A powerful new algorithm

I have been working for a month on applying an old idea of mine to a problem that is becoming more and more outstanding in CDF. The algorithm is two years old. It is called "Hyperball Algorithm", and I invented it and used it with some success in the context of the Higgs Sensitivity Working Group in 2003 to prove that CDF can indeed improve the dijet mass resolution for a Higgs boson decay to the level a previous study had claimed to reach - a bit far-fetchedly - in 1998.

The tool is more general than the very specific problem it was born to solve, and I decided a month ago that some more play with it was advisable. The occasion was the CDF Collaboration Meeting in Barcelona, where I was to present the status of activities focused on the improvement of the jet energy resolution: setting up to put together a talk, I had discovered I had not much to say, and proceeded to produce something brand new, by applying the hyperball algorithm to the problem of improving the energy resolution of b-quark originated jets of hadrons.

This post would become too long if I were to explain what the algorithm does and its niceties. That will be the subject of the next post. Here, instead, my twentythree readers will have to make do with a slide from the Collaboration Meeting talk, which contains a plot of the improvement in the relative energy resolution (error on energy measurement divided by true energy) I am obtaining as a function of the true energy of the quark originating the jet (it is shown on the right, while on the left the fits to the energy curves before and after correction are shown with inverted colors - stupid me!). The red points in the plot on the right show the resolution before the correction, and the blue points show what one gets after the correction. I am confident that I can show this plot although it is not blessed, since it is only Monte Carlo - no government-owned real data!

Ok, so much for now, I will collect more energy and explain what this is all about later...

Comments

[manuel visaya]

here is a theory (MNT) that straddles between quantum theory and queueing theory originally brought up with a statistician working on the 411 directory last thanksgiving day. upon observing sharks in a local aquarium i asked, do shark arrivals follow a distribution?

p(x,mu)=((e^(-(Sup(mu))))*(mu^x))/(x!) , (modified poisson distribution)