Data Center



UK supercomputer probes secrets of universe

CP violation, kaons and the Standard Model


A team of theoretical particle physicists at Edinburgh University has applied some hefty computing power to one of the great unanswered questions of particle physics - why does the universe contain more matter than antimatter when, on paper at least, there should be equal amounts of both and we shouldn't exist at all.

The researchers have focused their efforts on probing this mystery, known as the "CP violation", which was confirmed in 1964 by James Cronin and Val Fitch in the decay of neutral kaons ("strange" particles containing strange quarks).

CP violation runs contrary to the principle of "CP symmetry", which states that the sum of two symmetries - charge conjugation (C), which transforms a particle into its antiparticle, and parity (P) - should result in an equal amount of matter and antimatter.

However, it ain't necessarily so. Enter kaons, aka K-mesons, whose decay indicates CP violation.

Edinburgh University has now, in collaboration with Columbia University, Japan's Riken, the US's Brookhaven National Lab and IBM Research, provided evidence as to how CP violation sits with the Standard Model - the "internally consistent" theory describing the electromagnetic, strong nuclear and weak nuclear interaction between particles.1

The Standard Model has to date stood up to scrutiny, albeit with the proviso that the "Higgs Boson" - the as yet elusive particle required to give mass to the remaining 16 particles described by the Standard Model - actually exists.

Edinburgh Uni's QCDOCTo scrutinise CP violation in terms of the Standard Model, the Edinburgh researchers and their collaborators built (see pic) a "14k node class of computer, QCDOC (QCD-on-a-chip) based on our own custom compute chip with integrated communications hardware using then pretty leading edge PowerPC SoC and Edram technology", as team member Peter Boyle put it.

Boyle elaborated: "We designed a custom 4MB on-chip embedded dram memory and a custom 6d torus message passing network, which they integrated with an embedded PowerPC 440 core and FPU (similar to those later used in BlueGene/L) on their QCDOC ASIC, which draws a frugal 5W of power.

"We have three sister machines - one in Edinburgh, and two at Brookhaven National Lab - each having a peak performance of over 10Tflop/s and sustaining around 5Tflop/s. The nodes run a custom embedded operating system developed by the team, which spends most of its time running optimised assembler code at around 40 per cent of peak performance."

The results of the beast's three years of "Lattice QCD" calculations are, according to Boyle, "quite beautiful".

He told El Reg: "Two of the tightest theoretical constraints on fundamental Standard Model parameters related to 'CP asymmetry' have been produced by QCDOC - CP violation being a requirement for protons to be left over in excess of anti-protons after the big bang so that we weren't all annhiliated in the cosmic mush. However, the Standard Model still can't explain why there is quite so much left over."

Boyle concluded: "Whether or not the CP violation in kaons is correctly predicted by the Standard Model is difficult to know without better calculations because it is was only known to around 10-15 per cent. We reduced that uncertainty to five per cent.

"As we know the Standard Model does not well explain the amount of matter in the universe we expect studying this area with ever greater precision will eventually turn up new physics laws that do explain the matter content of the universe."

Whether the CP violation demonstrated by kaons will indeed display "small anomalies" that are evidence of new physics which can explain the preponderance of matter remains to be seen. That's a job for CERN's Large Hadron Collider (LHC) - which will later this year probe what happens when you smash protons into each other at rather high speeds. Watch this space... ®


1 The Standard Model has yet to encompass the force of gravity, which is its fundamental shortcoming.

Sponsored: Accelerated Computing and the Democratization of Supercomputing