[aspect-devel] Problem with active tracers & compositional fields

[John Naliboff]

Hi Felipe,

To add on what Wolfgang outlined, you can also use the “DG" (Discontinuous Galerkin) discretization (+/- limiter) to help maintain sharp interfaces. 

There are a number of ASPECT papers that track distinct compositions and even compare different methods (compositional fields verse particles, etc). I suggest going through these papers (https://aspect.geodynamics.org/publications.html <https://aspect.geodynamics.org/publications.html>) before running too many simulations, as they likely have implications for what method is most suitable for your specific problem.

Cheers,
John



> On May 20, 2018, at 2:05 AM, Wolfgang Bangerth <bangerth at colostate.edu> wrote:
> 
> 
> Felipe,
> 
>> For a target material body (e.g. a sphere) that has a compositional-density contrast with respect to the environment, and having the system convecting (by any means), what is the difference (in output, usefulness, etc.) between:
>>    a) Having the system domain fully mapped by active tracers and compositional fields, some of which will track the kinematic evolution of the target body.
>>    b) Having the system defined by compositional fields, and having the target body mapped by passive tracers that track its kinematic evolution.
> 
> Using particles or fields achieves basically the same thing.
> 
> Traditionally, the mantle convection community has used particles, and so that's what you can do in ASPECT as well. On the upside, it allows you to keep interfaces sharp. On the downside, it may lead to poorly load-balanced problems if you work on large parallel computations. It also requires interpolating particle properties from particle locations to the locations of quadrature points, and this may introduce an error.
> 
> On the other hand, compositional fields are always well load balanced and don't require this interpolation, but they do introduce diffusion.
> 
> So it is essentially a *choice* whether you want to go with one or the other. You should give both a try.
> 
> Best
> W.
> 
> -- 
> ------------------------------------------------------------------------
> Wolfgang Bangerth          email:                 bangerth at colostate.edu
>                           www: http://www.math.colostate.edu/~bangerth/
> 
> _______________________________________________
> Aspect-devel mailing list
> Aspect-devel at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/aspect-devel/attachments/20180520/1debbfa3/attachment-0001.html>