A Study of Hadronic Final States from e e- Annihilation as a Function of Center of Mass Energy by Michael H. Capell B.S. (Physics), University of California, Davis (1979) Submitted to the Department of Physics in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology January 1986 Massachusetts Institute of Technology 1986 Signature of Author Signature redacted SigrSjgnature redacted Certified by Department of Physics January 6, 1986 grProfessor Ulrich J. Becker Thesis Supervisor Signature redacted Accepted by Professor George F. Koster MASSACHUSETTS INSTITUTE OF TECHNOtOGY Chairman, Departmental Graduate Committee FEB 14 1986 LIBRARES Archives 1 Abstract A Study of Hadronic Final States from e+e- Annihilation as a Function of Center of Mass Energy by Michael H. Capell Submitted to the Department of Physics in partial fulfillment of the the requirements for the degree of Doctor of Philosophy, January 1986 The multihadron data from e+e- annihilation collected with the MARK-J detector at PETRA over the center of mass energy range 22 < Vs- < 47 GeV have been examined. Several measures of gluon bremsstrahlung and fragmentation models were used. The strong coupling constant has been determined from the entire energy range to be: a,(Vw/= 34.6 GeV) = 0.119 0.005 (statistical) 0.020 (systematic) by comparing the predictions of perturbative QCD through complete second order to the data. Including lower energy measurements the strong coupling constant has been shown to run with center of mass energy, the functional dependence being that predicted by QCD. The measured dependence of event topologies on F has been directly compared with the perturbative calculation via the Energy-Energy Correlation Asymmetry. Use of the data over a restricted range allowed the determination of an a, value consistent with those obtained with current fragmentation models over the full range. An improved description was obtained over almost the full range with a simple, quantitative, parameterization of the non-perturbative effects which yielded a QCD scale parameter of A = 114 +I4 (statistical) MeV in the MS renormalization scheme. Thesis Supervisor: Ulrich J. Becker Title: Professor of Physics 2 Acknowledgements My first acknowledgement is made with pleasure to the people who have constructed and run the MARK-J experiment. My thesis supervisor, Ulrich Becker, has a clear insight into what it is useful to measure and how to measure it, which is the root of experimental physics. I thank him for teaching me some parts of this and for sharing his experience in the field. The groundwork for the analysis presented herein was laid by Prof. J. G. Branson, H. Newman and R. B. Clare. This study and I have profited from the attentions of Prof. M. Chen, G. M. Swider, R.-Y. Zhu and Ll. Garrido. I thank Prof. Samuel C. C. Ting for his leadership in building the MARK-J detector and collaboration, and for the efficient way he has helped to solve the resulting physics questions. I thank A. Dworak, J. Hudson, R. Meisel, P. Slade and Dr. S. M. Ting for their administrative support. A debt of thanks is owed to the DESY directorate and the PETRA Machine Group for proving a good place to do high energy physics, and to the people of America and Germany who have helped to support MARK-J, PETRA, and this study. Part of this work was supported under DOE contract DE-AC02-76ER03069. Other, but no less necessary, support has been provided by my friends and family, especially Susan. 3 Table of Contents A bstract ................................................................................. A cknow ledgem ents ....................................................................... 2 3 Chapter 1 Introduction ............................................................... 5 Chapter 2 2.1 2.2 2.2.1 2.2.2 Experim ent ................................................................ 7 The MARK J Detector at PETRA ........................................... 7 Hadronic Event Selection and Reconstruction.................................9 Topological Considerations.................................................9 Selection and Analysis .................................................... 10 Chapter 3 3.0 3.1 3.2 3.3 3.4 Theories, Models and Their Implementation .......................... 12 Present Assumptions Used to Describe the Data e+e-- Hadrons............12 12 T he Initial State ............................................................ Perturbative Q CD .......................................................... 13 H adronization .............................................................. 17 Detector Simulation.........................................................18 Chapter 4 Determination of the Strong Coupling Strength as a Function of Center of Mass Energy ............... 4.0 4.1 4.2 4.3 4.4 4.5 Methods in Measuring Gluon Bremsstrahlung ............................... a, from the Energy Flow Variables..........................................22 a. from the Jet M ultiplicity.................................................25 a, from the Energy-Energy Correlation Asymmetry ......................... C om parison ................................................................. Measurement of the Running of the Strong Coupling Strength ............... Chapter 5 A Stringent Test of QCD Versus Energy...............................34 Chapter 6 C onclusions...............................................................37 R eferences .............................................................................. Tables .................................................................................. Figure C aptions.........................................................................51 F igures ................................................................................. 4 20 20 26 29 31 38 43 55 Chapter 1 Introduction The strong or nuclear interaction is long known but not well understood from basic principles. The strong force manifests itself in the interaction of hadrons. This study investigates multihadron final states produced by e+e- annihilation in order to gain insight into the strong interaction. Quarks were proposed by Gell-Mann [1-1] and by Zweig [1-2] as the mathematical building blocks of strongly interacting particles. Just three different "flavors" of fractionally charged spin I quarks and the symmetry of the group SU(3) [1-3] were required to construct the dozens of known hadrons: mesons were "built" from quark-antiquark pairs and baryons from triplets of quarks. A success of the model was the prediction of the existence and properties of a new particle [1-4], the 11-, which was subsequently discovered [1-5]. However, the 1- pointed out the disagreement of the simple quark model with spin statistics. As a ground state baryon of spin 1 the l- was postulated to be made up of three identical spin i quarks, violating Fermi-Dirac statistics. Greenberg solved this by introducing an additional quantum number for the quarks, called color, which could take three values [16]. The f- was explained as being composed of 3 quarks in a color antisymmetric state. Antiquarks were assigned anticolor. Color tripled the number of "fundamental" particles, solving two other puzzles. Calculations of the w* decay rate [1-7a] and of the total hadronic cross section in e+e- annihilation [1-7b] both needed a factor of three more channels to agree with the data, and color provided this factor. The idea that quarks are more than a tool for building hadrons was advanced by the deep inelastic lepton-nucleon scattering experiments at SLAC [1-8]. These revealed pointlike constituents - partons - within nucleons, with all the properties of quarks, including fractional charge [1-9]. The lepton-parton scattering results also showed that, within the nucleon, partons behaved as quasi-freely moving particles. This property, that at short distances or large momentum transfers the partons are unbound, has become known as asymptotic freedom. Later experiments revealed roughly collinear fluxes of hadrons moving in the expected directions of the struck parton and the remaining target partons [1-10]. This led to the concept of confinement: partons do not appear as free particles at large distances, rather, they dress themselves into hadrons. If a parton has sufficient momentum in the Lab frame, this dressing appears as a "jet", i.e., a collimated, or limited PL, flux of hadronic particles or energy. To observe jets, they must be isolated by regions of low flux, such that the angle between jets is larger than the angular width of the jets. The simple quark-parton model with color did not explain a further result from deep inelastic scattering, that only one half of the nucleon momentum was carried by the charged quarks [1-11]. Direct evidence that quarks are the physical building blocks of hadrons came with the discovery of the J/0' [1-12]. This was interpreted as the bound state of a quark-antiquark 5 pair with a new quark flavor, charm, in the same way that analysis of the hydrogen spectrum revealed the component electron and proton [1-13]. Futhermore the binding potential could be derived from a simple model of confinement. Quantum ChromoDynamics (QCD) is the theory of