Dr. Steven Owen of Sandia National Laboratories continued his tradition
of leading the course "Introduction to Unstructured Mesh Generation".
This year's course was split into two sessions. The "Algorithms"
session covered tri and tet meshing by Delaunay, advancing front, and
octree methods; quad and hex meshing by medial axis, paving, q-morph,
sub-mapping, plastering, sweeping, and whisker weaving methods; and
mixed element methods. The "Real World Meshing" session covered
advanced topics such as geometry models, smoothing, mesh topology
improvements via edge flipping and tranformations, and mesh sizing
Dr. Peter Schroeder of the California Institute of Technology offered
a course on Digital Geometry Processing from the viewpoint of computer
"Global Optimization of Mesh Quality" was led by Dr. David Eppstein of
the University of California, Irvine. His course covered issues mostly
relating to Delaunay-generated meshes including connectivity
optimization, global point placement, and individual point placement
(for sliver exudation).
The last course of the day was Dr. Stephen Vavasis' "Optimality and
Guaranteed Quality in Isoparametric Mesh Generation". Dr. Vavasis
(Cornell University) presented mesh quality results for higher order
Mr. Jim Rusk, Vice President of Digital Validation Solutions for EDS'
PLM Solutions (EDS Product Lifecycle Management Solutions is the
business entity resulting from the acquisition of SDRC by UGS and the
subsequent consolidation of UGS under the EDS name) kicked off the
technical program with his keynote presentation on how meshing and
simulation fit into the much larger picture of product lifecycle
management. Issues such as automation, just-in-time meshing, data
archiving, and geometry abstraction are seen as important to meshing
within a digital enterprise.
Invited talks were also given by Drs. Andreas Fabri and Timothy Baker.
Dr. Fabri (INRIA) presented an overview of the CGAL Project
(www.cgal.org), a European collaborative effort to develop a
library of computational geometry software exhibiting robustness,
efficiency, and flexibility. The software, implemented in C++ and
featuring generic programming via the STL (Standard Template Library)
and "exact" computing, has now reached a stable phase and is being
used in teaching, in CG research, and in endeavors as diverse as
collision detection in computer games and geological modeling. A
transition to an open source model is being considered as is the
beginning of commercially available CGAL support.
Dr. Baker's presentation on "Mesh Movement and Metamorphosis" delved
into mesh adaption for time evolving domains including enrichment,
coarsening, and stretching (r-refinement) as required. Enrichment is
required when the local length scale exceeds the length density
function derived from the boundary length scales. Coarsening is
currently implemented via edge collapsing. Maintaining mesh quality
is paramount for successful implementation of the entire process and
was demonstrated with video displays of a 3D mesh of the human heart
muscle in motion.
The remaining technical presentations ran the gamut of subjects
including surface meshing (9), volume meshing (11), geometry (5),
mesh quality (3), and software issues & applications (7). The bound
proceedings are available from Sandia National Labs (publication
The highlight of every IMR is the Poster Session from which three award
winners are chosen annually: Meshing Maestro, Best Technical Poster, and
Best Student Poster. The Meshing Maestro is chosen from the ballots
cast by all of the conference attendees and represents the "best
looking mesh". This year's winning mesh, by Ted Blacker from Fluent,
will be featured on next year's conference t-shirt. Dr. Blacker also
was awarded the Best Technical Poster award. Mike Borden from BYU
was awarded Best Student Poster.
Birds and Panels
The two most lively sessions of each IMR are the Birds of a Feather
(BOF) and Panel Sessions. The BOF Session brings together researchers
with a common interest in a meshing subtopic to assess recent progress,
current state of the art, and open issues within their speciality.
This year's BOFs included surface meshing, volume meshing, mesh
quality, geometry issues, and software issues. The complete BOF
summaries will soon be presented on the IMR web site. Some of the
interesting problems needing a solution are surface mesh reconstruction
from sampling, re-parametrization of surface meshes, dynamic meshes,
surface adjustments from hex and tet meshing, adaption, and a posteriori
The Panel Session began with a brief introduction to NAFEMS, the
International Association for the Engineering Analysis Community given
by Prof. Cecil Armstrong, a member of the CAD/FEA Integration committee.
NAFEMS' goals to create awareness, deliver training, and stimulate
standards may serve as a foundation between the FEA and meshing
communities to meet the challenges faced by this technology.
The Roundtable ended with a panel discussion of two "hot" topics
within the meshing community: "What Makes a Mesh Good?" and
"What's the Future of Geometry?: NURBs or Faceted Models." The
panelists included Cecil Armstrong (Queen's University of Belfast),
Gordon Ferguson (Visual Kinematics), Pat Knupp (Sandia), Peter Schroeder
(Caltech), Tim Tautges (Sandia), and Steve Vavasis (Cornell). The
"good mesh" topic got things off to a good start by making clear that
although a posterior quality improvement via adaption would be best,
a priori quality metrics are needed now simply because adaption isn't
ready for prime time. So while it is apparent that work in the area of
adaption is required, no consensus was reached with respect to the use
of a priori quality metrics with the exception of guidelines for
ensuring that metrics are well-formed.
On the other hand, the "future of geometry" drew a near consensus from
the panel: faceted geometry will coexist with NURB geometry in the near
future due to its advantages in terms of simplicity and flexbility.
Faceted geometry also allows the meshing community to leverage its
existing technology for working with discrete data. Peter Schroeder
went further by predicting that faceted models will do to NURB models
what the PC did to Unix workstations. Some drawbacks of faceted models
were identified, though, including their relatively heavy "weight" and
the lack of data standards for facets.
More information on the 10th International Meshing Roundtable can be
found on the conference web site at http://www.imr.sandia.gov/. Watch
this web site for news on the IMR for 2002.