Physics307L F09:People/Klimov/Poisson
Poisson Statistics
The surface of the earth is constantly being bombarded with cosmic rays. While the most common particle is the proton, there are many others, ranging from energies of giga electron volts all the way up through zetta electron volts (source: Wikipedia). Because these cosmic bombardment events are so ubiquitous and frequent, it is not difficult to detect them. The bombardment events have been studied extensively and are known to occur with a stable average rate. For this reason, such events are ideal for the study of the various statistical distributions, in particular the Poisson distribution. In this lab we set out to find a distribution which can accurately model cosmic ray bombardment at the surface of the earth (or at least in albuquerque). Cosmic rays were detected and counted using a NaI scintillation counter and a computer operator.
Useful Links:
Results
SJK 17:36, 17 December 2008 (EST)
As far as I know, there's no reason to expect one cosmic ray to be correlated with another (except for bursts and other variations in the average rate). That is, I think they should be Poisson distributed, not Guassian. So, I think the most likely explanation is that for some reason (which I don't know yet) the instrumentation is screwing up and / or we're not detecting only cosmic rays. Your results showing he dwell time effect are a big clue, but I don't know what it means. :)
During the experiment, we obtained a vast amount of data, taking full advantage of the operators settings. While our data was extensive, it has not given me a clear indication of whether or not cosmic bombardment can be described well with a poisson distribution. And in fact, it turned out that our data was highly dependent on the operator settings, which opened lots of room for interpretation. While our data looks like a poisson for large expectation values it deviates significantly for smaller expectation values. This would suggest to me that cosmic bombardment would follow a gaussian distribution, which would approach the theoretical poisson in such a way. The gaussian also seems more favorable to me because it is more flexible than a poissonian, offering two variable parameters instead of one. Although a large amount of quantitative data was obtained, there is no single value that can sum everything up.
Conclusions
This lab has more depth than I believed, at first. The main reason lies in the fact that the operator settings have a huge influence on the data, and one can spend a lot of time exploring them. Furthermore, I think a third lab period could be well spent on this lab, obtaining more data, tweaking with the apparatuses, and the operator software. Overall, however, I am pleased with the amount of data that were able to gather.