the strong interactions which has grown out of these observations [1-14]. Similar to Quantum ElectoDynamics (QED), it is a renormalizable gauge theory [1-15]. The interaction of charges in QED is replaced by the interaction of colors in QCD. The Abelian U(1) group symmetry of QED is replaced by the non-Abelian SU(3) group symmetry of color. Massless spin 1 quanta called gluons carry the strong force between colors, analogous to photons in QED. In contrast to electrically neutral photons, gluons carry one unit of color and one unit of anticolor. With three colors (and three anticolors) there are nine gluons. The group symmetry arranges these nine into an octet and a singlet (3 0 3 = 8 E 1). The coupling of the color neutral singlet gluon is set to zero. The colored octet gluons interact with both quarks and with other gluons. Gluons account for the missing momentum in the nucleons. The direct evidence for gluons came with the observation that about 10% of the multihadron events in e+e- annihilation at Fs > 25 GeV appeared as three jets [1-16]. The agreement of both the rate of these events and the angular distribution of the jets within the events with the expected process of gluon bremsstrahlung from the quark or antiquark [1-17], was a major success of QCD. In QCD the strength of the quark-gluon and gluon-gluon interactions is determined by the strong coupling constant, a,, in the same way as the fine structure constant, a, determines the strength of the electric charge-photon interaction. To account for both asymptotic freedom and confinement the coupling in QCD "runs". It is small at short distances and large at long distances (fractions of a fermi). In momentum space this means that the coupling is small when gluons are transferring large amounts of momenta (a few GeV). The perturbation series expansion of QED in terms of a is paralleled in the expansion of QCD in terms of a, to provide a calculable theory. Perturbative QED is useful over a wide range of energies, but, because a is large for processes involving low momenta, the expansion in QCD is only expected to be valid for reactions involving large momenta. Even in this region a, is larger than a >> e = ) and the effects (a,of ~higher to orders are worrisome. a This study uses the data collected with the MARK-J detector to precisely determine the strong coupling constant, a., over the wide energy range available at PETRA, 22 < v/ < 47 GeV. The functional dependence of the running of the coupling constant is tested over this range and with the inclusion of lower energy data. The energy dependence of the data is used to probe the validity of the present theoretical understanding and to quantify the as yet uncalculated effects. 6 Chapter 2 2.1 Experiment The MARK-J Detector at PETRA The e+e- storage ring PETRA [2-11 and its injection apparatus are diagrammed in Fig 2.1. Electrons are injected from Linac I into DESY, accelerated to 7 GeV, and injected into the PETRA ring to form two bunches. Positrons, produced in Linac II and accumulated in PIA, are likewise accelerated in DESY and injected into PETRA to form bunches. Each bunch contains (1 to 2)-1011 particles. As the ring circumference is 2.3 km this yields a current of 2 to 4 mA/bunch. The two pairs of counter-circulating bunches are further accelerated and focused to collide inside the four detectors, with a crossing frequency of 250 kHz. After the first physics runs in November 1978, the ring elements were modified to provide higher luminosities and higher beam energies. The peak instantaneous luminosity obtained with the MARK-J detector was 1.6 -10 31 /cm 2 /sec, with up to 650 nb- 1 collected in one day. The integrated luminosity collected with the MARK-J detector over the resulting large range of center of mass energy, ,F, from 12 GeV up to the world's present highest e+e- energy of 46.78 GeV, is shown in Fig 2.2. The MARK-J detector, Fig 2.3a and 2.3b, is composed of a vertex detector, electromagnetic and hadron calorimeters, and a muon spectrometer. Particles leaving the e+e- interaction region traverse the detector layers shown schematically in Fig 2.3c. Outside of the beampipe is the vertex detector (labelled DT or drift tubes in Fig 2.3). The next layer is the electromagnetic or inner calorimeter (A,B,C). Surrounding this calorimeter are the inner drift chambers of the muon spectrometer (S,T). Proceeding radially outward the magnet toroids of the spectrometer form the absorber for the hadronic or outer calorimeter (K). The outermost part of the detector completes the muon spectrometer with trigger counters (D), more magnetized iron, and drift chambers (P,R). Hadronic events are selected using the vertex detector and the calorimeters, the analysis of these events utilizes the reconstructed energy depositions in the calorimeters. The remainder of this section provides a description of these detector elements (more detail can be found in [2-2]). The following section describes event selection and reconstruction. The vertex detector measures the charged multiplicity and the vertex position along the beamline. It is composed of 2616 cylindrical drift tubes arrayed perpendicular to the beams in four rectangular layers. Each tube is 1 cm in diameter and 30 cm long. The array actively covers the polar angle from 0 = 10' to 170* and the entire azimuth less 70 in each corner. The resolution per tube is 0.03 cm. Tracks are fit to hits in each layer. The best overall fit for an event is determined by constraining the tracks to a common vertex along the beam direction, z, and minimizing the X 2 per track. The number of tracks pointing to this common vertex yields a measure of the charged multiplicity. The r.m.s. widi of the best fit vertices for hadronic events is just that expected from the PETRA bunch length, Az = 1.3 cm. 7 The inner calorimeter provides the bulk of the information about hadronic events. It is subdivided into three azimuthally segmented layers of shower counters (A,B,C in Fig 2.3). Each counter is made up of 0.5 cm thick pieces of scintillator alternated with 0.5 cm thick lead plates. At normal incidence this yields 3, 3 and 12 radiation lengths or a total of 1 absorption length. The 20 A counters are arrayed parallel to the beamline outside of an iron box which surrounds the vertex detector. They cover the polar angle 0 = 120 to 1680 with no azimuthal holes. The 24 B and 16 thicker C counters are arranged similarly to the A counters, but are offset in 0 and shorter, the B counters covering 0 = 160 to 164' and the C counters 0 = 260 to 1540. Each of these sixty counters is instrumented with a phototube at each end [2-3]. The time and magnitude of both phototube pulses are digitized with TDCs and ADCs and recorded. The longitudinal hit position is measured by a weighted average of the positions determined from the time difference between the two pulses and from their relative magnitude. Comparing to the the positions extrapolated from tracks fit in the vertex detector this method yields a resolution for single hits of A0 = 50 per counter. The azimuthal segmentation combined with shower sharing between counters yields a resolution of AO = 7*. The energy deposited in each counter is determined from the two pulse heights corrected for attenuation. From the direction with respect to the interaction region and magnitude of the energy deposited in a counter an "energy vector", $4, is formed. This is not a proper vector. Under addition the resultant direction is taken as the vector sum and the resultant magnitude as the algebraic sum. The leakage of electromagnetic showers from electrons and gamma rays into the outer calorimeter is less than 4%. The resultant resolution is AEIE = 7% at E = 17 GeV. On average, a hadronic event deposits 35% of its energy in the A counter layer, 15% in B and 25% in C. Distributions of the energy deposited in the inner calorimeter are shown in Fig 2.4 and 2.5. The outer calorimeter absorbs the remaining 25% of the energy from a hadronic event. 