Professor Emeritus of Mathematics, and of E.E. & Computer Science
863 Evans Hall (Math), and 411 Soda Hall (CS)
Now that I am "retired", I work in my
offices sporadically at least once or
twice a week during each semester.
Phone: (510) 642-5638 (rings in both offices)
E-mail to me at this address obfuscated to diminish spam:
wkahan at (omit underscores) E_E_C_S_D0T_B_E_R_K_E_L_E_Y_D0T_E_D_U
My e-mail is read at most once every week or two. There is much too much of it.
Better ways to communicate with me are ...
... by paper mail: EECS Dept., MS#1776, Univ. of Calif., Berkeley CA 94720-1776
... by telephone (or leave a message for me to call back), or
... by visiting when I am in my office (telephone me first).
Use e-mail only if you must; but be advised that the University's
e-mail is not so secure that you can trust its confidentiality.
Awards etc.
Selected Publications
7094II System Support for Numerical Analysis
IBM SHARE Secretarial Distribution SSD#159 Item C-4537 (1966), and in Error in Numerical Computation, Univ. of Mich. Eng'g Summer Conf'ce #6818 Numerical Analysis (1968) (A retyped reproduction)(PDF file)
Calculating the Singular Values and Pseudo-Inverse of a Matrix
(with G.H. Golub), pp. 205-224, J. SIAM Numer. Anal.SER. B. Vol. 2 #2 (1965)(PDF file)
Analysis and Refutation of the LCAS
ACM SIGNUM Newsletter, Vol. 26,
No. 3, July 1991, pp. 2-15.
Also, ACM SIGPLAN Notices, Vol. 27,
No. 1, January 1992, pp.61-74.
Accurate Singular Values of Bidiagonal Matrices
(with J. Demmel), SIAM J. Scientific Statistical Computation,
Vol. 11, No. 5, 1990, pp. 873-912.
Branch Cuts for Complex Elementary Functions, or Much Ado About
Nothing's Sign Bit
in The State of the Art in Numerical Analysis,
(eds. Iserles and Powell), Clarendon Press, Oxford, 1987.
A Wordsize- and Radix-Independent Standard for Floating-Point
Arithmetic
(with W. J. Cody et al.), IEEE Micro, August 1984,
pp. 86-100.
Mathematics Written in Sand
Proc. Joint Statistical Mtg. of the
American Statistical Association, 1983, pp. 12-26.
(A legible reproduction is available below) (PDF file)
Norm-Preserving Dilations and Their Applications to Optimal Error
Bounds
(with C. Davis and H. F. Weinberger), SIAM J. Numerical
Analysis, Vol. 19, 1982, pp. 445-469.
Residual Bounds on Approximate Eigensystems of Nonnormal
Matrices
(with B. N. Parlett and E. Liang), SIAM J. Numerical Analysis,
Vol. 19, 1982, pp. 470-484.
A Family of Anadromic Numerical Methods for Matrix Riccati Differential Equations
(with Ren-Cang Li), Mathematics of Computation,
Vol. 81 #227, Jan. 2012, pp. 233-265. (PDF file)
Unconventional Schemes for a Class of Ordinary Differential Equations
--
-- With Applications to the Korteweg-deVries Equation
(with Ren-Cang Li), J. Computational Physics,
Vol. 134, 1997, pp. 316-331.
Composition Constants for Raising the Orders of Unconventional Schemes
for Ordinary Differential Equations
(with Ren-Cang Li), Math. of Computation,
Vol. 66, 1997, pp. 1089-1099.
Pinchings and Norms of Scaled Triangular Matrices
(with Ren-Cang Li & Rajendra Bhatia), Linear & Multilinear Algebra,
Vol. 50, 2002, pp. 15-21.
Is there a small skew Cayley transform with zero diagonal ?
Linear Algebra and its Applications417 (2006) pp. 335-341.
(A version with fewer misprints is available here:) (PDF file)
Error bounds from extra-precise iterative refinement
(with J.W. Demmel et al.)
ACM Transactions on Mathematical Software (TOMS) archive 32, Issue 2 (June 2006) pp. 325 - 351 (PDF file)
Files available from this homepage, last updated 11 Sept. 2019
Presented to "Matrix Computations and Scientific Computing" Seminar 9 Sept. 2009 (PDF file)
Needed Remedies for the Undebuggability of ... Floating-Point Computations ...
(PDF file)
Desperately Needed Remedies for the Undebuggability of ... Floating-Point Computations ...
(PDF file)
Desperately Needed Remedies for the Undebuggability of ... Floating-Point Computations ... as presented at Heidelberg Laureates Forum #2
(PDF file)
on Pete's Unsung Contribution to IEEE Standard 754 for Binary Floating-Point
(PDF file) for a Panel on the Future of Numerical Linear Algebra: Undebuggable (PDF file)
Deflations Preserving Relative Accuracy, for IWASEP IX in Napa 4 June 2012
(PDF file)
For the ACM's Turing Centenary Celebration 15-16 June 2012 in San Francisco
(PDF file)
A Numerical Analyst thinks about Artificial Intelligence & Deep Learning
(PDF file)
Automatic Step-Size Control for Minimization Iterations
(PDF file)
For ARITH 23: Critiques of John L. Gustafson's Unum Computation and SORN arithmetic
(PDF file)