Skip to content →

neverendingbooks Posts

NOG master class update


Yesterday I made a preliminary program for the first two months
of the masterclass non-commutative geometry. It is likely that
the program will still undergo changes as at the moment I included only
the mini-courses given by Bernhard
Keller
and Markus Reineke but several other people have
already agreed to come and give a talk. For example, Jacques Alev (Reims),
Tom Lenagan (Edinburgh),
Shahn Majid (London),
Giovanna Carnovale (Padua) among others. And in
may, Fred assures me, Maxim Kontsevich will give a couple of talks.

As for the contents of the two courses I will be
teaching I changed my mind slightly. The course non-commutative
geometry
I teach jointly with Markus Reineke and making the program
I realized that I have to teach the full 22 hours before he will start
his mini-course in the week of March 15-19 to explain the few
things
he needs, like :

To derive all the
counting of points formulas, I only need from your course:

the definition of formally smooth algebras basic properties, like
being
hereditary
– the definition of the component
semigroup
– the fact that dim Hom-dim Ext is constant along
components. This I need
even over finite fields $F_q$, but I
went through your proof in “One quiver”,
and it works. The
key fact is that even over $F_q$, the infinitesimal lifting
property implies smoothness in the sense Dimension of variety =
dimension of
(schematic) tangent space in any $F_q$-valued
point. But I think it’s fine for
the students if you do all
this over C, and I’ll only sketch the (few)
modifications for
algebras over $F_q$.

So my plan is to do all of
this first and leave the (to me) interesting problem of trying to
classify formally smooth algebras birationally to the second
course projects in non-commutative geometry which fits the title
as a lot of things still need to be done. The previous idea to give in
that course applications of non-commutative orders to the resolution of
singularities (in particular of quotient singularities) as very roughly
explained in my three talks on non-commutative geometry@n I now
propose to relegate to the friday afternoon seminar. I’ll be
happy to give more explanations on all this (in particular more
background on central simple algebras and the theory of (maximal)
orders) if other people work through the main part of the paper in the
seminar. In fact, all (other) suggestions for seminar-talks are welcome
: just tell me in person or post a comment to this post.

Leave a Comment

Bill Schelter’s Maxima

Bill
Schelter was a remarkable man. First, he was a top-class mathematician.
If you allow yourself to be impressed, read his proof of the
Artin-Procesi theorem. Bill was also among the first to take
non-commutative geometry seriously. Together with Mike Artin he
investigated a notion of non-commutative integral extensions and he was
the first to focuss attention to formally smooth algebras (a
suggestion later taken up by a.o. Cuntz-Quillen and Kontsevich) and a
relative version with respect to algebras satisfying all identities of
n x n matrices which (via work of Procesi) led to smooth@n
algebras. To youngsters, he is probably best know as the co-inventor of
Artin-Schelter regular algebras. I still vividly remember an
overly enthusiastic talk by him on the subject in Oberwolfach, sometime
in the late eighties. Secondly, Bill was a genuine Lisp-guru and
a strong proponent of open source software, see for example his
petition against software patents. He maintanind
his own version of Kyoto Common Lisp which developed into Gnu
Common Lisp
. A quote on its history :

GCL is
the product of many hands over many years. The original effort was known
as the Kyoto Common Lisp system, written by Taiichi Yuasa and Masami
Hagiya in 1984. In 1987 new work was begun by William Schelter, and that
version of the system was called AKCL (Austin Kyoto Common Lisp). In
1994 AKCL was released as GCL (GNU Common Lisp) under the GNU public
library license. The primary purpose of GCL during that phase of it’s
existence was to support the Maxima computer algebra system, also
maintained by Dr. Schelter. It existed largely as a subproject of
Maxima.

Maxima started as Bill’s version of
Macsyma an MIT-based symbolic computation program to which he
added many routines, one of which was Affine a package that
allowed to do Groebner-like computations in non-commutative
algebras (implementing Bergman’s diamond lemma) and which he
needed to get a grip on 3-dimensional Artin-Schelter regular
algebras
. Michel and me convinced Fred to acquire funds to
buy us a work-station (costing at the time 20 to 30 iMacs) and have Bill
flown in from the States with his tape of maxima and let him
port it to our Dec-station. Antwerp was probably for years
the only place in the world (apart from MIT) where one could do
calculations in affine (probably highly illegal at the time).
Still, lots of people benefitted from this, among others Michaela
Vancliff
and Kristel Van Rompay in their investigation
of 4-dimensional Artin-Schelter regular algebras associated to an
automorphism of a quadric in three-dimensional projective space.
Yesterday I ran into Bill (alas virtually) by browsing the
crypto-category of Fink. There it was, maxima, Bill’s package! I tried to install it
with the Fink Commander and failed but succeeded from the command line.
So, if you want to have your own version of it type

