_for_double_point controls (and possibly others).

However, that said, there may still be some issues with MESA's double point creation, as it wasn't tested as thoroughly as I would have liked. If you post the MESA inlist you use to create your stellar model, then I could figure out whether/how MESA can add the double points correctly.

cheers,

Rich

[gali8292]

Hi GYRE staff,

When I check gyre output, I found an annoying bug.

l m n_pg Re(freq) Im(freq) Delta_g
1 0 -50 0.5008682947046021E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -49 0.5109359500054714E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -48 0.5214721525874243E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -47 0.5324933402526769E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -46 0.5440221467085419E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -33 0.5560894673277147E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5687164858455284E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -43 0.5819431235954060E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5958068775113836E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -41 0.6103520639155496E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -40 0.6256287702993355E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -39 0.6416841190403432E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -38 0.6585664051204567E+000 0.0000000000000000E+000 0.7464874418973579E+000

Those two bold lines show abnormal radial orders. I have no idea what happen. The MESA profile and gyre.in are attached.

Cheers

Gang

Abnormal radial orders are often a sign that your stellar model harbors density discontinuities. Looking at the rho(r) profile for your model, I can see that this is indeed the case -- there's a jump in the density around r/R ~ 0.1.

To handle density discontinuities properly, GYRE needs a model containing a double point at the discontinuity (i.e., two adjacent points with the same radius values but different density values). You can instruct MESA to write out models containing double points by using the add_double_points_to_pulse_data and threshold_grad_mu_for_double_point controls (and possibly others).

However, that said, there may still be some issues with MESA's double point creation, as it wasn't tested as thoroughly as I would have liked. If you post the MESA inlist you use to create your stellar model, then I could figure out whether/how MESA can add the double points correctly.

cheers,

Rich

Hi Rich,

Many thanks. Here is my MESA inlist, please help me add the double points. The inlist is designed for gamma doradus variable, if you find anything not correct, please also point them out.

Cheers

Gang

[gali8292]

Hi GYRE staff,

When I check gyre output, I found an annoying bug.

l m n_pg Re(freq) Im(freq) Delta_g
1 0 -50 0.5008682947046021E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -49 0.5109359500054714E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -48 0.5214721525874243E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -47 0.5324933402526769E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -46 0.5440221467085419E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -33 0.5560894673277147E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5687164858455284E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -43 0.5819431235954060E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -42 0.5958068775113836E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -41 0.6103520639155496E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -40 0.6256287702993355E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -39 0.6416841190403432E+000 0.0000000000000000E+000 0.7464874418973579E+000
1 0 -38 0.6585664051204567E+000 0.0000000000000000E+000 0.7464874418973579E+000

Those two bold lines show abnormal radial orders. I have no idea what happen. The MESA profile and gyre.in are attached.

Cheers

Gang

Abnormal radial orders are often a sign that your stellar model harbors density discontinuities. Looking at the rho(r) profile for your model, I can see that this is indeed the case -- there's a jump in the density around r/R ~ 0.1.

To handle density discontinuities properly, GYRE needs a model containing a double point at the discontinuity (i.e., two adjacent points with the same radius values but different density values). You can instruct MESA to write out models containing double points by using the add_double_points_to_pulse_data and threshold_grad_mu_for_double_point controls (and possibly others).

However, that said, there may still be some issues with MESA's double point creation, as it wasn't tested as thoroughly as I would have liked. If you post the MESA inlist you use to create your stellar model, then I could figure out whether/how MESA can add the double points correctly.

cheers,

Rich

HI Rich,

Did you get any progress?

Cheers,

Gang

[rhtownsend]

HI Rich,

Did you get any progress?

Cheers,

Gang

Yep, I did!

You need to edit your MESA inlist to add the following two lines to the &controls namelist group:

Code: Select all

   mesh_delta_coeff = 0.5    ! Ensure adequate grid resolution
   add_double_points_to_pulse_data = .true.    ! Add double points at density discontinuities

Then, re-run MESA and GYRE. This should fix the radial order issue.

cheers,

Rich

[gali8292]

HI Rich,

Did you get any progress?

Cheers,

Gang

Yep, I did!

You need to edit your MESA inlist to add the following two lines to the &controls namelist group:

Code: Select all

   mesh_delta_coeff = 0.5    ! Ensure adequate grid resolution
   add_double_points_to_pulse_data = .true.    ! Add double points at density discontinuities

Then, re-run MESA and GYRE. This should fix the radial order issue.

cheers,

Rich

Hi Rich,

But the results are still not perfect...