192 scintillation counters are arranged in four layers interleaved with 2.5 to 10 cm of iron for a total of 2 absorption lengths at normal incidence. The two inner (outer) layers cover the polar range 0 = 430 to 137' (260 to 1540) with azimuthal holes of 100 (40). In this calorimeter the 4 resolution is ~ 2* better than in the inner calorimeter because of the finer segmentation. The longitudinal resolution is worse because only one end of each counter is viewed by a phototube. To improve this resolution clusters of hits in the same ( 90 in 4 and t ~ 0.1 in cos 0) solid angle are assigned to a "counter track". The longitudinal hit positions in the outer calorimeter are adjusted on to these counter tracks. The resulting angular distribution is shown in Fig 2.6. Using the position information from the inner and outer calorimeters results in the energy weighted 4 distribution of counter tracks shown in Fig 2.7. The angular resolution for the axis of a hadron jet, determined from Monte Carlo studies, is A0 =Z\ cos 0 - A0 = 70 for the entire detector. The active solid angle is 95% of 47r. 8 Hadronic Event Selection and Reconstruction 2.2 2.2.1 Topological Considerations The "signal" events for this study are multihadron final states resulting from e+e- annihilation into a virtual photon. The procedure for selecting these events from the other channels and backgrounds which constitute the "noise" is best prefaced by descriptions of their respective appearances in MARK-J. Fig 2.8a is a schematic representation of a typical one photon hadron event; the five major sources of background are diagrammed in Fig 2.8b-f. Fig 2.9 shows a computer reconstruction of a typical "signal" event, taken at a beam energy of 22.05 GeV. The characteristic high multiplicity of penetrating particles appears in MARK-J as several tracks in the vertex detector (11 for this event, Fig 2.9b) originating from the e+einteraction region and significant energy deposition in the outer calorimeter (9.2 GeV). This is distinct from electromagnetic final states which are contained in the inner calorimeter. Two other distinctive features visible in Fig 2.9 are that the energy depositions in the calorimeter are balanced across the origin and the sum of the energy depositions, EV, is close to the avail2 E~eam. Specifically, this event has net energy imbalances able center of mass energy, Fs of 6.0, 4.2 and -0.4 GeV in the x, y and z directions and a total reconstructed energy of 41.5 GeV out of 44.1 GeV available in the center of mass. Tau pair production, Fig 2.8b, where both taus decay semi-leptonically into hadrons, is characterized by the missing energy carried off by the neutrinos and by two narrow jets of hadrons. This is observed as an event with a net energy imbalance, reduced Ey and low multiplicity, though a significant fraction of energy may penetrate to the outer calorimeter. Tau pairs can also contaminate the sample by one tau decaying leptonically into an electron and the other into hadrons. This appears as a narrow, low multiplicity hadronic jet back-toback with an electromagnetic shower. The appearance of two photon multihadron events, Fig 2.8c, depends on the number of tagged electrons. If both electrons are scattered sufficiently to enter the detector they are observed as nonpenetrating showers in the inner calorimeter with a single matching track in the vertex detector. An electron escaping down the beampipe leaves behind an event which is reconstructed with a net energy imbalance along the beam direction and with reduced Ey. The hadronic system may be roughly balanced but its energy is substantially less than 2EBeam, hence it leaves a reduced fractional energy deposition in the outer calorimeter. Bhabha scattering (ee- _- e+e- including radiated photons), Fig 2.8d, leads to < 2 tracks in the vertex detector with 2 or 3 electromagnetic showers in the inner calorimeter and no energy in the outer calorimeter. The bulk of these events occur at low scattering angles. Electron-beam gas scattering, Fig 2.8e, appears as a highly unbalanced event with E ~ EBeam and usually a reconstructed vertex away from the e+e- interaction region. High energy cosmic rays, Fig 2.8f, interacting in the outer parts of the detector can deposit large amounts of energy in the calorimeters. These events are distinguished by their 9 external origin, as indicated by the muon spectrometer and by time-of-flight between the muon trigger counters (D in Fig 2.3). Cosmic rays which sneak through the spectrometer only rarely produce tracks in the vertex detector which point to the e+e- interaction region, and the energy depositions are typically concentrated in one area of the detector. 2.2.2 Selection and Analysis The selection of signal events from the noise is done stepwise. The first step in the event selection is implemented by the online data acquisition system [2-2,4]. Of the eight triggers employed to initiate data collection, two are significant for multihadron events. The first relies on the coincidence of hits in several calorimeter layers, the second on the coincidence of energy depositions in groups of counters opposite each other in 0. On receipt of either of these triggers a fast sum of the total energy deposition is calculated. This approximate sum is conservatively required to exceed fI/6, suppressing beam gas, cosmic ray, and two photon events where neither electron is tagged. Combined with the other triggers the typical trigger rate is 5 Hz and the accepted trigger rate is 2 Hz, with a deadtime of less than 9%. Monte Carlo studies indicate the trigger inefficiency for the signal events is less than 1%. The energy depositions from the accepted triggers are then reanalyzed offline by a fast, but approximate, algorithm. At this step the total energy requirement is increased to EV > 0.30fi and a loose balance cut of energy depositions greater than 0.10of in opposing # quadrants is applied. These two cuts reduce the sample 75-90% (depending on beam conditions) by further rejecting beam gas and low energy two photon events. Monte Carlo studies indicate 95% of the signal events pass this step. The remaining candidates are accurately analyzed. vertex detector (e.g., see Fig 2.9b). Tracks and a vertex are fit in the The energy vectors, fi, of each hit in the calorime- try are precisely determined from the combination of ADC and TDC data with calibration values. Counter tracks are formed as described above. The visible energy require- ment, EV =_ E|J4|, is increased to 0.35N/s and the net energy imbalances perpendicular, AE.1 = /( Ei ) + (, Ej , and parallel, AE,, =1 - . + - [ , to the beamline are re- quired to be less than 0.60EV. The remaining cosmic ray events are rejected via the difference in the hit times between the outer muon trigger counters. After this step the sample contains one and two photon hadronic events, Bhabha events, and tau pair events. A complicated pattern recognition algorithm then attempts to sort each event into one of these categories. Each counter track is assigned probabilities of being hadronic, electromagnetic, or the result of a tau decaying into hadrons. This assignment is based on: the charged multiplicity of the counter track, estimated from the number of matching vertex detector tracks; the energy fraction of the track in the outer calorimeter, this fraction being a function of the direction of the track; and the total energy of the track, this also being a fraction of the track direction. The event is then viewed as belonging to each of the categories and the assigned probabilities folded with the possible sources. Events which do not strongly agree 10 with exactly one hypothesis or have an unusual feature are classified by physicists scanning detailed event pictures such as Fig 2.9. (Examples of unusual features are a vertex fit > 3away from the interaction region or large energy hits in the inner calorimeter.) About 20% of the sample at this stage requires this manual classification, half of which is accepted. Of the other 80%, 5-10% are scanned as a cross check. The data used in this study were collected over the last six years and over a wide range of energies, so systematic variations in the detector response are a concern. These can be induced by scintillator aging, by decreasing phototube gain, by ADC pedestal shifts, etc. During data taking and in the subsequent analysis chain these variations are largely eliminated by recalibration. The calibration values for each counter and its associated electronics are derived from studies using Bhabha events, cosmic rays taken between beam crossings, and multihadron events [2-5]. They are readjusted every 2-4 weeks. For hadronic events this has maintained the total energy resolution at A(Ev/v/) ~ 18% independent of - as shown in Fig 2.10. V The dark circles in Fig 2.11 show the mean values of detector related quantities versus measured using hadron events, and display the consistency of the detector response versus both time and V/-. Studies using both data and Monte Carlo indicate that events with Ev 0.50fi are predominately multihadron final states from the one photon annihilation channel, with an acceptance of 83% at F > 32 GeV, and with a contamination from two photon processes of 1.3% and from tau pairs of 4.5%. To reduce these two remaining backgrounds to negligible levels and to insure accurate reconstruction of the event shape by the calorimetry, the events 0.70v/-, as marked in Fig 2.10, and AE 1 < used in this study are required to have Ev 0.50Ev, AE, < 0.50Ev, as marked in Fig 2.12. The data are distributed in F as shown in Fig 2.13. For the bulk of the analysis presented below, these data have been combined into the bins summarized in Table 2.1. The data at 14 GeV is not used in this study because it is near to the b-meson threshold and both b-production and b-decay are not well understood in this region. 