Two full non-stop days of work at Canada’s Perimeter Institute, filled with intensive study of the papers, notes and conference talks from the ATLAS and CMS experiments at the Large Hadron Collider, have left me slightly exhausted but pretty well up-to-date. This higher level of knowledge will start to percolate into these pages almost immediately.
Meanwhile, a bit more insight into the CRESST dark matter situation (see yesterday’s post). There have already been silly press articles. The BBC article says
- Researchers at the Cresst experiment in Italy say they have spotted 67 events in their detectors that may be caused by dark matter particles called Wimps.
No, that’s not what they said. (Not surprisingly, there are other serious scientific errors in the BBC article.) The CRESST paper said they see 67 events that look like a dark matter particle hitting an atomic nucleus in the detector, but some number of them — probably most of them — are caused by backgrounds. That is, they know there are other sources, perfectly conventional, for many of these events.
They list four types of known backgrounds, from various forms of radioactivity, all of which can create effects in the detectors that are sometimes not distinguishable from a particle of dark matter striking an atomic nucleus in the detector. The difficult task of the experimenters is to determine whether the known backgrounds could have given them 67 events. They claim the known backgrounds could not have produced so many fake events, and that instead between 1/3 and just over 1/2 the events are extras, which might therefore be from dark matter collisions.
Maybe. The question is whether the excess events that they see are from dark matter, or whether they are from a fifth type of background that they haven’t understood yet. Both are plausible at this point. Quoting from the paper:
We have estimated these four backgrounds and have found using a likelihood ratio test that, at a significance larger than 4 standard deviations, these backgrounds are not sufficient to explain all the observed events. Scatterings of WIMPs may be the origin of this effect and, under this assumption, we have derived the corresponding WIMP parameters.
I should add that there is something I currently find disconcerting about the way the analysis was done. In the data analysis they assume a standard WIMP (weakly-interacting massive particle, where weakly-interacting means interacting via the weak nuclear interaction) but a standard WIMP of this class is ruled out by two independent experiments, CDMS and XENON100, which are quite different from each other and from CRESST. To avoid CDMS and XENON100 constraints, presumably the dark matter particle should not be a standard WIMP. However in that case wouldn’t the method that CRESST used to determine the mass and interaction strength of the dark matter particle also need to be changed? possibly altering the statistical significance and the number of events? Will try to get insight from experts here.
Unfortunately CRESST did not plot all of their data in their paper. They have eight detectors, but they only showed the full data from two of them. This makes it difficult to do more detailed analysis; one just has to take what they say at face value. Perhaps they can be convinced to show the data from the remaining detectors.
I’m grateful to Neal Weiner of NYU, and Philip Schuster and Natalia Toro of the Perimeter Institute, who are more expert than I, for some insights and discussions.