## Peter Neumann to Speak 4/11/2008 at 3 PM

HCS regularly brings in interesting speakers on topics in computer science, technology, and their relationship with the rest of the world. This month, we're proud to be hosting Peter G. Neumann, one of the original developers of Multics, moderator of the ACM Risks Forum, and a graduate of Harvard. Dr. Neumann will speak about his personal experience with mathematical computer science and will present short stories on lots of exciting topics (read on to find out what). We'll also have plenty of time for questions and discussion. We'll likely go to dinner afterwards.

Date: April 11, 2008

Time: 3:00 PM

Place: Maxwell-Dworkin 119 (the Grace Murray Hopper Memorial Conference Room)

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

A Short Personal View of

Mathematical Beauty in Computer Science

Peter G Neumann (Harvard '54, SM '55, PhD '61)

SRI International Computer Science Lab

Menlo Park CA 94025-3493

Neumann@CSL.sri.com, 1-650-859-2375

http://www.csl.sri.com/neumann

Harvard Computer Society, April 11, 2008

Maxwell-Dworkin 119, 33 Oxford St., 3-5pm

This is intended to be a free-wheeling discusion of a few

elegant roles of mathematics in computer science. It might

include any of the following suggested topics, according to the

interests and experiences of the group, and the time available.

Join us for an early dinner if you wish.

* Error-correcting codes: Huffman's simple graphic depiction of the

basic Hamming code. Huffman's little-known nonoptimal but very

clever graph-based codes, where the minimum distance of the code is

the smallest loop in the graph. Group-theoretic codes became the

norm in the 1960s (Galois fields, etc.). There are also some links

with cryptography -- which is its own mecca of mathematical

mechanizations (Chris Peikert's MIT thesis).

* A clever application of a surprisingly counterintuitive solution to

a variant of the hat problem with 2**m-1 people where only one

guess is permitted overall from the group. Intuition says maybe

the probability of success is one-half. Hamming codes are the

basis for an algorithm for achieving a probability of success just

less than one! See Mira Bernstein, The Hat Problem and Hamming

Codes, Focus, The Newsletter of the Mathematical Association of

America, 21, 8, November 2001, pp. 4-6.

* An interesting link between a quasi-Fibonacci linear recursion and

primes, a union of additive and multiplicative properties: f(n) =

f(n-3) + f(n-4) for n>3, f(0)=4, f(1)=f(2)=0, f(3)=3 has the property

that f(p) is a multiple of p whenever p is a prime! This was MAA Math

Monthly problem 10655, with two proofs given in the March 2000 issue.

Stephen Wolfram's A New Kind of Science has a somewhat simpler similar

example on page 891. (He suggested to me that the recursive sequences

on pages 131 and 907 also merit consideration.)

* Statistical linguistics: An interesting but almost unknown 1957

derivation of Zipf-Pareto-Mandelbrot formulas by Vitold Belevitch

(On the Statistical Laws of Linguistic Distributions, Annals of the

Society of Science, Brussels, Belgium, 73, III, 1959,

pp. 310--326), and the relationship with power laws. Tony

Oettinger got me started on that in his mid-1950s courses on

statistical linguistics.

* A mid-1950s effort to model common-meter hymn tunes using Markov

chain probabilities, and synthesize new tunes of varying constraint

lengths, by Fred Brooks, Al Hopkins, PGN, and Bill Wright on the

Harvard Mark IV: An Experiment in Musical Composition, IRE

Transactions on Electronic Computers, EC-6, September 1957,

pp. 175-182. Also inspired by Tony Oettinger's StatLing course.

* Some thoughts on composable architectures for developing

trustworthy computer systems, including formal methods for

analyzing complex systems. See

P.G. Neumann, Principled assuredly trustworthy composable

architectures, Technical report, Computer Science Laboratory,

SRI International, Menlo Park, California, December 2004.

http://www.csl.sri.com/neumann/chats4.html, .pdf, and .ps.

and

P.G. Neumann and R.S. Boyer and R.J. Feiertag and K.N. Levitt

and L. Robinson, A Provably Secure Operating System: The

System, Its Applications, and Proofs, Computer Science

Laboratory, SRI International, Menlo Park, California, Report

CSL-116 May 1980. Tagged strongly typed capabilities.

Perhaps dated, but ground-breaking.

http://www.csl.sri.com/neumann/psos/psos80.ps

* My Multics experience with F.J. Corbato, Ted Glaser, Jerry Saltzer, Bob

Daley, at MIT, Vic Vyssotsky, Joe Ossanna, Ken Thompson at Bell Labs,

and many others, clearly made on enormous impact on appreciating the

importance of elegant system architectures and structured development

processes.

* Huffman's continuous deformations of semirigid surfaces and

amazing artistic creations. D.A. Huffman, Curvatures and

Creases: A Primer on Paper, IEEE Transactions on Computers,

C-25, 10, pp. 1010--1019, October 1975. (I have some

fascinating slides of curved surfaces that almost defy belief

-- unless you have read the paper.)

* One other relevant reference:

W.H.J. Feijen, A.J.M. van Gasteren, D. Gries, and J. Misra, editors,

Beauty is our Business, A Birthday Salute to Edsger W. Dijkstra,

Springer-Verlag, Berlin, 11 May 1990. (I contributed

Chapter 39, Beauty and the Beast of Software Complexity:

Elegance versus Elephants.)

* As EXAMPLES of what the lack of beauty can mean, a large collection

of misuses of mathematics and the lack of sensible system

development are found in the ACM Risks Forum archives

(http://www.risks.org). These include major problems in our

infrastructures and user applications, relating to safety,

security, reliability, and particularly the lack of integrity and

accountability in very badly designed election systems.

BIO: After 8 years at Harvard, 2 years on a Fulbright, two

doctorates, ten years at Bell Labs in Murray Hill (where I was

heavily involved in the Multics development jointly with MIT and

Honeywell), and a year teaching at Berkeley, I've been in SRI's

Computer Science Lab since September 1971. I tend to work on

computer systems and networks, trustworthiness, high assurance,

security, reliability, survivability, safety, and many risks-related

issues such as voting-system integrity, crypto policy, social

implications, and human needs including privacy. I moderate the ACM

Risks Forum, edit CACM's Inside Risks columns, chair the ACM

Committee on Computers and Public Policy, and chair the National

Committee for Voting Integrity (http://www.votingintegrity.org). I

created ACM SIGSOFT's Software Engineering Notes in 1976, was its

editor for 19 years, and still contribute the RISKS section. I'm on

the editorial board of IEEE Security and Privacy. I've participated

in four studies for the National Academies of Science: Multilevel

Data Management Security (1982), Computers at Risk (1991),

Cryptography's Role in Securing the Information Society (1996), and

Improving Cybersecurity for the 21st Century: Rationalizing the

Agenda (2007). My 1995 book, Computer-Related Risks, is still

timely. I'm is a Fellow of the ACM, IEEE, and AAAS, and also an SRI

Fellow. I received the National Computer System Security Award in

2002 and the ACM SIGSAC Outstanding Contributions Award in 2005. I'm

a member of the U.S. Government Accountability Office Executive

Council on Information Management and Technology, and the California

Office of Privacy Protection advisory council. I co-founded People

For Internet Responsibility (PFIR, http://www.PFIR.org). I have

taught courses at Darmstadt, Stanford, U.C. Berkeley, and the

University of Maryland. My website (http://www.csl.sri.com/neumann)

has testimonies for the U.S. Senate and House and California state

Senate and Legislature, papers, bibliography, further background, etc.