Computing kaon decay

Lattice QCD – an approach to solving the theory of quarks and gluons – has been around for ages. I almost took my PhD in it, and that was … mumble, mumble … years ago. Lattice QCD involves formulating QCD in a discrete rather than continuous space-time: formulating it on a grid, or lattice, in other words. You can then use a supercomputer to investigate problems, such as quark-gluon plasma formation or confinement, that are beyond the realm of perturbative field theory techniques.

I had the feeling that practitioners were always promising great things of lattice QCD “next year” (a bit like some physicists have always been saying that cheap, reliable fusion power is “just five years away”). Things are really moving now, though: the computing power that was available back when I was a student is laughable compared to what’s available now. A recent paper in Phys. Rev. Letters demonstrates this: a team of US and UK physicists, running a calculation that took 54 million processor hours on the IBM BlueGene/P supercomputer at the Argonne National Laboratory, have simulated the decay of a kaon into two pions. The results of the calculation agreed with experimental results – but also provided a parameter that was hitherto unknown.

This is really exciting stuff. The discovery of CP violation (the phenomenon that generates the asymmetry we observe between matter and antimatter in the universe) came about through the study of kaon decays. It may well be that lattice QCD will be the tool by which we start to fully understand CP violation.