¿Hey, whatcha doin on this page?

It's just where I put stuff that I'm experimenting on for possible f¯uture use.

류 광

- (1)

Stanford's Center for the Study of Linguistics and Information
is currently preparing a series of books that contain
archival forms of my published papers,
and they want to pay someone to put most of the illustrations into
MetaPost
form. If you are interested in exploring this possibility, please
send a message to
`dikran@csli.stanford.edu`.

The next talk in the series is entitled

and the date is

and the room number is B01 and the time is 4:15. (This lecture will feature online demonstrations of some new software that allows infinite experimentations with the parameters of high-performance computer design. Using the new MMIX architecture, one can specify any degree of superscalar, superpipelined machines with any number of functional units each implementing arbitrary subsets of the operations. Each operation can take a specified amount of time in each of a specified number of pipeline stages. Many varieties of caches, many varieties of branch prediction, etc., are supported in a machine that does speculative execution with dynamic scheduling.) PS file for testing

류귑 컴퓨터 프로긔래밍의 예술 기초 알고리즘 준수치적 정럴 과 검색

I'll be playing organ (continuo) with strings, trumpets, and timpani in a performance of Mozart's Vesperæ Solennes de Confessore on Sunday, April 27, 4:00 pm, at First Lutheran Church in Palo Alto (corner of Homer and Webster streets).

CharlesFollowing up on the successful ``West Coast MMIX Day'' held at Stanford in August, we're getting ready for the ``East Coast MMIX Day'' at Harvard, on Tuesday October 26. This event, hosted by Norman Ramsey, begins at 10am, in a room to be announced (somewhere in the new Maxwell Dwork CS building). I will demo the MMIX assembler and its two simulators and answer everybody's questions. We plan to run for two hours, then break for lunch, and continue for another two or three hours or until everybody is too tired and/or saturated to continue...

Everybody is invited; please send email to
`pennell@eecs.harvard.edu`
if you have questions or if you want to be sure there is room for you!

It would be best if people are familiar with at least the tutorial
information about `MMIX` that is described below; even better if
attendees have tried to download and install the software themselves;
even better if they have tried to write and run at least one short program.
You are welcome to come even if you know nothing about the machine; but
you (and I) will get more out of the day the more you know beforehand.

and the date is

and it will take place in

(Please note the nonstandard time and place.)

The next talk in the series is entitledand the date is

and the room number is B01 and the time is 4:15.

The spanning trees of a graph or network of *n* points are the
ways to connect those points by choosing *n*-1 of the available
point-to-point interconnections. I will discuss an interesting procedure
by which we can run through all possible spanning trees in such a way
that only one branch of the tree is changed at every step. The method
is particularly nice when the given graph is a series--parallel network.

This talk will be the 10th annual "Christmas Tree Lecture", and it is also part of the Stanford Computer Forum's Emeritus Lecture Series.

The next talk in the series is entitled

and the date is

and the place is

and the time is 4:30.

I will discuss the amazing fact that the so-called Horton-Strahler number of a river has the same statistical properties as the so-called complexity of single-stranded nucleic acid structures, as well as being equivalent to the number of registers needed to evaluate algebraic expressions in a computer. If time permits, I'll also discuss other surprising appearances of this statistic, for example in new combinatorial objects called Kepler towers.