Geek On Conference: Chaos Mathematics Day 2-5

The discoveries so far are:
1. "Mathematics is the language of the nature"... and I don't speak this language - but it's never too late to learn.
2. Chaos is NOT the same as randomness. There are relations between the behavior of elements in the chaotic system, while there are no patterns in noise.
3. With Alex and Stephan we build Zeeman machine (see photo below), collected data from it and fitted chaos model into it. This taught me a very interesting lesson about how catastrophe theory works.

4. European Parliament is cool.
5. Psychology is complex physics.
6. More Belgian Beer!
Carolus Classic (8.1%) - classic, strong, distinctive, dark, expensive... strong...
Chimay (~8%) - typical street-type, cheap, not-so-special
Westmalle Trappist Dark (6.5%) - very German, both in taste and design
Primus (5.1%) yet another cheap beer that is very light comparing to above, and taste a bit like Polish lager...
Duvel (8.5%) - a king's beer, my favorite so far, I have at least one every evening since I came here, beautiful beer that is very strong in the effect, and has very complex taste, you can get addicted, honestly
Leffe (?%) - typical, cheap, barrel bear, quite casual comparing to other possibilities
Abbeye des Rocks (9%) - it rocks, made but some monks I think, they know the shit, it makes your brain bubble withing 2 minutes ;-)
Palm (6.1%) - dark and slightly fruity, not impressed actually

Westmalle Trappist White (9.5%) - strongest I had so far, don't get cheated by light design of label, it is a killer, I only had small glass from this bottle and... I woke up the next day...

And the last photo presents a common viewpoint in Belgium ;-)

Geek On Conference: Chaos Mathematics Day 1

The Belgian beer experience day 1:
Mystic - 3.5% fruit(ish) thing for ladies
Haacht - 5.1% standard, tastes like... typical lager
Prior Tongerlo - 9% killer beer - you actually feel your brain melting when you drink it - very strong and distinctive taste

Beside this we are having day of 'R' training, so nothing really exciting to report... beside the fact that there are much more types of beers here to explore ;-)

Here is a link for the ESMP website.

Geek On Conference: Application of Chaos Mathematics in Psychological Research

(La Foresta, Belgium, 22nd May - 1st April)

This is an ultimate ERASMUS training. Fifty PhD students from 12 different universities and 14 professors closed in old monastery (now converted into conference center) for 10 days. We will try to figure out how to apply non-linear dynamic systems, catastrophe theory and Bayesian modeling in psychological research. Every day there are 10 hours of workshops, seminars, tutorials, practical sessions. It is hard. Extremely hard. But the interdisciplinary mix of people from psychology, neuroscience, linguistics, mathematics and computing are making it a unique workshop.

We only arrived today, but I will report regularly from La Foresta to update you on this amazing and strange training.

We will teach Robots our skills

The new exciting project has been launched, with 7 million euro funding from EU, as a part of our pan-european cognition and robotics strategy. AMARSi or Adaptive Modular Architecture for Rich Motor Skills, is a 4-year long interdisciplinary research, which aims to develop a new breed of robots, that will learn motor skills by constant interacting with human counterpart. The idea is to combine dynamic neural networks build on reservoir computing, and new approach to hardware design. Robots will be interacting with human worker using strategies of imitation, but the human caretaker will be able to physically correct robot actions at any time, giving him more direct feedback for learning, without time-consuming reprogramming. It's like when you are showing your kid how to do things, those robots will be intended to work in a very similar way. The design will be based on two existing platforms: infant robot iCub (RobotCub) and four-legged Cheetah (Biorobotics). So to summarize - AMARSi is more about developing a software for a new learning strategies for robots, rather than redesigning hardware. Nevertheless, creating the motor skills architecture to enable such open-ended behaviour is a real challenge, so I am looking forward to see how they manage to achieve that.

[photo credit: RobotCub]

What are Supercomputers doing at the moment?

As China joins the list of fastest Supercomputers on the 5th place, it is maybe worth to look at what are the main application areas for the top 500 Supercomputers.

It is always difficult to look at details of major projects run by supercomputers, because there is so much they are doing every second. For example the fastest supercomputer in the world, Jaguar, is able to run 1759 petaflops (1 petaflop = quadrillion operations per second) and has 224,162 processors. From their website it seems that astrophysics (supernova collapse), superconductors and fusion plasmas are top priorities, although other disciplines have reasonable share in Jaguar processing time.

It is hard to find out what research the Chinese supercomputer, Tianjin-1, is going to be used for. They indicated that petroleum exploration and the simulation of large aircraft designs would be major areas. Still, it belongs to The Chinese National University of Defense Technology (NUDT), so...

Now, what about general patterns of application for other remaining 500 supercomputers? Lets take first 6 major ones. While it is 'not specified' what is the application of 117 of them, 78 is used for research, 47 - finance, 34 - information service, 31 - geophysics, 30 - logistic services, 22 - semiconductors. Not bad. The question is - what kind of research (78)? If geophysics or superconductors are there, together with aerospace, medicine and life sciences, this must be cell biology or advanced physics.

Table 1: Supercomputers application area (source: top500.org)

However, things get complicated when we look at segments of market the computers are working for. The leading is industry - using 62.4% of them (312). This is followed by research - 18.2% (91) and academic segment - 15.8% (79). That would indicate (only by subjective correlation) that the majority of 'research' (from the application area table) is done for industry, and therefore have some commercial application. This might be pretty much everything, from new face-scrub recipe and GM crop to design of super-submarine or space craft. However, deeper inquire reveal that massive calculating power is bought by stock market companies for so called 'high-frequency trading' (HFT). In HFT supercomputers make calculations, predictions and decisions by analyzing ongoing flow of data from stock market and test it against historical data. And it all occurs very, very fast - they can handle 10,000 orders per second. HFT is a topic I will discuss on different occasion, some people contribute credit crunch on this strategy, but certainly we live in strange and beautiful times, your company has miliseconds advantage over the others, and it matters, and it already occurs without human intervention - how could it be differently now?

I guess that nicely summarizes the topic - supercomputers are where the money are. And what the do - well, calculate those money ;-)

Table 2: Countries with highest number of supercomputers and number of supercomputers used in different market segments (source: top500.org)

[data source: www.top500.org]
[photo credit: ORLN]