Sunday 1 July 2007

The Greatest Magic Trick Ever

...is this one, by Derren Brown.

So how does it work? Quite simple really. The brain tells itself:

Hey, this isn't the guy I was talking to just before! I was talking to Derren Brown!

Then the brain goes:

No no, I've just confused myself because someone walked past me with a poster of Derren Brown. I was never talking to Derren Brown.

Perhaps not in so many words. But that's the meat of it.

The trick doesn't lie in the switch - to switch two people behind a large poster is easy. All the magic takes place in the mind of the target, prompted by the image of Derren Brown.

While most magic tricks are designed to do the seemlingly impossible and leave the audience to try to figure it out, in this one, figuring it out is fairly straightforward. Nobody is left stupified.

Not us, the audience; there is no sleight of hand, no secret.

Nor the target; they don't even see the magic - not right away at least. A bit like how children sometimes miss the magic because for them, levitation and disappearing acts aren't impossible, but entirely ordinary.

It is very elegant.

But why do I think it is the best magic trick ever? Well, I'm not going to tell you.

This post is itself a magic trick - but while you are probably the target, you are not necessarily the audience.

Sunday 27 May 2007

Sports and Fairness

Today FIFA announced a decision to prohibit World Cup matches beyond altitudes of 2500 meters above sea level. The reason, ostensibly, is that the advantage offered to teams who regularly play at such altitudes is unfair.

Makes me wonder why FIFA doesn't ban nations with more than 50 million citizens from entering. With their larger pools of potential football players from which to draw their teams, these nations gain an unfair advantage, right?

Alternatively, FIFA could mandate that national teams are drawn at random. Just an idea.

Wednesday 23 May 2007

Determining the Transition Matrix

Back when I was introduced to Markov chains in Robert Messer's Linear Algebra, the prime real-world examples of what Markov chains might model were interacting populations in a biosphere. Oddly though, there was no mention of how one might determine the transition matrix based on observations of actual populations. Nor can I seem to find a simple explanation of this online. So here is my own calculation.

We wish to determine the Markov Chain transition matrix based on an observed set of transitions where the dynamics are unknown. The update function is
where p is the probability distribution, q is the old probability distribution, and M is the transition matrix. In order to find the optimal transition matrix, we minimize the variance
where d is the dimensionality and E is the size of the statistical ensemble over time and space to which p and q belong. The index of the ensemble e is suppressed. We find the differential
Setting the differential to zero we get
which, assuming Maj is known for all j ≠ b gives the solution
Our strategy then is to iteratively increase the size of the ensemble of observations and compute the transition matrix for each iteration. For j ≠ b we use the value Maj from the previous iteration. For the initial matrix there are many possible choices. We will use

To be continued...

Saturday 19 May 2007

I Wish I Could do This

Link.

I've tried all the Melbourne Shuffle instructional videos out there, but I just can't quite get the hang of it. What I need is an instructional video by Leroy of Stampa med Leroy fame. He has a nice, relaxed style and takes things slow so even a layman like me can keep up.

Wednesday 16 May 2007

Names for Cars

The automobile industry is obviously in trouble and needs my help. Nissan Xterra? Subaru Impreza? Need I go on?

Here are my suggestions:

Cherenkov
Doppler
Landau
Moebius
Sobolev
Tycho
Camus
Gödel
Dante
Dyson
Fourier
Napier
Schafer

Someone should pay me money for this.