11 Theories, Models and Their Implementation Chapter 3 Present Assumptions Used to Describe the Data e+e--+ Hadrons 3.0 Only a partial understanding of the process of e+e- annihilation into hadrons is existing. This study aims to increase this understanding. This chapter sketches the theories and models used to describe the data, and how they have been implemented for the analysis presented. At present e+e- annihilation into hadrons is postulated to proceed as sketched in Fig 3.1, [3-1]. From left to right the abscissa can be taken as increasing time, decreasing particle momentum, or, roughly, decreasing understanding. The three subprocesses, (1) e+e- _- y* (2) -y* -+ qq, -+ qqg and (3) q, g -+ hadrons, are assumed to have independent probabilities The first step is well understood using QED with Electroweak which can be multiplied. The next step, involving the high momentum quarks and gluons known as modifications. partons, is described by perturbative QCD. The final step, the evolution and association of partons into real hadrons, is called hadronization or fragmentation. This step has no solid theoretical basis and resort is made to models. At PETRA energies the association during fragmentation results in the hadrons appearing collimated in jets. The jets of hadrons are the observables and they reflect the original parton energies and directions. The utility of the factorization is that it allows the calculable, perturbative steps (1) and (2) to be examined without requiring a full understanding of the non-perturbative step (3). Because of the multistep and complex nature of the predictions, the comparison of data and theory relies heavily on Monte Carlo (MC) techniques. "Events" are generated with different parton configurations, distributed according to the matrix elements of steps (1) and (2). Step (3) is then implemented on an event by event basis using different models. 3.1 The Initial State A beauty of e+e- annihilation physics is that the initial state is governed primarily by QED. This gives the cross section for e+e- -+ Y* -+ ff (massless, pointlike, unit charge, 2 a is the fine structure constant and q is the momentum transferred by the -*. Hence, for e+e- - y*, q is equal to the center of mass energy: q = N. spin I pairs) as aPt ra /q , where = For hadron production the total cross section is: ao(e+e - * - Q2 hadrons) = N, (3 2 pt -+1 32 , (3-1) f where N, = 3 is the number of colors, the sum runs over the active flavors (at PETRA energies, ./ > 14 GeV, f=u,d,s,c,b), Qf is the quark charge (+ for u,c and -. for d,s,b), #2 = 1 - 4m2 /s and mf is the quark mass (taken to be, in GeV: ~ 0.01 for u, d; ~ 0.15 for s; ~ 1.2 for c; and ~ 5.0 for b). These masses are not well determined. They are inferred from the bound states and mesons displaying flavor associated features. The initial state radiative corrections (RC) are given by QED and were included in step (1) to order cO [3-2]. Should a 12 hard photon be emitted, steps (2) and (3) are carried out in the boosted hadronic center-ofmass frame with a decreased \Fs. The detector response to the hadrons and the photon is, of course, simulated in the Lab frame. At PETRA energies the interference of virtual Z 0 's with the annihilation photon is imElectroweak theory ~ [3-3] predicts the total hadronic c to !Mzo- The standard portant, cross section is modified to: ( OEw(e e - Z*,ZO) -+ hadrons) = Ne S [Q2 + 2Qf~wsge gl 2 + X2 s 2 (g + g 2 )(gf 2 + gi 2 )], (3-2) f where Xw = [4sin 2 OW gky = T'3L- 2Qj s sw2 gA TL = TiL = 0 cos 2 Ow(M2 - -26 at ) F = 35 GeV, is the weak vector coupling, is the weak axial coupling, +1/2 for i = c -1/2 for i = d, s, b, e- is the third component of the weak isospin, w is the weak mixing angle, Mzo is the Z 0 mass, and the width of the Z0 and the quark masses have been neglected. The bulk of the modification comes from the interference term, oc Qfgf. In this study variables normalized to the total cross section are used exclusively. For this case the modification to the fractional contribution of the heavy flavors is PL gf /Qf. Within the statistics, this is not observable. 3.2 Perturbative QCD QCD, the theory of the strong interactions, was introduced in Chapter 1. In this section the calculations of the perturbation series expansion of QCD in terms of the strong coupling constant, a,, is outlined up to second order in a, in the form used for the analysis in this study. Many excellent descriptions of these calculations exist (e.g. see [3-1]). The calculation to first order requires the diagrams of Fig 3.2. (The interference of the simple quark pair production diagram of Fig 3.2a and the virtual correction diagrams of Fig 3.2c is first order.) The quarks are taken as massless to facilitate the calculations. For qqg final states evaluation of the gluon bremsstrahlung diagrams of Fig 3.2b leads to the differential cross section [3-4]: where CF = (N2 - , QC2 CF 27r +X q (1 - Xq)(1- 2P ~ -(3-3) _- )' X'" r , d 20 o0 dxq dxq 1 1)/2N, = 4/3 is the Casmir operator from the evaluation of the color SU(3) transformation of the quark-gluon coupling. Eqn (3-3) displays the infrared or soft (Xq and x- -+ 1; Pg - 0) divergence, and the collinear (xq or 13 xq - 1; Pg | or Pg || P1) divergences. It also shows the cross section is low for "Mercedes-Benz" events, where all three partons have equal energy (xq = X- = Xg = 2/3), and are separated by 1200. The soft and collinear singularities are familiar from QED, where a similar divergence arises in the calculation of e+e- - P+/-ry when the -y is not resolved. Evaluation of the diagrams is facilitated via a resolution cut which separates the distinguishable 3-parton events from 3-parton events where the gluon is not resolved. Events where three partons can be distinguished are called "resolvable 3-parton" events. For events from the 3-parton diagrams which fail the resolution cut a pair of partons is recombined, these events are called "quasi-2parton". That is, a "resolvable-n-parton" event is an event where an n-parton configuration has passed the resolution cut, or an (n+ 1)-parton configuration has failed the resolution cut. When the (n + 1)-parton configuration has failed the cut it is also know as a quasi-n-parton configuration. In the calculations the divergences in the quasi-2-parton amplitudes cancel those from the interference between the diagrams in Fig 3.2a and 3.2c [3-5]. Summing the finite remainder and the zeroth order diagram, Fig 3.2a, yields the total "resolvable 2-parton" cross section. In agreement with the "Kinoshita-Lee-Nauenburg" (KLN) theorem [3-6] the resulting resolvable 2- and 3-parton cross sections are finite. Summing them gives the total cross section to first order [3-7]: 01= (1+ 3CF )uo 4 ?r (3-4) This defines the strong coupling constant to be: Swith bo log(s/A 2 )' 11N-2Nf bo 6 - 23(3-5) 6' where Nf = 5 is the number of flavors and where A is the QCD scale parameter. A is the free parameter of the theory. The resolution cut deserves some consideration [3-7a]. In the pyuy case the functional form of the cut depends on the quantity observed, e.g. the muon acollinearity or the photon energy. The cut for qqg is selected similarly. The resolution cut must be less than the resolution for the variable being measured, taking into account the smearing from hadronization. Conversely, the cut must be hard enough to avoid double counting of the soft gluons which are also involved in hadronization. More importantly, to fixed order in the perturbation expansion, too soft a cut can lead to negative probabilities, which is incompatible with the MC technique. In addition, the observables of an event which moves from one category to another as the cut is varied should change smoothly, as the combining of partons in configurations that fail the resolution criteria is more in the nature of bookkeeping than physics. Three reasons to undertake the extension to second order, which involves the diagrams in Fig 3.3, are: (1) a, measured in first order is large, - 0.17 at I = 35 GeV, which necessitates at least next order calculations. (2) The non-Abelian nature of QCD, e.g. the triple gluon vertices in Fig 3.3, only appears in greater or equal to second order diagrams. (3) The relation 14 of the scale parameter A to c, depends on the renormalization scheme. In first order this is poorly specified, e.g. a rescaling of A(-+ A' oc A) yields a constant which can be absorbed or not during the renormalization. Equivalently "N/s" is not well specified, so that neither the same measurements at different center of mass energies nor measurements involving different reactions can be compared. In second order the scale is uniquely defined for e+e- annihilation into hadrons. Here the modified minimal subtraction (MS) renormalization scheme [3-8] is used exclusively, where the scale, i.e. fi, is exactly that of the virtual photon. The total cross section can not yet be evaluated by summing the partial dressed 2-, 3and 4-parton cross sections. Instead, much as in the optical theorem, the relation between the total cross section and the imaginary part of the inverse photon propagator is exploited and the latter is evaluated via the diagrams in Fig 3.4. This yields the total cross section to complete second order [3-9]: 3 a 2)2),o , CF- + Kjs( ?rX 3 Kis - CF(0.538Nc - - CF - 0.08645Nf) = 1.405, a2 = with (1 + 4 (3-6) 32 where the coupling strength is given by: 2x' as = 2r(3-7) S bo log(s/A 2 ) + (bi/bo) - log(log(s/A 2 )) wit with b1 bi - The presence of both the log( 17N2 - 5NeNf wv 6 - 3CFNf _ 68 6 ) and the log(log() terms indicates the unique relation between a, and A in second order. The full contribution of the interference of the fourth order, 2-parton diagrams, Fig 3.3c, with the zeroth order diagram, Fig 3.2a, is not yet finished [3-10]. Instead the resolvable 2-parton cross section, finite by the KLN theorem, is taken as the total cross section less the resolvable 3- and 4-parton cross sections. The 4-parton tree level diagrams, Fig 3.3a, are evaluated above a suitable resolution cut yielding the resolvable 4-parton cross section [3-11]. As in first order the assorted divergences below the resolution cut have been shown to cancel with interference terms [3-12a]. The quasi-3-parton cross section has been calculated by several groups and the results have not agreed. The earliest calculation [3-12a] was precise but difficult for experimentalists to apply because the results were expressed in terms of a single variable, thrust (this variable is defined in 4.1). This variable, it turned out, is very dependent on the resolution cut and the calculation was done with a cut orders of magnitude smaller than that dictated by the requirement of non-negative cross sections and the desire to avoid double counting soft gluons. A subsequent attempt [3-12b] obtained an analytic expression in a general form with reasonable resolution cuts, but achieved this by approximating a few terms and by an approximate scheme of parton recombination. These approximations turned out to be not a 15 good idea and the calculation was found to be accurate only to the order of the resolution cut. These two calculations were eventually shown to be equivalent in the limit of vanishing resolution cut [3-13]. This study uses a third calculation [3-12c], where the precise methods of [3-12a] were extended to relevant values of the resolution cut by a MC integration of the 4-body phase space, and the results were expressed in terms of the x qand x- defined in Eqn (3-3), which completely specify a 3-parton event. The accuracy of this calculation is limited by the Monte Carlo statistics. Ten million events were used and the theoretical uncertainties exceed the statistical error. Work continues on this topic, giving some hope of a useful analytic evaluation [3-12d]. The calculation of [3-12c] implemented two different forms of resolution cuts over a range of values, the scaled pseudo-invariant mass cut, defined by Y = 2 minPP and the Sterman-Weinberg cuts C = min i S (>Z Pi) where [3-7a], P-P Ecut - P' 2 (Zs P ) i~ ' > Ycut, (3-8) defined by: and cos(26) #- -PN max i+1 PPi < cos(26cut), (3-9) P, P are the momentum vector and magnitudes of the partons i, j,k. These momentum based definitions allow the QCD calculations for massless partons to be extended to massive quarks. The Sterman-Weinberg cuts have simple interpretations: if a parton has too low an energy it cannot be resolved and if two partons are too close together they cannot be distinguished from one parton. The invariant mass cut achieves the same effect by excluding large parton energies but can be calculationally more convenient. For example in the qqg case the cut excludes large parton momenta: Y > Ycut -- > xqq,g < 1 - Ycut, where x is from Eqn (3-3). To give a concrete example MC events can be placed in one of four categories, labeled 2, 3, 3' and 4. "2" refers to all events which contain 2 resolved partons. "3" refers to the first order resolvable 3-parton events, i.e. those events from Fig 3.2b which pass the resolution cut. "3'" refers to the second order resolvable 3-parton events. Contributions to this category come from: the finite terms in the evaluation of the diagrams of Fig 3.3a which fail the resolution cut once and the interference between the third order virtual correction diagrams of Fig 3.3b with the simple first order single gluon bremsstrahlung diagrams of Fig 3.2b. "4" refers to the second order resolvable 4-parton events, i.e. configurations passing the cut from the tree-level diagrams, Fig 3.3a. At Fi = 35 GeV for A = 0.10 GeV(=> a, = 0.12), with ecut = 0.1 and cos(2Scut) = 0.9 the fractional partial cross sections are calculated to be: 1 - (F 3 + Fs' + F4), )/Utot = 0.57 F2 = u (resolvable 2-parton F = u(resolvable 3-parton, O(a))/uott = 0.31 F-' = u(resolvable 3-parton, O(a2))/otot = 0.06, 16 = o-(resolvable 4-parton, 0(a))/utot = 0.06 . F For the analysis presented below F' and F 4 were calculated from the generation, including radiative corrections, of one million events of both parton types at each center of mass energy. The resulting statistical error is negligible (- 3 parts in 104). 3.3 Hadronization Because the evolution of partons into hadrons involves the non-perturbative or "confine- ment" regime of QCD and is not yet calculable, the comparison of data and theory requires the use of fragmentation models. While these models are partially motivated by theory, they are not useful if they cannot be adjusted to agree with the data. Two models are used in this study, the model due to Ali et al., an independent jet model [3-14], which is an extension of the earlier Feynman-Field qq fragmentation parameterization [3-15], and the Lund color string model [3-16]. Of the models which can be "tuned" to reproduce the data, these two represent extreme viewpoints with respect to the effects of the initial parton topology on the direction and energy of the hadrons produced. The model dependence of the results is taken as a conservative estimate of the uncertainty owing to the ignorance of the details of fragmentation. The models are well described in the references given (see also [3-17]), but a few points need to be reiterated. Each model produces hadrons by popping a series of sea quark pairs out of the vacuum. Several free parameters describe the generation of these pairs, these parameters are fixed using inclusive and exclusive particle spectra [3-18]. Of these parameters, the mean transverse momentum, o-q, has the largest effect on the resulting event shape. The Ali model fragments quarks one at a time into hadrons through a chain of decays: quark -+ meson + sea quark, where the meson contains the initial quark and the sea antiquark. Gluons are treated as a collinear quark pair, each of which is fragmented in turn, so gluons produce a broader jet (i.e. two overlapping jets). A problem with this approach is that the production of massive jets from massless partons explicitly fails to conserve energy and momentum. This is fixed post facto by a suitable Lorentz boost of the jets of hadrons. which introduces a mild global dependence of the final jet directions on the initial parton configuration. In a qqg event this results in the boost usually being directed antiparallel to the gluon direction, which moves quark jets off of their original parton directions and towards the gluon jet. Hence the event appears somewhat more like a 2-jet event. Another problem with the model is its discontinuous behavior with parton pairs just passing or just failing the resolution cut. Because the number of links in the fragmentation chain varies about as the square root of the initial parton energy, a (qg) subsystem just passing the resolution cut produces about twice as many hadrons as when it fails the resolution cut and is recombined into a single quark. The Lund model fragments strings, which are narrow tube-like fluxes of the color field, that run between the partons. This is a Lorentz invariant approach a n d avoids discontinuities 17 across the resolution cut. Formally the Lund model does not produce "quark" jets or "gluon" jets and the intuitive association of jets with partons present in the Ali model is lost. The fragmentation of a qqg event is depicted for the two schemes in Fig 3.5, the Lund model producing particles along hyperbolas connecting the quark and gluon and connecting the gluon and antiquark. Functionally, the Lund model produces jets shifted from the original parton directions, the "quark" jets swinging towards the "gluon" jet. This is a larger effect than the imposition of energy-momentum conservation via a Lorentz boost in the Ali model. Consequently a 3-parton event appears less 3-jet like when fragmented with the Lund model than with the Ali model. This implies that to predict a measured number of 3-jet events the Lund model requires a systematically larger partial cross section for 3-parton events than the Ali model, that is, a larger a.. The magnitude of this shift is seen to be 14% in chapter 4, it would be nice to eliminate this shift beforehand by excluding one of the models. The MARK-J data does not rule out either model. Investigations which have claimed to favor the Lund model have been based on the soft particle fluxes between jets (e.g. see [3-19]), a region neither model was designed to simulate. There are several other models proposed to mimic fragmentation. Hoyer independent jet model, These include the [3-201, which, in the usual implementation, achieves energy- momentum conservation by rescaling the hadron energies while maintaining their directions, and numerous "gluon-shower-cascade" models [3-21]. To date they do not yield satisfactory descriptions of our data and remain under development both at MARK-J and elsewhere. 