### Today's Editor:

- Cleve Moler
- The MathWorks, Inc.
- moler@mathworks.com

- Encrypted SIAM Membership List Available
- Looking for Tricks of the Trade
- Huge Sparse Eigenvalue Problem
- Unix "tar" on DOS.
- Contents: Linear Algebra and its Applications
- Contents: SIAM Optimization

-------------------------------------------------------

From: Eric Grosse 908-582-5828 <ehg@research.att.com>

Date: Sun Aug 9 22:52:50 EDT 1992

**Subject: Encrypted SIAM Membership List Available**

The SIAM membership list, a useful source of up-to-date addresses and

phone numbers, has long been searchable via netlib. Now you can also

download it to your own machine for faster searching.

To preserve privacy, i.e. to keep the list from being used by mass mailers

and telemarketers, the database is encrypted. Given a person's last name

or phone number, you can decrypt that one database entry. But there is

no feasible way to crack the entire list. To learn how we do this, read

J. Feigenbaum, E. Grosse and J. Reeds (1992) "Cryptographic Protection of

Membership Lists", Newsletter of the International Association for

Cryptologic Research, 9:1,16-20. (This paper is available from netlib by

"send 91-12 from research/nam".)

You can use the system without understanding the mechanism. First, get

the decryption program and (1.2 megabyte) database by

ftp research.att.com

login: netlib

password: <your email address>

binary

cd research

get decryptdb.c

get siamdb

quit

then follow the instructions at the start of decryptdb.c to install.

For now, you must have ftp access and a C compiler; if demand

warrants, SIAM headquarters may make the system available on other

media at a later time.

The database, which is updated quarterly, will continue to be

searchable via netlib's "whois" command. But fast local access allows

new uses; for example, my computer is connected to my phone and, when

caller-ID is functioning, automatically translates the calling number

into a name.

------------------------------

From: Steve Stevenson <steve@hubcap.clemson.edu>

Date: Mon, 10 Aug 92 08:44:59 -0400

**Subject: Looking for Tricks of the Trade**

In the past several months, I have read a couple of texts which have made

a big deal about Horner's rule, like it was a new can for beer.

Other texts seem to be totally oblivious to certain computational facts of

life, like using extrapolation. Yet another trick is converting

Taylor series from interative (natural) to recursive. Most of this

stuff goes back to the pre-computer days when things had to be

done by hand.

I would like to compile a list of all the old and new computational techniques

which people use to accelerate computations. If you would please send me

just the name and a reference, I'll summerize.

Thanks.

Steve

Steve (really "D. E.") Stevenson steve@hubcap.clemson.edu

Department of Computer Science, (803)656-5880.mabell

Clemson University, Clemson, SC 29634-1906

------------------------------

From: Ching-ju Ashraf Lee <leec2@rpi.edu>

Date: Thu, 13 Aug 92 12:26:24 -0400

**Subject: Huge Sparse Eigenvalue Problem**

Dear Sir/Madame:

I have a huge sparse eigensystem Ax=lambda*Bx to solve. A and

B are real, symmetric and even positive definite. I only need

the smallest eigenvalue of this system and such eigenvalue in my

system is always simple and believed to be well-isolated. An

immediate numerical method for such problem is the inverse iteration

method: (A-mu*B)x(k+1) = Bx(k). Unfortunately, the best method I

come up with in solving the above system is the Cholesky decompo-

sition(I am able to get a lower bound of lambda, hence mu may be

taken to be the lower bound). But since A and B are 70,000 by

70,000, the half bandwidth of A and B is usually around 30,000.

Even though I only have at most 24 nonzero entries per row in the

matrices, Cholesky decomposition will fill nonzero entries inside

the band. So the virtual memory required by the method is way

beyond the limit on my local machine(700 megabytes). Note also

that if mu is close to lambda, therefore a good initial approxi-

mation of the system, A-mu*B is quite singular. So the ordinary

iterative methods will not work well in solving the system

(A-mu*B)x(k+1)=Bx(k) for x(k+1) (correct me if my impression is

false). I would like to utilize the special sparseness I have in

this system, if possible, before jump on a larger machine. So any

thoughtful suggestions or reference of the related literature will

be greatly appreciated.

