http://arxiv.org/abs/1604.06758
We apply the generalized Lomb-Scargle (LS) periodogram proposed by Zechmeister and Kurster, to the solar neutrino data from Super-Kamiokande (Super-K) using data from its first five years, and to Sudbury Neutrino Observatory (SNO) on data from both its salt and $\mathrm{D_2O}$ phases, in order to detect periodicities in these datasets. For each peak in the LS periodogram, we evaluate the statistical significance in two different ways. The first method involves calculating the False Alarm Probability (FAP) using non-parametric bootstrap resampling, and the second method is by calculating the difference in Bayesian Information Criterion (BIC) between the null hypothesis, viz. the data contains only noise, compared to the hypothesis that the data contains a peak at a given frequency. Using these methods, we assess the significance of various claims in the literature of sinusoidal modulations in the Super-K and SNO datasets. For the Super-K dataset (binned in 5-day intervals), the only marginally significant peak (from our periodogram analysis) is at a frequency of 0.0258/day, corresponding to a period of 38.75 days. The FAP for this peak is about 5.7% and the difference in BIC (between pure white noise and this peak) is about 4.8. All other claimed peaks in the Super-K datasets are consistent with noise. We also find no evidence for any statistically significant peaks in the SNO datasets.
S. Desai and D. Liu
Mon, 25 Apr 16
40/40
Comments: 8 pages, 4 figures. This manuscript is dedicated to the memory of Prof. Danuta Kielczewska
You must be logged in to post a comment.