Landmark Calculation Clears the Way To Answering How Matter Is Formed 205
First time accepted submitter smazsyr writes "An international collaboration of scientists is reporting in landmark detail the decay process of a subatomic particle called a kaon – information that may help answer fundamental questions about how the universe began. The calculation in the study required 54 million processor hours on the IBM BlueGene/P supercomputer at Argonne National Laboratory, the equivalent of 281 days of computing with 8,000 processors. 'This calculation brings us closer to answering fundamental questions about how matter formed in the early universe and why we, and everything else we observe today, are made of matter and not anti-matter,' says a co-author of the paper."
Re:Just another step closer... (Score:5, Interesting)
There conceivably could be an infinite number of "parellel" universes, but there's a real philosophical problem with that. So long as we use the real physicists definitions and not something out of Stargate SG1, those parallels will always remain undetectable. SF writers tell stories about interacting with other universes - physicists define them in ways that show they can't be interacted with to be verified.
An untestable idea isn't part of science. If it can't be disproven, it's philosophy or religion or something instead. An infinite number of untestable ideas is even worse. Philosophers get to whip out Occam's Razor at that point. If I claim that there is not only a God, but 7 different orders of angels totaling 144,000 beings working for him, those numbers are still simpler, in the sense Occam's Razor usually means, and so are to be preferred as a hypothesis. The same goes for a Million gods with an avarage of four arms each and a bunch of hidden cyclic time periods totalling quintillions of years for them to do their work in, or any of those models with a reasonably sized bunch of gods, and maybe some giants, dwarfs, dark elves, ninja turtles piza delivery robots, a billion clones of an invisible pink unicorn who died for your sins, riding on a gigantic fiberglass replical of L. Ron Hubbard, and so on. Just about any other idea looks preferrable to an idea that postulates an infinite number of unverifiable consequents.
Re:Missing Details and Corrections (Score:3, Interesting)
Thank you for the explanation, it is far more informative than the blog post.
I'm a little depressed with how little of that explanation I understood - I'm a 3rd year physics PhD student, writing a thesis on matter-antimatter interactions (specifically, low energy swarm theory with liquids), and even I only have a very loose grasp of what you're talking about. I suppose it says something for how specialized physics really is. 99.9% of people in the world would think that we're studying the same thing.
Re:And this is why we need distributed computing. (Score:5, Interesting)
there is still a niche for traditional supercomputers left...
I don't think you know much about supercomputers. Sure, there are a few problems that are embarrassingly parallel [wikipedia.org], but most aren't. For those that aren't, the bandwidth and latency of the interconnect between different processors is more important, and often more expensive, than the speed of the processors themselves. This is why most supercomputers use exotic interconnects like Infiniband, Myrinet or 10GigE, and linking nodes together using complex topologies such as a torus.
Case in point: on the website of the QCDOC supercomputer, which was partially used in this study, they say that a highly optimized lattice QCD simulation achieves up to 50% CPU utilization, and this is considered very good. The rest of the time is mainly spent waiting for the interconnect.