Friday, April 29, 2011

More Muon1 Information

I guess this will also become the semi-official Muon1 blog, at least for now.

There are some common questions that are asked about Muon1, including 'what is it about', 'why should I bother?', 'what's so special about it?' and 'what does it do?'

So let's answer them, shall we?

What's it about?

It's about finding the most optimal design for part of a particle accelerator. Specifically an accelerator that will generate Neutrinos, in an attempt to detect them. Neutrinos, by their nature, barely interact – there are BILLIONS passing through your body each second. Being able to create a reliable, sensitive detector not only tells us a lot more about the particles, but can open up new possibilities, such as communication links that go THROUGH the Earth, rather than around it. If you want to see a bit more about the overall Neutrino Factory Project, there's this handy infographic

Why Should I Bother?

Advances in science often come with the fundamentals. Understanding atoms meant we could use nuclear materials effectively. Our understanding of photons has led to lasers, and fibre-optic cables. Understanding Neutrinos can lead to a massive shift forward in technology and scientific research. While [email protected] is looking for radio signals from Extra-Terrestrials, perhaps the signal is being send by Neutrino, since it has less 'interaction' and thus less degradation

What's so special about it?

Muon1 is one of the very few projects that doesn't follow the traditional 'brute force' approach.
In most projects, there's a fixed batch of work. You request some work from the project, process it, and send it back, then request some more. You can only work if you're internet connected, usually.

Muon1, by contrast uses a genetic algorithm method. The workspace for each lattice (sub-project) is so huge, and each one so time-intensive, that to attempt a brute force method would be pointless. Instead, the client focuses on what works, and tries to make it work better. Starting with random designs, the client tweaks things to make things better, using its own results as a baseline. These results, when returned, can be used to compile team or project-wide 'best design' lists, which the client can incorporate in its data. It's an intelligent way of analysing a problem, following design leads, rather than mechanically processing things sequentially. It also allows for human interaction since work is not dictated, participants are free to try their own designs.

Here's a visual representation of the differences.
click to enlarge

It's also incredibly modular. To try another design theory, a new lattice file is all that's needed, which tells clients the parameters for generating simulations. The time between thinking of a new design idea, to having the first results come in can be as little as 12 hours. It's a level of flexibility no other project can touch.

What does it do?

In short, “It simulates particles in a particle accelerator”

The more complex answer is it tracks particles given off by a rod that has been struck with a high-energy proton beam. It follows tens of thousands of them, as they move down a series of components, which attempt to focus the particles and get them to the right energy. It works in 0.01nanosecond (0.0000001 of a second) steps, as they go through the design. Then, when it's done, it tweaks the design based on that result, and previous results, and repeats.

It's not something that's easy to describe, and it is better shown. So, there's now this video showing a typical simulation in v4.45. It's a little long, but is shown in it's entirety. (it may be better to run it full-screen)








If you want to know more, leave a comment here, or on the project forum, and you can join the project at it's website:
Or join the facebook group, and get weekly updates on progress, as well as other news.





No comments:

Post a Comment