Inconsistent frequencies solved under various "diff_scheme"

[yaguangli]

Hello GYRE users,

I am trying to solve the adiabatic frequencies for a solar model, but I found the frequencies returned from the various schemes (colloc_gl2/colloc_gl4/colloc_gl6/magnus_gl2/magnus_gl4/magnus_gl6) are very different from each other.

This plot shows a comparison between them (the three panels plot the same data but with diff y-axis ranges)

The attachment schemes.png is no longer available

"colloc_gl2" have the most worrisome results. It has a ~5muHz difference at numax and it confuses me with the surface effect at first.
"magnus_gl2" is better but still diff ~0.1muHz at numax, which exceeds the typical observational uncertainty (~0.01muHz).

Is this behaviour intended/unexpected? What do you think the most recommended choice for diff_scheme?
I have attached my gyre inlist and the stellar structural file if you are interested in reproducing my results.
Thanks for any help in advance!

Cheers,
Yaguang

[warrick]

Hi Yaguang,

Can attach or link to the model you're using, as well as provide your GYRE input (and GYRE version if not the latest 6.0.1)?

Cheers,
Warrick

[yaguangli]

Hi Warrick,

Sorry, I think my attachments were mysteriously gone.
Here's a link: https://1drv.ms/u/s!AloivC8jPpg0nKIinCL ... g?e=i27eHg
And yes, I am using 6.0.1.

Thanks!
Yaguang

[warrick]

Here's what I get for ℓ=0,1 and 2 in Model S using the input file in the GYRE tests (i.e. $GYRE_DIR/test/ad/fgong/solar/gyre.in) by cycling through the MAGNUS and COLLOC methods of order 2, 4 and 6. The frequencies are all in μHz. It's qualitatively similar to your plots but the worst errors (for 2nd order methods) are an order of magnitude better.

The attachment gyre_methods0.png is no longer available

gyre_methods0.png (36.62 KiB) Viewed 2836 times

[warrick]

Here's a link: https://1drv.ms/u/s!AloivC8jPpg0nKIinCL ... g?e=i27eHg

That folder appears to be empty. Has the link expired?

[yaguangli]

That folder appears to be empty. Has the link expired?

Sorry, the new link is here:
https://1drv.ms/u/s!AloivC8jPpg0nK1yIHE ... A?e=SgGxUq