Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Inconsistency between 15140 and 23.05.1-rc2: emergence of pre TP-AGB oscillatory feature #543

Open
giulia-cinquegrana opened this issue Jun 1, 2023 · 16 comments
Open

Inconsistency between 15140 and 23.05.1-rc2: emergence of pre TP-AGB oscillatory feature #543

giulia-cinquegrana opened this issue Jun 1, 2023 · 16 comments

Comments

@giulia-cinquegrana
Copy link

Discovered with @mjoyceGR, following the issue #537 raised by @mjoyceGR during attempts to replicate results from TP-AGB inlists used with version 15140 with version 23.05.1

Using supposedly identical physics, we see the emergence of a pulsation feature prior to the TP-AGB when using version 23.05.1. This feature is not present when using 15140 (see three figures attached).

We attach also two sets of inlists: one pair from 15140 and one pair or 23.05.1, inlist_common is identical in both.

The changes in inlist_TP_AGB are:

(1) to the H limit controls discussed in issue #537

(2) the need to comment out the following controls, present in 15140,

   dH_div_H_limit_min_H = 1d-5
   dH_div_H_limit = 0.5

for the 23.05.01 model to run.

@evbauer suggested looking into TDC, we do not think this is likely the issue given that Henyey MLT was explicitly defined in the new version of the inlists. Any thoughts as to what has caused this change? Has there been significant adjustment to how resolution controls are handled since 15140?

snellar_1
snellar_2
snellar_3_thesnail

inlists.zip
@mjoyceGR mjoyceGR self-assigned this Jun 1, 2023
@mjoyceGR
Copy link
Member

mjoyceGR commented Jun 1, 2023

Additional context:

  • in both cases, the same externally-built AESOPUS low-T opacity tables are used

@mjoyceGR mjoyceGR removed their assignment Jun 1, 2023
@evbauer
Copy link
Member

evbauer commented Jun 1, 2023

To get resolution controls equivalent to what you had in 15140, I think you need to add this into your new inlist:

dX_limit(1) = 0.1
dX_div_X_limit_min_X(1) = 1d-5
dX_div_X_limit(1) = 0.5

If that still doesn't resolve the discrepancy, I think it would be useful to check if this issue also shows up in r22.05.1 or r22.11.1. That way we could get a rough sense of when the change occurred.

@Debraheem
Copy link
Member

Debraheem commented Jun 1, 2023 •
edited
Loading

It looks to me like you might be resolving He-subflashes in rc23.05.1. Here(edit fixed the link) is a movie of a solar Z, 2.1Msun model encountering sub-flashes in r15140 before going up the RGB, look similar?

This could indeed be due to differences in timestep resolution between versions. Maybe check to see if there is any oscillation in log_Lneu in tandem with log_L. N-alpha captures during the sub-flash produce strong neutrino signals through the beta+ decay of f18 -> o18. (or just look at log_LHe)

Maybe Evan's suggestion will fix this by matching the timestep resolution between versions.

Another test is to re-run the r15140 model with delta_lgL_He_limit = 0.01 -->0.005 (or lower) and maybe see if you can resolve these sub-flashes in r15140.

p.s. I tested your r15140 inlists, and I get the same result. I will test the rc23.05.1 version with evan's suggestion next.

@Debraheem
Copy link
Member

Debraheem commented Jun 2, 2023 •
edited
Loading

I ran the inlists provided through both versions exactly as provided with the edits @evbauer suggested, and I can reproduce the same results, confirming that there are sub-flashes in rc23.05.1, but not in r15140 (among seemingly identical inlists). it seems like it could be something time-step related, as @giulia-cinquegrana and @mjoyceGR have pointed out. However, I could imagine a situation where slight changes in the default reaction rates adopted between both versions would result in a slightly different evolution through this territory (especially when it comes to the 3-alpha, c12ag rates, which I believe are different between r15140 and r23.05.1). Another difference is that the default solver tolerance level was changed from ( use_gold2_tolerances = .true.) to ( use_gold2_tolerances = .false.) from r22.11.1 to rc.23.05.1 (just another thing to check).

I'm quite curious what is ultimately causing these differences!

here is r15140:
r15140

here is rc23.05.1:
rc23 05 1

One drawback of these inlists is that they do not fully capture the weak reactions that occur during He-burning so the log_Lneu value is not to be trusted. For example, there should be a spike in log_Lneu during the He flash to at least log_Lneu ~ 7.

A note on resolution for He burning: If you decrease delta_lgL_He_limit -> 1d-3, you can watch the off-center He flash propagate inward in multiple sub-flashes, followed by the off-center flashes near the outside of the core, before star climbs the RGB. At this resolution, you might also notice that the thermal-pulses also show multiple subflashes in the interpulse period between each pulse as opposed to the single flash present at lower resolutions.

(MESA model directories run are attached)
model_directories.zip

@giulia-cinquegrana
Copy link
Author

Thanks @evbauer for the suggestion-I'm running those tests now.

@Debraheem cheers for running them. Are you referring to a specific example of he subflashes on the early agb? And what's your definition of sub flashes? Yeah of course, there are toss ups for TP-AGB models because we want them to be able to converge with a rate faster than 1 TP / year haha, that was my initial aim with this inlist.

@Debraheem
Copy link
Member

Are you referring to a specific example of he subflashes on the early agb?

