Andrew Cooke: C[omp]ute

AGU 2011

From: andrew cooke <andrew@...>

Date: Mon, 5 Dec 2011 18:59:53 -0300

I'm at the AGU (Geophysics conference in San Francisco).

I've been here before.  It sucked.  This time, though, I wnated it to be
different.  I really did.  But it *still* sucks:

 - Wifi sucks.  Not only do connections drop, but each new connection hijacks
   DNS to display a pointless "you are now connected" message on the first web
   connection.  Why?  It serves no fucking point at all - the only effect it
   has is to screw up SSL connections (so if you're connecting to read gmail,
   you need to first open some other page just to see how much AGU sucks).

 - The talks suck.  If you're giving a talk on your amazing toolkits I am
   going to check them out on the web (once I get past the suck Wifi).  If the
   last commit was March and there is no online documentation at all - just a
   google code source dump - I am not going to use it.  No matter how cool you
   think it is.

 - The posters suck.  Students: people really don't care about your equation.
   Yes, you spent a long time working out what it meant, but in the end it's
   just some code.  When I look at your poster I want to learn somethig new.
   And I don't give a fuck about the exact details of one particular equation
   you are solving.  Did you learn *no* general lessons?

And I'm only half way through the first day.

Andrew

In AGU's Defense (OpenSuse Kernel Bug)

From: andrew cooke <andrew@...>

Date: Tue, 6 Dec 2011 13:44:10 -0300

The Wifi issues turned out to be a kernel/wifi bug.  If you're using 3.0 (as
OpenSuse 12.1 does) and you're getting weird behaviour and messy logs
featuring "rcu_preempt_state detected stalls" then switch to kernel 3.2 via
"kernel of the day".  Instructions at
http://en.opensuse.org/openSUSE:Kernel_of_the_day

That's an RC kernel, and I have lost my keyboard after a hibernate (once, with
the default - I switched to deskptop and so far, touch wood, it's not happened
again), but it's better than 3.0.

Also, the bug seems to be specific to certain wifi connections.  So you may
want to switch to one that seems stable before updating your kernel.

Alternatively, if you're as dumb as me, then "rpm --rebuilddb" will fix your
rpm database if you end up having to do a hard shutdown in the middle of an
update...

Andrew

PS Also, Python is reporting the Linux OS as "linux3" now that the kernel
major version has changed.  So any code that explicitly checks for "linux2" to
detect Linux is going to break.

An Idea a Day

From: andrew cooke <andrew@...>

Date: Sat, 10 Dec 2011 16:41:48 -0300

It's the weekend and I'm looking back at the AGU in the context of the
negative post above.

Fortunately, things worked out a little better. I decided to give myself a
break - to stop worrying about doing what I thought was expected for me from
work and instead trying to enjoy myself a little more.

So I started walking to the exhibition, exploring San Franscisco on the way,
and including various local museums in my schedule. I also took more time off
to eat, and ate better, and didn't spend so much time hanging around our booth
in the exhibitor's hall.

I also changed how I would learn about new things. Instead of going to talks,
which were in dark, depressing rooms, and which took forever to get to the
point, and were often a disappointment, I focussed on posters. In the morning
(there were new papers each day) I would visit the computing / informatics
section, since that is most relevant, and pay each poster a little attention.
in the afternoon, I would walk the length of the entire (huge) poster hall,
zg-zagging my way past one third of the posters (every third aisle), trying to
pick out something interesting. My aim was to try find one new idea a day and
then learn more on that one subject from the internet.

So the papers worked like a kind of guided random exploration for some
intitial search term that I could then explore in more depth.

Below are the things I noted for three days (Tue-Thur):

- NARMAX. According to the poster, this seemed to be some magical way to
turn a stream of data into a model. I later found that it was a non-linear
version of
http://en.wikipedia.org/wiki/Autoregressive%E2%80%93moving-average_model
which is, in turn, related to Kalman filters.
http://www.apav.it/sito_ratio/file_pdf/ratio_14/capitolo_4.pdf

- Higuchi Estimator. Seems that this can give you the fractal dimension of
some data.
http://books.google.com/books?id=0Jb9412yVGEC&pg=PA110&lpg=PA110

- Swift Shell. Used for scripting scientific data processing.
http://www.ci.uchicago.edu/swift/papers/SwiftLanguageForDistributedParallelScripting.pdf

Also, bonus link (via SFMOMA Dieter Rams exhibition) Teenage Engineering's
OP-1 synth: http://www.teenageengineering.com/products/op-1

Andrew

Accomodation in SF

From: andrew cooke <andrew@...>

Date: Sat, 10 Dec 2011 17:06:02 -0300

While I was at the AGU I stayed at the Pacific Heights Inn.  They have a web
page at http://www.pacificheightsinn.com/

I wouldn't say that they live up to the web page, exactly (it's a typical
courtyard style, two storey motel kind of place - cars in the middle and rooms
around the edge; not as homely as you might think from reading the blurb), and
the breakfast was a little odd, but it was clean, reasonably priced, and the
location is good (I'll expand on that below).

For breakfast, the front office has a couple of coffee machines, some plastic
containers of orange juice, and a bak or two of muffins or doughnuts.  The
first day I went down and sat there (at a single small coffee table) and ate.
But then I realised that it makes much more sense to take something back to
your room.  So I would put water in the coffee maker in my room, go down to
collect juice and a half-muffin, and then return to my room to eat with the
freshly made coffee.  For me, social recluse, it was perfect.

Better than the motel, though, was the area.  Or, more exactly, being quite
far from the Moscone centre, I was forced to explore a little.  And it turns
out that San Francisco is a much nicer place than you would think, if you stay
near the exhibition hall.

For travelling around there is a good network of buses (and streetcars).  The
site http://511.org/ does route planning.  At a kiosk on street level near
Powell (Hallidie Plaza) you can buy a "passport" for 1, 3 or 7 days use.  And
the 45 runs from outside the Pacific Heights Inn to the Moscone center, so if
you want to get in or back quickly, there's no problem.

But it was also pleasant to walk down Van Ness or Polk until close to Market
(turn left at, say, O'Farrell).  That way you pass through (particularly on
Olk) a pleasant area, and a bunch of restaurants.  And if you head in the
opposite direction you're just 10 minutes from a bay.  East you can get to
chinatown and west (further down Union) are a bunch of fancy shops.

If I go to AGU I will stay there again.  Having somewhere quiet, my own, away
from the conference, in a pleasant area, was a big plus.

Andrew

Comment on this post

C[omp]ute

Personal Projects

Last 100 entries

AGU 2011

In AGU's Defense (OpenSuse Kernel Bug)

An Idea a Day

Accomodation in SF