[mesa-users] Timestep control

[Warrick Ball]

Hi all,

I wouldn't use the name

   dt_div_acoustic_radius

because the acoustic radius is a radius, not a time, so this number isn't 
a ratio.  I'd prefer* something like

   dt_div_sound_crossing_time

although it's more cumbersome.  Maybe invent

   dt_div_dt_sound

?

How different will this be from the dynamical time anyway?  I haven't 
personally come across stars where the sound crossing time doesn' scale 
(roughly) with the dynamical time.

Finally, I like some of the subsequent ideas in this thread.  Are there 
already hooks for the maximum timestep?  Or is it enough simply to add 
something like

   s% dt = min(dt, my_dt_limit)

to extras_check_model in run_star_extras?

Cheers,
Warrick

* Not that this is a democracy. :p

On Tue, 26 Aug 2014, Bill Paxton wrote:

> Hi Mathieu,
>
> Okay: let's add some more controls explicitly based on timescales.
> These need to be stated in the form of a ratio of timestep (dt) to timescale.
> If the ratio is too large, then the "select timestep" routine reduces the next dt.
>
> [let's not get into debates about order unity prefactors
> as long as we are explicit about the definitions we are using.]
>
> "global" timescales for entire star:
> dt_div_dt_thermal
> 	! dt_thermal = (3/4)*G*M^2/(R*L); Kelin-Helmholtz time
> dt_div_dt_dynamic
> 	! dt_dynamic = 2*Pi*sqrt(R^3/(G*M))
> dt_div_acoustic_radius
> 	! acoustic_radius = time for sound from center to photosphere
> 	! = sum over shells of local sound crossing time dr/csound.
> dt_div_dt_mass_loss
> 	! dt_mass_loss = -M/Mdot; only applies when Mdot < 0
>
> any other global timescales we should add?
>
>
> "local" timescales for shells of model:
> dt_div_dt_Courant
> 	! dt_Courant = min over shells of sound crossing time, dr/csound
>
> any other local timescales we should add?
>
>
> For the next release, I'll add at least these,
> and if anyone has other suggestions, please speak up now.
>
> Cheers,
> Bill
>
>
>
> On Aug 26, 2014, at 6:48 AM, Mathieu wrote:
>
>> Hi everyone,
>>
>> I have a (probably silly) doubt on the timestep control MESA does by
>> default.
>> If I am correct, in the two-stage process of picking the new timestep
>> there isn't any explicit comparison between the timestep, the thermal
>> timescale, mass loss timescale and other relevant time scales? So
>> nothing forces the timestep to be smaller than, say, the thermal timescale?
>> Or is the comparison implicitly done by the digital control-based scheme
>> (I am still a bit confused by this)?
>>
>> Thanks,
>>
>> Mathieu
>>
>>
>> ------------------------------------------------------------------------------
>> Slashdot TV.
>> Video for Nerds.  Stuff that matters.
>> http://tv.slashdot.org/
>> _______________________________________________
>> mesa-users mailing list
>> mesa-users at lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/mesa-users
>
>
>
>


------------
Warrick Ball
Postdoc, Institut für Astrophysik Göttingen
wball at astro.physik.uni-goettingen.de
+49 (0) 551 39 5069