3.4 Detector Simulation After event selection the analysis of hadronic events uses primarily the energy vectors, Ej, of the hits in the various calorimeter layers. For this reason a complete detector simulation is more important for MARK-J than for detectors which measure, for example, the momentum of charged tracks. The bulk of this study uses a fast, but complete, simulation. Particles are tracked through the detector and their intersections with active detector elements calculated. The energy deposited in each counter is determined from tables that give the dependence on penetration depth, angle, incident energy, and particle. Energy resolution and longitudinal shower fluctuations are also simulated using tabulated information. These tables were generated using test beam data for 0.5 to 10 GeV electrons and pions, experimental calorimeter studies [3-22], and a shower Monte Carlo program [3-23], but have been much improved by studies using the data collected at PETRA from multihadron events and the very collinear hadron jets from r decays. After correction for time of flight, attenuation and propagation delays in the scintillator, time slewing due to varying pulse heights. and multiple hits, the counter ADC and TDC information is digitized. The drift chambers are also simulated in detail, and the result is written out and analyzed using the offline programs, including the same cuts, described in 2.2. 18 As a cross check another, more accurate, detector simulation called GHEISHA is used [324]. This program carefully follows not only each particle but also its associated secondaries through the complete detector. Multiple and nuclear scattering are included, as are many other effects. Again, the output is digitized and analyzed with the programs used for data. For technical reasons, only a fraction of the events simulated with the fast shower program are available. The results of the simulations are compared with the data in Fig 2.4-2.7, 2.10 and 2.11. Both simulations reproduce the response of the inner and outer calorimeters reasonably well. The effect of the inconsistencies, e.g., the shift in the inner calorimeter energy distribution seen in Fig 2.4b, are investigated in the analysis presented below. 19 Chapter 4 4.0 Determination of the Strong Coupling Strength as a Function of Center of Mass Energy Methods in Measuring Gluon Bremsstrahlung The fundamental process of QCD is the interaction of gluons. The strength of this interaction is given by the strong coupling constant, a.. This chapter presents the determination of a. over a wide energy range using the rate of hard gluon bremsstrahlung. The presence of a single radiated gluon can be tagged because it modifies the topology of the event. Qualitatively, a two parton event is observed as two narrow back-to-back "quark" jets. An event with a gluon emitted at low energy or small angle to one of the quarks appears as a narrow "quark" jet opposed by a broader "quark + gluon" jet. Events with a higher energy gluon emitted at larger angles show three distinct coplanar jets. Four parton events appear in a variety of topologies: two broad jets, a very broad jet opposing a narrow jet, and more isotropic configurations. Hence three parton, qqg events are identified by analysis of the spatial distributions of observed hadron momentum vectors. Intrinsic to QCD is the variation of the strong coupling with energy, accounting for confinement at low energies and asymptotic freedom at high energies. Eqn (3-7) relates the coupling constant, a,, at a given center of mass energy, fi, to the QCD scale parameter, A. The perturbation expansion leading to this equation is expected to be valid for 5> A, and well above quark flavor production thresholds, \s > 2Mf. Measurements of a, over the PETRA energy range which meet these conditions (i.e., excluding the data \ < 14GeV ~ 2Mb) can then be related, provided a renormalization scheme is specified and a full second order calculation is used. This study uses the MS renormalization scheme and such a calculation, as related in 3.2. In the MS scheme, for e+e- -- hadrons, the scale of the reaction is unambiguously defined to be the energy of the annihilation photon, so that the rate of gluon bremsstrahlung yields a, immediately, i.e. a, is the physical quantity and A is the theoretical parameter. For this reason the determinations of the strong coupling strength in 4.1, 4.2 and 4.3 which combine several center of mass energies are given in terms of a, evaluated at an V'8, with the expected variation of a, with F from Eqn (3-7) taken into account. For this study the mean \/F of the high statistics data 33 < \Is < 36 GeV is taken as \F: o = 34.63 GeV, and the conversion to AT is delayed to 4.4. The primary difficulty in determining the strong coupling strength is to disentangle the effects of hadronization from the phenomena described by perturbative QCD. This is where the use of a wide range of center of mass energies becomes advantageous. The perturbative effects have a slight, logarithmic, dependence over the energy range: for A = 100 MeV , a,(\/, = 22GeV) = 0.154 and a,(x/5 = 46GeV) = 0.118. The fragmentation effects are expected to 1/ 8 b, where the value of b ranges from } to 2 depending vary much more quickly, roughly on the variable considered [4-1]. Hence, using variables which show only a small, ~ logarithmic, dependence on 5F to determine the strong coupling strength has the advantage that 20 the uncertainty due to the incalculable power corrections from non-perturbative effects are minimized. A wealth of variables exist to quantify the observed topologies of hadronic events. Their suitability for determining a, can be examined with respect to the related criteria of: insensitivity to the details of fragmentation; infrared stability; independence from the parton resolution cut; and small higher order corrections. The particle multiplicity of a jet is an examplary detail of fragmentation. Variables which depend quadratically or more on the particle momenta or energy are sensitive to the multiplicity. These variables are unstable against particle decays, for example p -+ 7rxr and against the emission of soft or collinear gluons. In addition there is no manner to connect such variables with parton level calculations. Parton resolution cuts cannot then be related to the experimental resolution and their functional form cannot be chosen to minimize their effect on the measurement. Variables which depend linearly on the particle momenta avoid these problems and are called "infrared safe". The higher order (0 (a 3 )) calculations do not exist and can be roughly estimated in two fashions: the relative size of the first and second order contributions, and the dependence on the parton resolution cut. The reasoning behind the former guess is clear, the latter guess is motivated by the parton level calculations of these variables, where the expansion in powers of the resolution cut parallels the perturbation expansion. Stability as the resolution cut varies indicates that next order corrections may be stable, stability as the resolution cut vanishes indicates some hope of small higher order corrections. In this study two additional criteria need to be imposed: adaptability to the calorimetric measurements of the MARK-J detector and good discrimination of "3-jet-like" events as opposed to the easier discrimination of "not-2-jet-like" events. The MARK-J calorimetry does not measure the individual particle momenta in a jet, but a series of energy depositions in the different shower counter layers, resulting from the interaction of an undetermined number of charged and neutral particles. This dovetails with independence from the exact particle multiplicity and imposes no further requirements. The discrimination of 3-jet events fits in equally well with the criteria of small higher order corrections, because the next order is expected to have a larger relative effect on the > 4-parton distributions than on the 3-parton distributions. Assuming a suitable variable has been chosen, the point is to compare the prediction, which depends on a,, with the data. To do this for some variable, call it z, the prediction is given by: XThy(a)= ZzAF(as)/ AF(es) (4-1) where the sum runs over the parton types defined at the end of 3.2, i= 2,3,3',4. A' is the acceptance per parton type, and xi is the value of the variable per part on type evaluated using accepted MC events. The denominator on the right side of Eqn (4 1) is the acceptance, as Ai is 5 for all parton types, the acceptance is practically independent of ci. The dependence 21 of XThy on a, enters solely through the fractional partial cross sections, Fi, and the quality of a variable in picking out single gluon bremsstrahlung is then the difference between x 3 , X' and x 2 , x 4 . The coupling strength is determined by the condition XThy (a,) - extension to obtaining the prediction for a distribution is straight forward. xData = 0. The To determine ci (Vs-o) from measurements at several energies, Eqn (3-7) is used to evaluate a, at each v's. The determination of o, using three suitable but different variables is presented in 4.1, 4.2 and 4.3. The three methods are compared in 4.4. The "running" of the strong coupling "constant" is discussed in 4.5, non-perturbative effects are quantitatively examined in chapter 5. 