Uploads by Aaa014

This special page shows all uploaded files.

Date	Name	Size	Description	Versions
11:20, 1 June 2010	Thesis aasvold.pdf (file)	1.28 MB	Neutrinos escaping detection is one of the main problems in mass reconstruction with tau leptons. They must be neglected or corrected for in some way. This thesis discusses two methods of handling the neutrinos in different ways. The Collinear Approximation (CA) builds upon the assumption that the neutrinos travel in the same direction as the visible tau decay products. The boost method neglects the neutrino energy contribution in the leading visible tau, as seen from the mother particle's reference system. The methods have been studied with simulated \Z°→τ+τ- , \H°→τ+τ-, and QCD samples, and with early data from ATLAS. This work shows that the weaknesses of CA is that the transverse angle of the transverse missing energy has to lie between the transverse angle of the two visible taus, and that it collapses with back-to-back taus in the transverse plane. A strength of the CA is that it uses the missing energy, which is all information available about the neutrinos. The CA works better for boosted taus, i.e. taus decaying from heavy particles, like H° and Z°. The weaknesses of the boost method are that it does not use the transverse missing energy information, and that the distribution is not easily fitted, but the method is still under development on these points. The strength of the boost method is that it works for all tau pairs, making it a good complimentary method to the CA. Both methods work in \Z°→τ+τ- and \H°→τ+τ- events, and can be potentially applied to other decay chains as well. In the future, many studies will include mass reconstruction from tau leptons, where both the CA and the boost method will be important methods.	1
12:05, 1 June 2010	Masteroppgave.pdf (file)	705 KB	Hensikten med denne masteroppgaven er aa lage en laboratorieveiledning til en o�velse i et fag for laveregradstudenter ved Universitetet i Bergen. Foruten aa fi�nne en interessant fysisk verdi er hovedhensikten at studentene skal laere behandle resultatene paa en hen- siktsmessig maate. Dette innebaerer blant annet aa behandle usikkerheter, aa �finne en verdi som representerer hele maaleserien og aa kunne vise resultatene paa en god maate. O�velsen gaar ut paa aa maale forholdet mellom elektronets ladning og masse, e/m, med et katodestraalero�r. I forarbeidet blir tre metoder pro�vd ut; avb�oyning av elektroner i et magnetisk felt der en verdi av avb�oyningen bestemmer radiusen, ingen avbo�yning ved kryssede E-og B-felt og avbo�yning i magnetisk felt der flere punkter blir benyttet for aa estimere radiusen. Den sistnevnte metoden gaar ut paa aa estimere sirkelbanen og dens sentrum ved hjelp av min- imalisering av chikvadratet. Denne viste seg aa bryte sammen. Kun resultatene fra den fo�rste metoden inneholdt den kjente verdien av forholdet innenfor usikkerhetsintervallet, men usikkerheten var kun en sto�rrelsesorden lavere enn verdien for forholdet. Maalingene var veldig avhengig av usikkerheten i avlesningen av y-verdiene og denne usikkerheten kunne vaert forminsket ytterligere. Den andre metoden hadde lavere usikkerhet, men inneholdt ikke den kjente verdien i usikkerhetsintervallet. Denne metoden hadde ogsaa noen utfordringer knyttet til innsamlingen av data. Derfor blir kun den f�orste meto- den gjennomgaatt i laboratorieveiledningen. Denne blir utformet med en generell teoridel som skal gj�ore studentene mentalt forberedt paa det de skal laere og den inneholder en utf�orelsesdel som skal fungere som et stillas for studentene. Der staar det hint om hvordan �ovelsen kan gjennomf�ores og sp�orsmaal til diskusjon som skal besvares.	1
14:04, 1 June 2010	Thesis Svandal.pdf (file)	1.39 MB	The ATLAS working group has defined some safe variables that is well understood at the ATLAS early data taking. With these safe variables they have developed some safe cuts by using TMVA, that removes approximatly 70%(Loose), 50%(Medium), and 30%(Tight) of the signal. In this thesis I have made a likelihoodfunction based on the safe variables, figure 4.19, if we make cuts on this likelihood we manage to keep the same amount of signal while we reject more background than the safe cuts do. As shown in table 4.2 a cut on this likelihoodfunction will reject 5 times more background than the loose cut, and 2 times more background than the medium cut. This means that it can be a good idea to use those likelihoodcuts instead of using the safe cuts.	1
07:07, 2 June 2010	Masteroppgave Olausen.pdf (file)	705 KB		1