Yozo Hida

yozo (at) cs (dot) berkeley (dot) edu

Mathematics, rightly viewed, possesses not only truth, but supreme
beauty -- a beauty cold and austere, like that of sculpture without
appeal to any part of our weaker nature, without the gorgeous
trappings of paintings or music, yet sublimely pure and capable of
stern perfection such as only the greatest art can show.
-- Bertrand Russell, 1872-1970.

[ Main | Classes | Links | Random ]

I am a graduate student at U.C. Berkeley in Computer Science interested in scientific computing and symbolic mathematical computing. I am advised by Prof. James Demmel.

Research

Double-Double and Quad-Double Arithmetic. I worked on efficient algorithms (and implementation) on arithmetic on double-double and quad-double numbers. Double-double and quad-double numbers are unevaluated sum of two and four IEEE doubles capable of representing 106 and 212 bits of significand, respectively. The library is written in C++, taking full advantage of operator overloading. C, Fortran 77, and Fortran 90 interfaces are also provided. This work was done at Lawrence Berkeley National Laboratory, NERSC Division, with Xiaoye S. Li and David H. Bailey. Details can be found in the technical report here. Current implementation of qd package (which includes both double-double and quad-double arithmetic) can be downloaded: qd-2.3.6.tar.gz (2008-03-16). A Windows version (MS Visual C++ Express 2005) is also available: qd-2.3.4-windll.zip. (This is essentially identical to the UNIX version, except it has config.h pre-generated and includes MSVC++ project files. For Cygwin or MinGW setup I recommend just using the UNIX version as they usually are more up to date [I don't use Windows much].)

Arbitrary Precision Arithmetic. I have also worked (in some part) on the arbitrary precision package ARPREC. Current stable implementation of arprec package can be downloaded here (from David Bailey's webpage). This is a successor to David Bailey's MPFUN, written in C++, taking full advantage of C++ operator overloading. C, Fortran 77, and Fortran 90 interfaces are also provided. I also maintain my own development version arprec-2.2.1.tar.gz (2008-01-22), which should eventually get merged into the main line found at David Bailey's page. A Windows version (MS Visual C++ Express 2005) is also available arprec-2.2.1-windll.zip. (This is essentially identical to the UNIX version, except it has config.h pre-generated and includes MSVC++ project files. For Cygwin or MinGW setup I recommend just using the UNIX version as they usually are more up to date [I don't use Windows much].)

Extended and Mixed Precision BLAS. I have also been doing research with Jim Demmel and Xiaoye S. Li, working on the XBLAS (Extended Linear Algebra Subprograms) Project. More information about this project is found here. It's basically improving the numerical accuracy of various basic linear algebra routines (such as iterative refinement) by using higher internal accuracy than the input/output. It also supports mixed precision arithmetic. The XBLAS is a part of BLAS, documented at the BLAS Technical Forum. The most recent devel version is also available xblas-1.0.243.tar.gz. A Matlab (R) mex files for extended precision matrix multiply routine (handling double, double-extended, double-double, and quad (on Sun)) is available here.

Hyperbolic Tree. During the summer of 1999 I was a research intern at Xerox PARC, and I worked on an extension of the Hyperbolic Tree Browser. Originally developed by John Lamping and Ramana Rao at Xerox PARC, this uses non-euclidean geometry for new user interfaces. The Hyperbolic Tree Browser (now called Star Tree) is now a product at Inxight.

Papers

  1. J. Demmel, Y. Hida, W. Kahan, X. S. Li, S. Mukherjee, E. J. Riedy. "Error Bounds from Extra Precise Iterative Refinement", Technical Report UCB/CSD-04-1344, February 2005. [ ps.gz | pdf ]
  2. J. Demmel and Y. Hida. Fast and Accurate Floating Point Summation with Application to Computational Geometry. Numerical Algorithms 37 (1-4): 101-112, December 2004. Originally presented at 10th GAMM-IMACS International Symposium on Scientific Computing, Computer Arithmetic, and Validated Numerics (SCAN 2002). [ ps.gz ]
  3. J. Demmel and Y. Hida, "Accurate and Efficient Floating Point Summation", SIAM J. Sci. Comput., 25(4):1214--1248, 2003.
  4. James Demmel and Yozo Hida, "Accurate Floating Point Summation", Technical Report UCB/CSD-02-1180, May 2002. This is essentially the same paper as above. [ ps.gz | pdf ]
  5. D. H. Bailey, D. Broadhurst, Y. Hida, X. S. Li, and B. Thompson, "High Performance Computing Meets Experimental Mathematics", Proceedings of the IEEE/ACM SC2002 Conference, 2002.
  6. X. Li, J. Demmel, D. Bailey, G. Henry, Y. Hida, J. Iskandar, W. Kahan, S. Kang, A. Kapur, M. Martin, B. Thompson, T. Tung, D. Yoo, "Design, Implementation and Testing of Extended and Mixed Precision BLAS", ACM Transactions on Mathematical Software, Vol. 28, No.2, June 2002, pp. 152-205. This is a shortened version of Technical Report LBNL-45991 [6]. [ pdf ]
  7. Yozo Hida, Xiaoye S. Li and David H. Bailey, "Algorithms for Quad-Double Precision Floating Point Arithmetic", ARITH-15, Oct. 2000. This is a shortened version of Technical Report LBNL-46996 [5]. [ ps.gz ]
  8. Yozo Hida, Xiaoye S. Li and David H. Bailey, "Quad-Double Arithmetic: Algorithms, Implementation, and Application", Technical Report LBNL-46996, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, Oct. 2000 [ ps.gz ]
  9. X. Li, J. W. Demmel, D. H. Bailey, G. Henry, Y. Hida, J. Iskandar, W. Kahan, A. Kapur, M. C. Martin, T. Tung, D. J. Yoo, "Design, Implementation and Testing of Extended and Mixed Precision BLAS", Technical report LBNL-45991, Oct. 2000. [ ps.gz ]

Teaching

Yozo Hida < yozo (at) cs (dot) berkeley (dot) edu > Last Updated 19 September 2005.