[mesa-users] Calibration of Kjeldsen-Bedding-JCD surface effect power law
Laurent Piau
lrnt_p at yahoo.com
Wed Nov 13 11:17:11 EST 2013
Hi Warrick,
I have two comments about your recent questions on correction
of the surface effects in solar eigenfrequencies calculations.
1) You report a systematic obs/model difference of ~0.5 muHz below
2000 muHz.
In this low frequency regime I have not seen a difference
larger than 0.1 muHz. You can have a look at to Fig 6 and 7
of Piau et al. (2013). Thecorresponding computed frequencies have not
been corrected by the method addressed by K08.
So I expect that either this 0.5 muHz is due to the K08 correction
or that the solar model you use is slightly too cool in
internal regions. This could be related to the microscopic diffusion
process (not efficient enough?). If the model is too cool the
neutrino flux will be below observations.
2) It is true the b value you get is only for the Sun. Yet
K08 suggested that the same solar b allows to obtain the same
average density from models of AlphaCenA,B and BetaHyd.
This supports that a solar b value has some predictive power
for the physics of other stars.
I hope it helps!
I m interested to know how the situatio evolves.
Cheers,
Laurent.
--------------------------------------------
On Mon, 11/11/13, Warrick Ball <wball at astro.physik.uni-goettingen.de> wrote:
Subject: Re: [mesa-users] Calibration of Kjeldsen-Bedding-JCD surface effect power law
To: stello at physics.usyd.edu.au
Cc: mesa-users at lists.sourceforge.net
Date: Monday, November 11, 2013, 5:24 AM
Dennis,
> Your approach is a sensible one to take [using MESA to
find a solar model at find 'b' by comparing it to, say,
BiSON]. However, one should note that just because
b=<your fit result for the Sun> works for the Sun,
does not mean it is the most appropriate b-value for another
star....not to mention using a power law to correct the
freqs.
> What I am trying to say is: there is a limit to which
one should seek to obtain the 'right' b-parameter.
Yip, I agree completely. What I'm trying to work out
is what is a "best" b-parameter. If such a thing
really exists! :p
> That said, which solar calibration should you use? MESA
now has a second option for doing a solar calibration, which
is in 'astero'. [dont have access to my MESA stuff so
can give you details]. But you could try use both
options in MESA to make your own solar calibrations and see
how much that in the end changes your b-parameter.
I've tried two options. First, there's a test case
called "example_solar_model", which looks like it's a
previously calibrated model. That's what I sent.
I've also tried running the calibration in the test case
"solar_calibration", which uses astero_search_controls, with
the BOBYQA minimization algorithm. I've attached the
best-fitting model from that, for comparison, but the
frequencies are about the same because the final parameters
are also basically the same.
The default inputs are a mix of non-seismic
constraints. It looks like a combination of the
solar-specific options (like Y=0.2485 ± 0.0034, c_s_rms = 0
± 2d-6, R_CZ = 0.713 ± 0.001) and what would be pretty
good spectroscopic constraints for a distant star (e.g.
T_eff = 5777 ± 65). I presume some of the solar
values are actually much better constrained, so I'll play
around with the parameters and see if I can get a
calibration that reproduces the lower frequencies
better. There are some frequencies in there but the
default case has chi2_seismo_fraction = 0.
I'm not quite sure what the second option is. It looks
to me like the separate methods for solar calibration and
seismic model-fitting have been somehow consolidated into
the same exercise?
> Keep playing!
Will do!
Cheers,
Warrick
> Quoting Warrick Ball <wball at astro.physik.uni-goettingen.de>:
>
>> Fellow MESAlians*,
>>
>> I'm trying to recover the calibration of the
Kjeldsen-Bedding-JCD surface-effect law [1] in the
astero_search routines.
>>
>> In
mesa/star/astero/defaults/astero_search_controls.defaults,
one finds
>>
>> ! surface corrections (see K08 --
ApJ 683, pg 175)
>> correction_factor = 0
>> ! use this
fraction of the correction; set to 0 to skip doing
corrections.
>> correction_b = 4.90d0
>>
>> So the surface power law is b=4.90. This is
the canonical value in the paper, based on the calibrations
in their models. It is often used elsewhere.
>>
>> In
mesa/star/test_suite/example_astero/inlist_astero_search_controls,
there is, instead,
>>
>> ! surface corrections (see K08 --
ApJ 683, pg 175)
>> correction_factor = 1
>> ! use this
fraction of the correction; set to 0 to skip doing
corrections.
>> l0_n_obs(:) = -1 ! the observed
radial orders (ignored if < 0)
>> ! the
observed radial orders are used in calculating surface
corrections
>> ! if <= 0,
use default calculation for radial orders
>> correction_b = 4.25d0
>>
>> So b=4.25. Hmmm...
>>
>> For any given evolutionary code, what one should
probably do is use that code to calibrate a solar model,
compute the oscillation frequencies and fit a power law to
the differences with, e.g. the BiSON frequencies [2]. So I
ran mesa/star/test_suite/example_solar_model, saved the
final model as an FGONG file (attached) and handed that to
GYRE, with the JCD boundary condition, for l=0,1,2,3 in the
range 500 to 4500 uHz. I then calculated the
difference between the MESA model and the BiSON
frequencies. A plot is attached.
>>
>> I then tried fitting the parameters myself.
First, I did a fit (with scipy.optimize.leastsq) with all
the observed frequencies, in which case I got b=4.53.
I then did a fit just with the radial modes, in which case I
got b=4.68.
>>
>> What value should one use? Where did b=4.25
in example_astero come from?
>>
>> Now, I realise that this probably isn't a settled
issue. But in the absence of a definitively correct
answer, I'll settle for a well-justified one! :p As a
first note, I know that changing the range of frequencies
over which one fits modifies the power law. Second,
the calibrated model seems to show a small offset of about
0.5 uHz even at the lowest frequencies. Should I add
one or two frequencies to the calibration to try to fix
this, then fit again? Or is the model in
example_solar_model outdated, and I should run the
calibration in solar_calibration myself before
fitting? And should one fit only to the frequencies in
the target of the astero_search? e.g. HD 49835?
>>
>> All input is obviously welcome, and I'm happy to
work on this problem and report to the community how I
progress.
>>
>> Cheers,
>> Warrick
>>
>> * Is this the correct word? Is there an
official stance? :p
>>
>> [1] http://adsabs.harvard.edu/abs/2008ApJ...683L.175K
>> [2] http://adsabs.harvard.edu/abs/2009MNRAS.396L.100B
>>
>>
>> ------------
>> Warrick Ball
>> Postdoc, Institut für Astrophysik Göttingen
>> wball at astro.physik.uni-goettingen.de
>> +49 (0) 551 39 5069
>
>
>
>
----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging
Program.
>
>
------------
Warrick Ball
Postdoc, Institut für Astrophysik Göttingen
wball at astro.physik.uni-goettingen.de
+49 (0) 551 39 5069
-----Inline Attachment Follows-----
------------------------------------------------------------------------------
November Webinars for C, C++, Fortran Developers
Accelerate application performance with scalable programming
models. Explore
techniques for threading, error checking, porting, and
tuning. Get the most
from the latest Intel processors and coprocessors. See
abstracts and register
http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
-----Inline Attachment Follows-----
_______________________________________________
mesa-users mailing list
mesa-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mesa-users
More information about the Mesa-users
mailing list