武克胜
Telephone: 1(510)486-6609
E-mail: kwu
lbl.gov or
john.wuacm.org
Mailstop 50B-3238, 1 Cyclotron Road, Berkeley, CA 94720, USA
E-mail: kwu
lbl.gov or
john.wuacm.orgMailstop 50B-3238, 1 Cyclotron Road, Berkeley, CA 94720, USA
武克胜 |
Telephone: 1(510)486-6609 E-mail: kwu lbl.gov or
john.wuacm.orgMailstop 50B-3238, 1 Cyclotron Road, Berkeley, CA 94720, USA |
Analysis and management of large datasets
Parallel distributed software systems design and implementation
Component architecture for scientific software
Performance tuning for large distributed software
FastBit
[Publications]: an efficient
compressed bitmap index
technology for data intensive sciences. This project addresses the
challenges of efficiently searching growing amounts of data
collected/generated by various scientific applications, such as
high-energy physics,
combustion,
astrophysics, and
network
traffic analysis. The FastBit software has received
an R&D 100 Award;
here is
a photo
from the award receiption.
Connected Component Labeling
[Publications]: an efficient
connected component labeling algorithm. This grows out our work on
feature tracking for a combustion data analysis. The key new insight is
that there is a way to make use of an implicit union-find data structure
to speed up the connected component labeling algorithms, which in turn
leads to faster algorithms for finding regions of interest. In
particular, using compressed bitmaps as representations of points in the
regions of interest, we can find the regions in time that is
proportional to the the number of points on the boundary of the regions.
This is faster than
the best iso-contouring algorithms and
much faster
than similar region finding algorithms. This is also a basis of some of
the work on visualization
and visual analytics.
DEX
[Publications]: a query-based
visualization tool. This project provides a new visualization
capability based on the FastBit
technology and the fast connected
component labeling technology. This effective combination was first
demonstrated on a project of analyzing
combustion simulation data. It is extensively documented in our
paper at IEEE Vis 2005, and also appeared in a SciDAC review
report about the Scientific Data
Management Center.
TRLan [Publications]:
Thick-restart Lanczos method for symmetric eigenvalue problems.
An implementation in
Fortran 90 is available with a
BSD-like license.
Kesheng Wu, Ekow Otoo and Kenji Suzuki. Optimizing two-pass
connected-component labeling algorithms. Pattern Analysis &
Applications v11 DOI 10.1007/s10044-008-0109-y. 2008.
Kesheng Wu, Ekow Otoo, and Arie Shoshani,
Optimizing bitmap indices with efficient compression.
ACM Transactions on Database Systems, v 31, pages 1-38, 2006.
Kesheng Wu, Ekow Otoo, and Arie Shoshani,
On the Performance of Bitmap Indices for High Cardinality
Attributes.
VLDB 2004, pages 24 - 35.
Kesheng Wu and Horst Simon, Thick-restart Lanczos method for
large symmetric eigenvalue problems.
SIAM Journal on Matrix Analysis and Applications.
Vol. 22, No. 2, pp. 602-616, 2001.
Publications listed elsewhere on the web:
Google Scholar,
DBLP,
CiteSeer,
and Microsoft Libra.