sudo fink
install maxima

from the Terminal and it will install without
problems (giving you also a working copy of common lisp). Unfortunately
I do not remember too much of Macsyma or Affine but there is plenty of
documentation on the net. Manuals and user guides can be obtained from
the maxima homepage and the University of Texas
(Bill’s university) maintains an online manual, including a cryptic description of
some Affine-commands. But probably I’ll have to send Michaela an
email asking for some guidance on this… Here, as a tribute to Bill who
died in july 2001 the opening banner

 iMacLieven:~ lieven$
/sw/bin/maxima Maxima 5.9.0 http://maxima.sourceforge.net 
Distributed under the GNU Public License. 
See the file COPYING. 
Dedicated to the memory of William Schelter. 
This is a development version of Maxima. 
The function bug_report() provides bug reporting information. 
(C1)
 
Leave a Comment

bandwidth measures


One day (hopefully) lots of MP3, JPEG and perhaps even
MPEG-files will be flying around our wireless home-network. But I
didn’t have any idea of how much data I could cram through the
Airport-connections. To estimate the available bandwith of a
network there is a nice free tool around, iperf of which you can download binaries for
almost any platform including OS X. So click on the MacOS X (Darwin 6.4)
binary button half way on the iperf-page and you get a Desktop
iperf-1.7.0-powerpc-apple-darwin6.4 Folder
which you may rename to
just iperf. Do this on two computers connected to the
Airport-network you want to measure. Now, decide which of the two will
play the ‘server’ and which the ‘client’ (the end result does not
depend on this choice). So fire up the Terminal of the serving
computer and type

sudo ~/Desktop/iperf/iperf -s

and you will
get a message saying that the server is listening on TCP port 5001. Go
to the SystemPreferences/Network to obtain the IP-address of the server
(say it is 10.0.1.5) . Walk over to the ‘client’-computer and type
into its Terminal

sudo ~/Desktop/iperf/iperf -c 10.0.1.5
-r

and after a few moments it will compute the bandwidth of the
connection for you. Here is a sample output of two Airport-card
iMacs connected to the same Airport-Extreme base station :

iMacLieven:~/Desktop/iperf lieven$ ./iperf
-s ------------------------------------------------------------\r\
nServer listening on TCP port 5001 TCP window size: 64.0 KByte
(default) -----------------------------------------------------------
- [  4] local 10.0.1.2 port 5001 connected with 10.0.1.7 port
49245 [ ID] Interval       Transfer     Bandwidth [  4]  0.0-10.3
sec  2.77 MBytes  2.27
Mbits/sec -----------------------------------------------------------
- Client connecting to 10.0.1.7, TCP port 5001 TCP window size:
65.0 KByte
(default) -----------------------------------------------------------
- [  4] local 10.0.1.2 port 49515 connected with 10.0.1.7 port
5001 [ ID] Interval       Transfer     Bandwidth [  4]  0.0-10.2
sec  2.73 MBytes  2.23 Mbits/sec indicating a bandwidth of approximately
2.25Mbits/sec. If we replay the same game with two
AirportExtreme-card iMacs on the same network we can nearly
triple (!) the bandwidth : 
[eMacAnn:~] lieven% cd
Desktop/iperf [eMacAnn:~/Desktop/iperf] lieven% ./iperf
-s ------------------------------------------------------------\r\
nServer listening on TCP port 5001 TCP window size: 64.0 KByte
(default) -----------------------------------------------------------
- [  4] local 10.0.1.5 port 5001 connected with 10.0.1.6 port
49314 [ ID] Interval       Transfer     Bandwidth [  4]  0.0-10.0
sec  8.50 MBytes  7.11
Mbits/sec -----------------------------------------------------------
- Client connecting to 10.0.1.6, TCP port 5001 TCP window size:
65.0 KByte
(default) -----------------------------------------------------------
- [  4] local 10.0.1.5 port 49320 connected with 10.0.1.6 port
5001 [ ID] Interval       Transfer     Bandwidth [  4]  0.0-10.9
sec  7.07 MBytes  5.45 Mbits/sec

However, if these two
AirportExtrame-card computers connect to each other via the
Graphite-Airport base station the bandwidth drops to a meagre 1.9
Mbits/sec which is roughly the same as two Airport-card computers
connecting (which gave me 2.45 Mbits/s). Anyway, there is no immediate
problem with bandwidth on either network for what I have in mind.
Another important number to know is the real speed of our
internet-connection (for instance if I want to replace our old router by
a better documented one and have a measure for the in/decrease of the
connection-speed). Here, a good URL is performance.chello.at which offers two tests :
String and String SSI. The later one has a graphical
resulting page such as

Leave a Comment