pothen | 12 Nov 2006 03:12
Favicon

PhD Fellowships in CSC at Old Dominion University


Ph.D.  Fellowships  in CSC at Old Dominion  University

We are recruiting  two  Ph.D. candidates with research interests  in Parallel Computing,  Algorithms, and/or  Scientific Computing,
to join the CSCAPES Institute funded by the Department of Energy's Office of Science.

The CSCAPES (Combinatorial Scientific Computing and Petascale Simulations) Institute, centered at Old Dominion University,
also includes researchers from Sandia and Argonne National Laboratories, and Ohio State  and Colorado State Universities.
We have received funding for the CSCAPES Institute for five years. The research involves developing algorithms for  combinatorial problems arising in
scientific computing, such as load-balancing, graph coloring, graph matching, and automatic differentiation,  on  tera- and peta-scale computers.
Students will be guided in their research by Alex Pothen, Assefaw Gebremedhin, and Florin Dobrian.
They  will have the opportunity to work at  Argonne and Sandia Labs during  the summers, and will  be co-mentored by Lab scientists.
Students will benefit from joint seminars  with CSCAPES member institutions and partner organizations through the Access Grid,
as well as joint workshops with other SciDAC research groups.
Further information about the project is available at http://www.cscapes.org.

Applicants should hold a Master's  degree or equivalent in computer science or computational mathematics.
Exceptionally well-qualified students with Bachelor's degrees will also  be considered. Excellent academic performance and letters of recommendation attesting to
research ability and motivation  are required. The research  fellowships include a competitive stipend, supplemented by full tuition remission and funds for
participating in conferences. Students will be able to pursue their studies in  the Computer Science or Mathematics departments depending on educational background
and interest. Further information about the departments are available at http://www.cs.odu.edu and http://www.math.odu.edu.

The application deadline is January 15, 2007 for students who wish to be considered  for the  Fall 2007 semester (or Summer 2007).
Applications  should include a completed graduate application, three recommendation letters, a brief research statement, and a CV.
See http://admissions.odu.edu/home.php for templates for the application and recommendation letter. Applications could be completed online (preferred)
or submitted as hardcopy.  For questions  about these positions, please contact Alex Pothen (http://www.cs.odu.edu/~pothen, Email: pothen "at" cs.odu.edu).
_______________________________________________
Csc mailing list
Csc@...
http://list.odu.edu/listinfo/csc
pothen | 22 Nov 2006 15:31
Favicon

Static pivoting


From Iain Duff...
Happy Thanksgiving!
--Alex

> ++++++++++++++
> > From isd Thu Nov 16 08:26 GMT 2006
> > Date: Thu, 16 Nov 2006 08:26:29 GMT
> > From: Iain Duff <isd>
> > To: csc-lVPF2Jp+hdv2fBVCVOL8/A@public.gmane.org
> > Subject: Static pivoting
> > Cc: isd
> > X-Status:
> >
> > Sorry about the delay in getting back to the CSC list on the topic of
> > static pivoting but I have just been travelling first in China and now
> > (on a quite separate journey) in Australia.
> >
> > I had many responses to my earlier questions, some offline from CSC.  In
> > fact, it is not really possible to summarize them because many were
> > answering the questions from quite different perspectives.  I will
> > thus try a different track that I have already field tested on some
> > colleagues.
> >
> > I would like to define "static pivoting" as any scheme for selecting pivots
> > in an elimination method that enables a static data structure to be used.
> > For example, in the case of supernodal or multifrontal methods, this would
> > mean that the data structures (eg assembly trees) determined by the
> > analysis phase would be respected by the numerical factorization.
> >
> > Comments please.
> >
> > Best wishes
> >
> > Iain
> >
_______________________________________________
Csc mailing list
Csc@...
http://list.odu.edu/listinfo/csc
Jason Riedy | 22 Nov 2006 19:58
Picon
Favicon

Re: Static pivoting

And Iain Duff writes:
> I would like to define "static pivoting" as any scheme for
> selecting pivots in an elimination method that enables a static
> data structure to be used.

Every pivot method allows for a static data structure: a dense
matrix.

Please let static pivoting remain the static selection of pivots.

We shouldn't conflate the numerical actions with the particular
implementation choices more than absolutely is necessary.

Jason

Gmane