Look at those three attached pdf files. They are from three continuous profiles from MESA. However, There are extremely sharp dips in 'profile927.pdf' and 'profile925.pdf', and the period spacing pattern in profile926.pdf changed suddenly.

Cheers,

Li Gang

[rhtownsend]

HI Rich,

Did you get any progress?

Cheers,

Gang

Yep, I did!

You need to edit your MESA inlist to add the following two lines to the &controls namelist group:

Code: Select all

   mesh_delta_coeff = 0.5    ! Ensure adequate grid resolution
   add_double_points_to_pulse_data = .true.    ! Add double points at density discontinuities

Then, re-run MESA and GYRE. This should fix the radial order issue.

cheers,

Rich

Hi Rich,

But the results are still not perfect...

Look at those three attached pdf files. They are from three continuous profiles from MESA. However, There are extremely sharp dips in 'profile927.pdf' and 'profile925.pdf', and the period spacing pattern in profile926.pdf changed suddenly.

Cheers,

Li Gang

Why do you think the results are wrong?

[gali8292]

Yep, I did!

You need to edit your MESA inlist to add the following two lines to the &controls namelist group:

Code: Select all

   mesh_delta_coeff = 0.5    ! Ensure adequate grid resolution
   add_double_points_to_pulse_data = .true.    ! Add double points at density discontinuities

Then, re-run MESA and GYRE. This should fix the radial order issue.

cheers,

Rich

Hi Rich,

But the results are still not perfect...

Look at those three attached pdf files. They are from three continuous profiles from MESA. However, There are extremely sharp dips in 'profile927.pdf' and 'profile925.pdf', and the period spacing pattern in profile926.pdf changed suddenly.

Cheers,

Li Gang

Why do you think the results are wrong?

The dips are too sharp aren't they? The previous literature ( Bouabid 2013, Miglio 2008 for theory and Van Reeth 2015 for observation) never show such sharp and deep glitches. And the diagram in profile926.pdf should be similar to other two diagrams.

Cheers,

Gang

[rhtownsend]

Hi Rich,

But the results are still not perfect...

Look at those three attached pdf files. They are from three continuous profiles from MESA. However, There are extremely sharp dips in 'profile927.pdf' and 'profile925.pdf', and the period spacing pattern in profile926.pdf changed suddenly.

Cheers,

Li Gang

Why do you think the results are wrong?

The dips are too sharp aren't they? The previous literature ( Bouabid 2013, Miglio 2008 for theory and Van Reeth 2015 for observation) never show such sharp and deep glitches. And the diagram in profile926.pdf should be similar to other two diagrams.

Cheers,

Gang

Hi Gang --

It looks like the sharp dip in the period spacings, for 925 and 927, are a result of an avoided crossing between modes trapped in different parts of the star. I'm not altogether sure what's happening with 926 -- you might want to inspect eigenfunctions for some of the modes to see if they look reasonable.

But more generally, I'm yet to be convinced that the behavior you're seeing is due to a bug in GYRE, or -- as was the case previously -- due to an inadequacy in the input model. If you can present clear evidence that there's a bug in GYRE, then I'm happy to work on fixing it. However, I'm afraid I don't have the time to fix issues with the input model -- you'll have to work on it yourself, and/or seek help elsewhere.

cheers,

Rich

[gali8292]

Why do you think the results are wrong?

The dips are too sharp aren't they? The previous literature ( Bouabid 2013, Miglio 2008 for theory and Van Reeth 2015 for observation) never show such sharp and deep glitches. And the diagram in profile926.pdf should be similar to other two diagrams.

Cheers,

Gang

Hi Gang --

It looks like the sharp dip in the period spacings, for 925 and 927, are a result of an avoided crossing between modes trapped in different parts of the star. I'm not altogether sure what's happening with 926 -- you might want to inspect eigenfunctions for some of the modes to see if they look reasonable.

But more generally, I'm yet to be convinced that the behavior you're seeing is due to a bug in GYRE, or -- as was the case previously -- due to an inadequacy in the input model. If you can present clear evidence that there's a bug in GYRE, then I'm happy to work on fixing it. However, I'm afraid I don't have the time to fix issues with the input model -- you'll have to work on it yourself, and/or seek help elsewhere.

cheers,

Rich

Hi Rich,

I solved that problem! We should use 'mesh grid' correctly to avoid it. The MESA inlist from C. Aerts in MESA summer school 2016 is a good guidance.

Cheers

Gang