4.1 a, from the Energy Flow Variables In this section the Energy Flow variables [4-2] are used to quantify event topologies, resulting in a measure of the rate of single gluon bremsstrahlung. Having this rate, the strong coupling strength is determined. As depicted in Fig 4.1, the variables are determined stepwise for each event. First, the thrust axis, , and thrust value, T, are found by maximizing T while varying t: i ZE, = [0.5,1] (4-2a) , Thrust = T = max t is also known as the event axis. Next, the major axis, mh, and value, M, are found by maximizing M while varying m, keeping r^ I t: - = 0,0.5] , Major = M = max l _ t, (4-2b) (t, rA) defines the event plane. The normal to this plane is the minor axis, n, the projected energy flow along n^ is known as minor, m: for planar events minor = m = E ini -Al = [0, M) , n^ = t X M^, (4-2c) minor is very close to the minimum projected energy flow. At the parton level 2-parton events have T = 1, M = 0, m = 0. A 3-parton event has T = 1 - Y, M = 2 pguon/,F/, m = 0, where pflun is the gluon momentum perpendicular to t and Y is from Eqn (3-8). Consequently the "Y-cuts" are used with the Energy Flow variables. A "Mercedes-Benz" configuration, 3 equal energy partons separated by 1200, has T = 2/3, M = 1/-V, m = 0. Fragmentation smears these directions. Including the detector resolution, MC studies indicate a resolution of 5* for the event axis determined in qq events, and 100 for the event plane (i.e. A) determined in qqg events. The values are also smeared, but the coplanar events from gluon emission still have a significantly larger Major than minor, i.e. they are oblate [4-3]: Oblateness = 0 = Major - minor 22 (4-2d) Again at the parton level, O(2-parton) = 0 and O(3-parton)= M = F. An advan- 2Plu"n/ tage of Oblateness over Major and Thrust is the subtraction which effects a partial cancellation of the hadronization smearing. Other advantages of Oblateness over Thrust as a variable to discern single gluon emission are that 0 specifically tags these events whereas T yields the "2-jettiness" of an event and that 0 is much less sensitive to the parton resolution cut [3-12c]. More detail is obtained by dividing the event into hemispheres using the (tm, n) plane. For q-qg events the hemisphere with less energy flow perpendicular to C, the narrow (N) side, usually (70% at \/F = 35 GeV) contains the "quark" jet. The opposite, or broad (B), side contains either the "quark + gluon" jet or two distinct jets. Recalculating the variables on each side yields TN, TB, MN,...,ON, OBThe data and MC predictions using both models with a.(fi 0 ) = 0.144 are plotted versus -/F for several of the Energy Flow variables in Fig 4.2a-e. Reassuringly, the agreement is good. The dot-dashed curves are the predictions for qq events alone. For the mean values (1 - T), (M) and (m), Fig 4.2a-c, the q-q effect, which is purely from fragmentation, is large. These are not good 3-jet measures. Better separation comes using the fraction of events with 3-jet like topologies. The fraction of events with large 0 ., Fig 4.2e, is the preferred measure of single gluon emission, having a small contribution from fragmentation and only a slight dependence. VS Qualitatively, requiring a large value of OB selects events with energy flow confined in a half plane on the broad side. Using this variable the strong coupling is determined. For example at 35 GeV with Yet = .04 QCD predicts (using the Ali model): X2 = (fraction of accepted 2-parton events ( ( 3-parton events ,0q(a), " ) = 0.25, 3-parton events ,Oi(ai), " =( 4-parton events ,O(ce.), " )= )= S= X X = 4 with 0B > 0.3) = 0.04, 0.24, 0.35. Folding in the acceptances and fractional partial cross sections and comparing these fractions to the fraction measured, XData = 0.13, & la Eqn (4-1), determines the strong coupling constant: a. = 0.14. This implies 76% of the events with 0B > 0.3 are from first and second order qqg events. Of course this percentage depends on the Ycut used to define qqg events. For each of the five center of mass energy points in Table 2.1 ten to forty thousand MC events were generated, fragmented, and put through the complete detector simulation for each of the fragmentation models. The strong coupling constant determined at each fi is plotted in Fig 4.3 (see also Table 4.1). Also plotted are the results from simultaneously fitting all the data, (22 GeV < Vs 47 GeV), which yielded a,(fO;OB) = 0.134 0.012(syst.) using the Ali Model and o,(xo; OB) = 0.154 the Lund Model. 23 0.004 (stat.) 0.004 (stat.) 0.011(syst.) using The systematic error estimation is summarized in Table 4.1, the following sources were considered: (1) Parton level resolution cut in the QCD calculation: This was varied from Ycut = 0.03 to 0.05. At 35 GeV this results in a variation F3 from 0.37 to 0.23 and of F3' from 0.14 to 0.09 but the variation in a, determined at each of the five energies and over the full range is less than a 5%. The measurement is stable over the experimentally relevant region. However, from [3-12c] the parton level calculation indicates a&,(Ycut -+ 0) ~ 20%. (2) 0 B 0 cut: Fig 4.4a shows the slight dependence of the measurement on the exact value. The ordinate is a,(,F0 ; 0), fitting the data 22 < fv . cut i.e. the strong coupling strength determined from < 47 GeV. (3) Acceptance cuts: The cuts on EV, E1 and E, which define the sample of accepted events (see 2.2.2 and Fig 2.10, 2.12) were varied. In addition, cuts on the maximum fractional Ey and on cos 0j, the angle between the event axis and the beam axis, were imposed. The motivation is two-fold: First, background events such as 2-photon hadron events or tau pairs decaying into hadrons are characteristically more imbalanced and have either too little visible energy (electrons down the beampipe or neutrinos) or too much (the energy of an electro-magnetic shower is over estimated when evaluated as a hadronic shower) as related in 2.2.1. Second, an event topology would be misconstrued if a large fraction of the event vanished into one of the holes in the detector response. The largest holes are around the beampipe ( 2.1), and the cut on cos 0j ensures the event is well into the active solid angle. The eight sets of cuts examined and the resulting estimated acceptances are given in Table 4.2. The acceptance varies by a factor of two but the result is stable for each of the five energies and, as shown in Fig 4.5, for the fit over the entire V/- range. At the two points with higher statistics the following sources could also examined: (4) Oq: Fig 4.6 demonstrates the stability of the determination over a wide range of The leftmost point was determined using the Ali model with Uq Uq values. ~ 0, and gives a feeling for the size of the corrections due to fragmentation. As mentioned, [3-18], the uncertainty on Uq is ; 12%, i.e. 270 < oq(Ali) < 330MeV and 380 < Uq(Lund) < 470MeV. (5) Detector simulation: The fast detector simulation used in this study is most suspect in its treatment of low energy particles. To examine this cuts on the minimum energy deposition per counter were imposed on the data and MC. Fig 4.7a shows the measured and predicted variation of EV from these cuts, and Fig 4.7b shows the resulting variation in the a 8 determination, which is minimal. Similarly, using only the information from the inner calorimetry causes at most slight shifts in the results (marked "ABC only" in Fig 4.7), as does use of the GHEISHA detector simulation (marked "GHEISHA" in Fig 4.7). The largest source of systematic uncertainty comes from the difference between the results for the two fragmentation models. Averaging the two results in terms of a8 24 (fin) and taking the difference into the systematic error, the strong coupling is determined using the fraction of large OB events to be ct,(V'o; 4.2 O) 0.004 (stat.) = 0.144 0.015 (syst.). a. from the Jet Multiplicity In this section the number of partons and hence the rate of gluon emission is deduced from the number of observed jets. An "observed" jet is a high density of energy flow surrounded by a region of lower energy flow, such that the width of the jet is less than the angle between jets. Experimentally this is quantified with the use of a cluster algorithm. Many such algorithms exist [4-4], here one designed for the MARK-J calorimeters is used. In this algorithm the largest energy hit is taken as the seed for the first cluster. within an angle jet/2 of the seed vector are added into the cluster: Hits the cluster direction equalling the vector sum of the Ei and the cluster energy being taken as the algebraic sum of Ej I. The seed for the second cluster is taken as the largest energy hit not added into the first cluster, and hits within Sj.t/2 of it are added to it, and so on, until all hits have been added into clusters. Depending on bjet this yields an average of 10-30 clusters/event. Clusters with an energy less the ejet - EV are then "dissolved", their constituent hits added hit by hit to the nearest (in angle) cluster. Finally, for cluster pairs separated by an angle less than 2 3j,,t, the lower energy cluster is dissolved. The jet multiplicity is taken as the number of surviving clusters. The (fejt,jet) used in the algorithm correspond to the Sterman-Weinberg (Ccut ,Sut) parton resolution cuts used in the QCD calculations, Eqn (3-9). Counting the