What statement are you referring to? Sorry, I know I wrote a lot. In general I was detailing that the He flash is actually composed of multiple flashes, a strong flash and multiple weaker "sub-flashes" that propagate inward. After this, there are some few weak sub-flashes that can occur at the outer edge of the He-core, which induces the oscillation in log_L you're observing, after which He-burning carries on quiescently. All this is pre TP-AGB.

And what's your definition of sub flashes?

I'm referring to any explosive he-ignition event aside from the initial He-flash, e.g. any time log_LHe increases rapidly from He-burning. You can count each spike in log_LHe from the Kipp.png as a sub-flash (followed by convection). Is this clear?

Yeah of course, there are toss ups for TP-AGB models because we want them to be able to converge with a rate faster than 1 TP / year haha, that was my initial aim with this inlist.

Totally understandable, I'm in agreement!

@giulia-cinquegrana
Copy link
Author

In general I was detailing that the He flash is actually composed of multiple flashes, a strong flash and multiple weaker "sub-flashes" that propagate inward.

Yeah agree, I'm happy for oscillations to occur around the actual helium shell flashes (just noting that the he flashes on the TP-AGB are not the same physical mechanism as the core he flash). But the definition of the outset of the TP-AGB is when the model undergoes the first thin shell instability flash, the early AGB is meant to be a period of relatively quiescent He shell burning before it undergoes the first pulse. So just based on my experience its a bit weird to me!

I'm referring to any explosive he-ignition event aside from the initial He-flash, e.g. any time log_LHe increases rapidly from He-burning. You can count each spike in log_LHe from the Kipp.png as a sub-flash (followed by convection). Is this clear?

Okay cool, I wasn't sure if you referring to a particular threshold definition.

@Debraheem
Copy link
Member

Debraheem commented Jun 3, 2023 via email

@giulia-cinquegrana
Copy link
Author

since the oscillations you are seeing precede he-core burning (the he-flash burns very little helium). They are the remainders of the he-flash!

I'm not sure we're talking about the same thing. The oscillations above are post core He burning. The model depletes central He, contracts, but is not massive enough to ignite core carbon burning so ignites He in a shell instead outside of the CO core as it ascends the e-agb.

@Debraheem
Copy link
Member

Debraheem commented Jun 3, 2023 via email

@giulia-cinquegrana
Copy link
Author

No stress! It's always helpful to talk out the physics expectations anyway.

@Debraheem
Copy link
Member

Debraheem commented Jun 3, 2023 •
edited
Loading

Just to quickly recap! We have He-core flashes, these strange He-shell flashes at the onset of the agb during he-shell burning, and then thin He-shell Thermal-pulses later on right?

Not to muddy the waters of this thread, but I just wanted to quickly note: In models with convective overshooting, I do not encounter any of these He-shell flashes at the onset of He-shell burning, regardless of initial mass . Instead, I get quiescent he-shell burning as you expected. I don't know if this is maybe some indication that overshoot or some other mixing mechanism is needed, but their inclusion might remove this feature. Nonetheless, thanks for discussing this topic with me! This conversation can certainly be continued outside of this thread.

@giulia-cinquegrana
Copy link
Author

Just to quickly recap! We have He-core flashes, these strange He-shell flashes at the onset of the agb during he-shell burning, and then thin He-shell Thermal-pulses later on right?

Yup! It's the:

strange He-shell flashes at the onset of the agb during he-shell burning

that I'm not sure about.

In models with convective overshooting, I do not encounter any of these He-shell flashes at the onset of He-shell burning, regardless of initial mass

Oh that's really interesting, I didn't include overshoot in this model because (a) it didn't make a difference to whether the model experienced the third dredge up and (b) I wouldn't expect this model to, based on its mass and initial Z. But overshoot was necessary in higher mass models to get them through any dredge up when they should. So maybe that's an indication of an issue with the location of the base of the convective envelope when just using schwarschild. I'll run some tests with and without overshoot and see if there is a big difference in where the base of the envelope lies. Cheers @Debraheem, always happy to chat about AGBs.

@Debraheem
Copy link
Member

Debraheem commented Jun 3, 2023 •
edited
Loading

So maybe that's an indication of an issue with the location of the base of the convective envelope when just using schwarschild.

This 2.1Msun solar Z model (r15140) was run with ledoux and predictive mixing, and overshoot only during H-burning, but still encounters these He-shell burning flashes. Looking more closely at the movie, it looks like the flashes only occur at the onset of He-shell burning, and that the burning does indeed proceed quiescently afterward.

@mjoyceGR
Copy link
Member

mjoyceGR commented May 1, 2024

This issue is very nearly a year old (!), but I happened to come across some useful literature on this feature a few days ago, as part of an unrelated investigation. Just sharing here in case it's helpful; no action needed.

https://ui.adsabs.harvard.edu/abs/2007A%26A...471..911G/abstract
https://articles.adsabs.harvard.edu/pdf/1986ApJ...308..706M
https://arxiv.org/pdf/1502.03318

Of course, this does not address the issue of inconsistency between MESA versions, but maybe we don't care about that...would have to check against 24.03.1 at this stage anyway, which I don't think is a priority right now.

@Debraheem
Copy link
Member

@mjoyceGR Lovely papers, thanks for sharing!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
4 participants
@mjoyceGR @evbauer @Debraheem @giulia-cinquegrana