Ching-ju Ashraf

leec2@rpi.edu

------------------------------

From: Luciano Molinari <molinari@cumuli.ethz.ch>

Date: Fri, 14 Aug 1992 11:33:23 +0200

**Subject: Unix "tar" on DOS.**

I am looking for an MS-DOS utility to dearchive and decompress Unix tar.Z

files. Can anybody help me?

Thanks,

L. Molinari

molinari@cumuli.ethz.ch

The Children's Hospital

Steinwiesstr.75

CH-8032 Zurich

------------------------------

From: Richard Brualdi <brualdi@math.wisc.edu>

Date: Wed, 12 Aug 92 07:28:17 CDT

**Subject: Contents: Linear Algebra and its Applications**

Contents of LAA Volume 174, September 1992

Joel V. Brawley (Clemson, South Carolina) and Gary L. Mullen

(University Park, Pennsylvania)

Scalar Polynomial Functions on the Nonsingular Matrices Over a Finite Field 1

Robert Brawer and Magnus Pirovino (Zurich, Switzerland)

The Linear Algebra of the Pascal Matrix 13

A. A. Chernyak and Z. A. Chernyak (Minsk, U.S.S.R.)

Joint Realization of (0, 1) Matrices Revisited 25

Lifeng Ding (Atlanta, Georgia)

Separating Vectors and Reflexivity 37

Massoud Malek (Hayward, California)

Notes on Permanental and Subpermanental Inequalities 53

Keith Bourque and Steve Ligh (Lafayette, Louisiana)

On GCD and LCM Matrices 65

K. H. Kim and F. W. Roush (Montgomery, Alabama)

Automorphisms of gl-Matrices 75

Charles Lanski (Los Angeles, California)

An Identity for Matrix Rings With Involution 91

P. J. Maher (London, England)

Some Norm Inequalities Concerning Generalized Inverses 99

Desmond J. Higham (Dundee, Scotland) and Nicholas J. Higham (Manchester,

England)

Componentwise Perturbation Theory for Linear Systems With Multiple Right-Hand

Sides 111

John A. Holbrook (Guelph, Canada)

Spectral Variation of Normal Matrices 131

Lei Wu (Dalian, People's Republic of China)

The Re-Positive Definite Solutions to the Matrix Inverse Problem AX=B 145

O. L. Mangasarian (Madison, Wisconsin)

Global Error Bounds for Monotone Affine Variational Inequality Problems 153

Barbu C. Kestenband (Old Westbury, New York)

Quadrics as Hyperplanes in Finite Affine Geometries 165

Max Bauer (Rennes, France)

Dilatations and Continued Fractions 183

Bernhard A. Schmitt (Marburg, Germany)

Perturbation Bounds for Matrix Square Roots and Pythagorean Sums 215

Vlad Ionescu and Martin Weiss (Bucharest, Romania)

On Computing the Stabilizing Solution of the Discrete-Time Riccati Equation

229

Daniel B. Szyld (Philadelphia, Pennsylvania)

A Sequence of Lower Bounds for the Spectral Radius of Nonnegative Matrices

239

BOOK REVIEW

Frank Uhlig (Auburn, Alabama)

Review of Topics in Matrix Analysis by Roger A. Horn and Charles R. Johnson

243

Author Index 247

------------------------------

From: Beth Gallagher <gallaghe@siam.org>

Date: Wed, 12 Aug 92 11:03:54 EST

**Subject: Contents: SIAM Optimization**

SIAM Journal on Optimization

November 1992 Volume 2, Number 4

CONTENTS

On the Behavior of Broyden's Class of Quasi-Newton Methods

Richard H. Byrd, Dong C. Liu, and Jorge Nocedal

New Results on a Continuously Differentiable Exact Penalty Function

Stefano Lucidi

On the Implementation of a Primal-Dual Interior Point Method

Sanjay Mehrotra

On Regularized Least Norm Problems

Achiya Dax

On the Continuity of the Solution Map in Linear Complementarity

Problems

M. Seetharma Gowda

Linear Inequality Scaling Problems

Uriel G. Rothblum

New Proximal Point Algorithms for Convex Minimization

Osman Guler

A Necessary and Sufficient Condition for a Constrained Minimum

J. Warga

Diagonal Matrix Scaling and Linear Programming

Leonid Khachiyan and Bahman Kalantari

Author Index

------------------------------

End of NA Digest

**************************

-------