Projects - Summer 2001
- AMR Data Volume Visualization, Oliver Kreylos, UC Davis.
-
Adaptive Mesh Refinement (AMR) is a technique to speed up large-scale
PDE simulations. Instead of resolving the domain of the simulation problem
as a single uniform grid, the simulation runs on a hierarchy of grids of
different cell sizes. This way, a grid can be locally refined only where
the PDE solver requires high resolution, and can use coarse resolution
otherwise. This summer project focuses on methods for direct volume
visualization of AMR data. The benefit to the scientific community from
this work will be the ability to directly visualize scalar AMR data
fields without the need for data reformatting or "grid flattening."
- Feature Identification in AMR Vector Fields with Image-based Clustering Techniques, James Chen, UC Davis.
-
The goal of the ongoing research work is to facilitate the identification and
visualization of vector field features, such as vortex tubes, in time-varying
AMR data. As with the multi-resolution nature of the AMR datasets, we would
like to implement the vortex tube visualization capability
in the same fashion. Another desirable feature of the resulting program is
to have the ability to track the movement of features both forwards and
backwards in time.
- Embedded Boundary Extraction and Visualization, Chris Co, UC Davis.
-
Certain classes of local AMR-based simulations are used to model fluid flow
around physical obstacles, such as air flow around an airplane wing. When
the simulation data is transferred to secondary storage, there is no
explicit representation in the dataset for the obstacle's boundary in
the computational grid. The purpose of this project is to derive and
visualize these boundaries from implicit information contained within
the AMR dataset, and to integrate the resulting methods into
ChomboVis,
a production tool for AMR data visualization created by the
Applied Numerical Algorithms
Group at LBL/NERSC.
- Investigatory Work in Visualization Portal Technology, Tom Hsu, San Francisco State University.
-
A web portal is intended to be a "server" that mediates and simplifies
access to a new class of visualization tools, such as LBL's
Visapult. This
project focuses on deployment of an alpha-version web portal that
combines elements of JSP, servlets, ssl in order to provide
secure authentication to "grid-based" visualization resources. This
project enables access to a new incarnation of Visapult through
a portal-based interface. (There is no web page yet for this project).
- Interaction-Accelerated Protein Structure Prediction, Oliver
Kreylos, UC Davis.
- Protein structure prediction from amino acid
sequences is currently a very challenging problem in computational
bioengineering. Sequences of amino acids which compose proteins are
visualized by representing their secondary structural characteristics
as a sequence of alpha helices and beta sheets. They are further
specified according to their tertiary structural characteristics. This
involves the determination of three dimensional relationships, which
are formulated based on determination of gloal energy minima, such
that the overall three dimensional structure is configured to
represent the most stable configuration produced by changes in
dihedral angles of covalent bonds and hydrogen bond interactions. The
number of possible configurations is enormous, and determination of an
optimal configuration a computationally demanding problem. Working
with researchers from the
Computational and Theoretical Biology Department in the Physical
Biosciences Division, we are using interactive visualization to
constrain and decrease the problem space to produce a more rapid
convergence to compute these global energy minima. (There is no web
page yet for this project.)
The following list of LBL Staff and UC Davis Faculty are contributing
to the technical management of these projects: