[CIG-SHORT] Updating CIG short-term tectonics priorities

[Bill Appelbe]

Hi Brad et al:

On 17/05/2007, at 2:58 AM, Brad Aagaard wrote:

> Hi all:
>
> NOTE: Feedback is requested BEFORE Mon May 21.
>
> The CIG Science Steering Committee will be meeting next Mon and Tue  
> to revise
> the rolling five-year Strategic Plan. There are currently two years  
> left in
> the (initial) NSF funding. The SSC would like to shift from relatively
> independent software development in each of the working groups  
> towards a more
> synergistic approach. This synergistic approach will likely involve  
> common
> science goals (for example, driving a crustal deformation model  
> with a mantle
> convection model) and common software implementation goals (for  
> example, use
> of common components).

We would certainly support this. There are two components to what you  
are proposing:
a)	Common science goals

b)	Common software implementation goals

I think that under a) there is also a question of the extent to which  
you want to drive it by specific models or benchmarks, and the extent  
to which those models reflect trying to model very specific   
geological systems (for many years we have used the "Grand Challenge"  
example of being able to model the evolution of the Tibetan plateau  
for example).  My personal opinion (and that of a non-expert  
geologist, just a sw. developer) is that the more concrete the  
benchmark or physical model that can be agreed on the better the  
outcomes (as opposed to rather general goals such as coupling two  
models without some specific benchmarks). It is probably also true to  
say that you are not going to create a single common goal, but  
instead try to have a "bottom-up" approach of a few shared goals.

Under b) I think that there is a fair amount of convergence already  
going on across several projects: PyLith, GALE, and the Magma code.  
Ourselves and Matt/ANL are all about building our software with well- 
defined interfaces, that make it possible to couple codes. At this  
early stage of evolution of codes, that makes coupling possible. It  
is really premature to try to define or enforce a common architecture  
or interface standard across all components, as in CCA).

Another set of issues are also important:

c)	Documentation
This is a key issue for GALE/Magma, and anything using our StGermain  
framework. For Magma, we are going down the route of building  
"cookbook" documentation, as we develop the code, analogous to the  
current GALE documentation, but with more detail on the low-level  
facilities of StGermain (the analogy is Knuth's Tex documentation).   
This is a lot of work, but important to make sure that more  
sophisticated codes reach a wider user base of non-experts.

d)	Scaleability
This is a key issue for us and others as well. Credibility in the  
Terascale computing community means scaling out to hundreds and  
thousands of processors on very large problems, and that means 3D  
decompostions. Scaling has an associated problem of load balancing  
for the type of non-homogenous dynamic problems that are typical in  
geosciences.  This all ties in with AMR, etc. This is a lot of work.

e)	Useability
If we want models to be used by non-experts, then we need to have  
high-level scripting (whether in XML, Python, or Matlab is an open  
question). This too is a lot of work

Between a)-e) I think SSC needs to decide on priorities, as any one  
of these areas is a fair amount of effort from a relatively small  
budget.  But I should add that we have ongoing funding from the  
Australian Government for development at Monash/VPAC for at least the  
next few years for our side of the software for collaborative projects.

My 2c worth -- Bill

>
> I would appreciate feedback/discussion on priorities for short-term  
> tectonics
> within CIG in the context of greater synergy across the working  
> groups.
> Specifically, prioritizing the list of ideas below. Of course, new  
> ideas are
> also welcome. Below I have listed some of the ideas proposed over  
> the last
> year for upcoming priorities as well as an update on PyLith  
> development.
>
> PLEASE PROVIDE FEEBACK BY SENDING EMAIL TO cig-short at geodynamics.org.
>
> Thanks,
> Brad
>
> IDEAS FOR SYNERGISTIC ACTIVITIES
> a. Adaptive mesh refinement
> b. Common interface/implementation of physical property databases
> c. Coupling models of different physical processes (e.g., mantle  
> convection
> and crustal deformation)
> d. Common software components (use of PETSc solvers, Pyre  
> framework, build
> systems) [this is more of an implementation issue]
> e. Benchmarking tools
>
> PYLITH DEVELOPMENT UPDATE
>
> Version 1.0 (planned release at CFEM workshop)
>
> Summary: Initial release of code for both quasi-static and dynamic  
> modeling.
> This is a completely new version of PyLith written in C++ and  
> Python which
> combines the functionality of PyLith 0.8 and EqSim. User input is
> dramatically simplified via direct import of meshes from CUBIT and  
> LaGriT and
> use of a spatial database to specify all boundary conditions.
>
> Features:
>   * Solves quasi-static or dynamic deformation problems using  
> implicit and
> explicit time-stepping schemes, respectively.
>   * 2-D and 3-D Cartesian geometry (1-D also included but not very  
> useful)
>   * Dirichlet boundary conditions
>   * Kinematic fault boundary conditions (specification of slip as a  
> function
> of time)
>   * Elastic and linear Maxwell viscoelastic bulk constitutive models
>   * Output in VTK files
>   * Runs in parallel
>
> Version 1.1 (planned release in late Summer 2007)
>
> New features:
>   * Dynamic fault boundary conditions (fault constitutive models)
>   * Nonlinear viscoelastic bulk constitutive models
>   * Absorbing boundary conditions (may be included in version 1.0)
>   * Traction boundary conditions (may be included in version 1.0)
>   * Output to HDF5 files
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short

         Cheers -- Bill

------------------------------------------------------------
Bill Appelbe                           +61 3 9925 4648
VPAC                      (mobile)+61 0418 557115
                                       (fax)   +61 3 9925 4647
www.vpac.org                             bill at vpac.org


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geodynamics.org/pipermail/cig-short/attachments/20070521/51007da8/attachment.htm