number of observed 3-jet events is a straight forward attempt to measure the rate of hard gluon emission, which has advantages and disadvantages. On the positive side, the intuitive interpretation of observing an event with 3-jets is much cleaner than, e.g., observing an event with 0 B =0 .35. On the negative side, the resolution criteria are tighter. For a q-qg event where 0. can tag a gluon jet partially merged into a quark jet (the 'broader "quark + gluon" jet' of 4.0) a cluster algorithm has trouble. Using a sufficiently loose (Ejet,Sjet) to identify 3 jets in such an event leads to jet multiplicities of 5 or 6 in other events, which are difficult to interpret. Related to this is the difficulty in estimating the effects of the parton level resolution cuts. With these provisos in mind the method is applied. Fig 4.8a-c show the measured jet multiplicities and the MC predictions as a function of center of mass energy for a typical value of (fjetet). Again using the Ali model at \/s = 35 GeV as an example, Table 4.3a gives the fractional jet multiplicities for each event type. Comparing the rate of 3-jet events between data and MC determines the strong coupling to be a, = 0.12. Table 4.3b gives the resulting predicted fractional contribution of each parton type to each jet multiplicity. A similar number of MC events as in 4.1 were generated and the strong coupling strength was determined by comparing the predicted and measured rate of 3-jet events. The results are given in Fig 4.9 and Table 4.4. For the simultaneous fit to the entire 25 fV range the values a(N/s; JM) = 0.123 cs(Fo; JM) = 0.003 (stat.) 0.143 0.035 (syst.) when using the Ali Model and 0.003 (stat.) 0.035 (syst.) when using the Lund Model gave the best description of the data. The systematic uncertainties, summarized in Table 4.4, were estimated as for the rate of large 0B determination, but the interplay of the (E, 6) cuts on the parton and experimental levels is much stronger than the corresponding relationship between Ycut and the cut on 0 .. Fig 4.10 shows the relationship between the a, value determined and the fraction of 3-jet events predicted to have come from qzg initial states for a range of (EjetSjet) and (Ecut,bcut). The points which are plotted as crosses yield a poor description of the rate of 2- and 4-jet events. The errant points have either a very tight (fiet ,jet) compared to (Ecut, cut) ® (fragmentation smearing) or a too loose an (Ejettjet). The errors quoted in Table 4.4 do not include these points. Fig 4.11 shows the stability of the result with respect to different criteria used to define the event sample. Fig 4.12 shows the variation of the result with Uq. Fig 4.13 shows the disappointing sensitivity of the method to a minimum energy per counter cut. The observed jet multiplicity decreases rapidly with increasing min(Ei) and indicates this method is dependent on the details of the detector response. This is despite Eqet being taken as a fraction of the energy used. Combining this with the systematic error from the difference between the two fragmentation models the QCD scale parameter is determined to be a,(A/so; JM) = 0.133 4.3 0.003 (stat.) 0.037 (syst.) using the observed rate of 3-jet events. a. from the Energy-Energy Correlation Asymnetry In this section the modification of event shape due to gluon emission is extracted from the correlations between the energy flow in different parts of the event. This is achieved with the Energy-Energy Correlation (EEC) function, Z1, first introduced by Basham, Brown, Ellis, and Love [4-5]. Experimentally, the pairwise products of fractional energy deposition, Ej Ej/E' , are plotted versus the cosine of the angle between them, cos Xij, for all events: 1 d~c (cos X) 01 dcos xdcs a X 1for XN Acos V Nevents i~j.A Ey2 Acos~ Icos X - cos Xij < -Acos X) . (4-3) The intuition is best served by considering the parton level: A 2-parton event has entries only at cos X = 1. A 3-parton event has a reduced entry at cos X = + 1 (the self correlation term), no entry at cos X = -1, and three entries spread across -1 < cos x < +1. A 4-parton event has a further reduced entry at cos X = +1 and six entries spread over -1 < cos X < +1. Both 3- and 4-parton events have an entry at cos X = cos 26, where cos 26 was used in the definition of the Sterman-Weinberg parton resolution cuts, Eqn (3-9). For this reason the Sterman-Weinberg cuts are used with the EEC and the EECA, defined below. The utility of the measure comes from the fact that qqg events contribute more at negative cos X than at positive cos X, while qq events contribute only at cos X = 26 1 and 4-parton events, being more isotropic than qqg events, contribute more symmetrically about cos X = 0. This is enhanced by forming the EEC Asymmetry (EECA), A, defined as: A(cos X) = 1 [d~c(cos(,r - X)) 01 d COS (r - X) cs7-X cs~ --1<5cos X50. (4-4) d.c(cos(X)) 1 , d cos( X)] _ Still at the parton level, qq events are zero for the entire distribution, while 3- and 4- parton (the self correlation term) and cos X = - cos 26. events have negative entries at cos X = -1 Hadronization distorts the parton level considerations. The back-to-back jets from a q-q event bleed inward from cosX = 1 in ZC, but this smearing is symmetric and cancels in A. The entries from a qqg event are likewise smeared, but this smearing is again symmetric. Averaging over events the EECA still bears the imprint of the underlying QCD matrix elements. Specifically, the region away from cos X = -1 is populated by 3-jet events. Therefore, to determine the strong coupling strength, the distribution is examined for cos Xo < cos X 5 0, where the functional dependence is given by the QCD matrix element and the area is ~ proportional to the rate of hard gluon bremsstrahlung, i.e., to a.. If - cos 26cut < cos X0 , this has the added advantage of excluding the region of the parton level resolution cut, where the perturbative calculations are least sure (this is further examined in the next chapter). Indeed, parton level calculations indicate that the EECA distribution away from cos X = -1 is stable with respect to the parton level resolution cuts (Cc ut), even in the limit of very loose cuts [3-12c], [4-6]. Another advantage of the EECA compared to the methods of 4.1 and 4.2, is that it requires no complicated algorithm to pick out event or jet axes (these algorithms can become confused by, for example, holes in the detector response). Looking at the definition of the EEC, Eqn (4-3), and remembering that hadrons enter the detector packed together in jets, a positive bin-to-bin correlation is expected. Considering an event, a large entry in one bin is accompanied by entries in the neighboring bins. This correlation carries through to the EECA. These correlations are accounted for as described below. The measured and predicted EEC and EECA distributions at the five center of mass energies are shown in Fig 4.14 and 4.15. Note that the MC histograms were generated with a single value for a,(v%), adjusted according to Eqn (3-7), and this is not the "best fit" value for the EEC(V ), Fig 4.14. The EECA falls quickly with cos X (note the logarithmic scales). Fitting the integrated EECA, A,(cosXo) = o A dcosX would reflect primarily the contents of the bins near cos Xo. Instead the strong coupling strength is determined using the EECA by minimizing the following X 2 : X 2 (a) = (Alat - AM 0 ) (ata _ AMC(&,)) k,L where the prediction for A in cos X bin k is, from Eqn (4-1), AMC(c,) = Z A'AF(a.)/acceptance % a,('s) see Eqn (5-1), and where the covariance matrix V is defined from the EEC as: 27 (4-5) +[same ~L~)1M k + i MC at a 2()D 2(E) ,(pc o 0_Lc)_ Vk k cos Xj =cos(7r -Xk) Z or kevent events Acos X . . ev1eNt1 Zevent\ ZkN IN (z N WE) (event events 2 E2 I N events for ICOSXk - COsXij < IAcos X V accounts fully for the bin to bin correlations. Fig 4.16 and Table 4.5 give the results of the fits at each energy. Fig 4.17, the contribution to the X 2 versus (k, 1) for one of the fits, shows the importance of the off-diagonal terms in V. To determine a. (,/59) the X 2 was summed over the five energy points. The best fits (one parameter, 88 degrees of freedom) were obtained at: a8 (/Fs 0 ; A) = 0.108 and ci(vF/o; A) = 0.121 0.005 (stat.) 0.006 (stat.) 0.010 (syst.) with a X 2 /d.f. = 1.54 using the Ali Model 0.010 (syst.) with a X 2 /d.f. = 1.50 using the Lund Model. These X 2 /d.f. are discussed below. The systematic errors were estimated as in 4.1 and 4.2. Particular attention was paid to the variation of the measurement and the chi-square with cos Xo. Fig 4.18a shows the X 2 /d.f. for different (Ecut, cos Xo) and Fig 4.18b shows the resulting variation in the determination of cs (VFo). For cuts away from the 2-jet region, cos X Z -0.9, both models show no variation with COS Xo. The results of the Lund model is also stable with respect to variation of the parton level resolution cut, Ecut. The results of the Ali model display a ~ linear dependence on ccut. As mentioned in 3.3, this discontinuous response of the model when varying the parton level resolution cut results from the anzatz of independent fragmentation. This deficiency appears despite the use of the EECA, which is relatively free from fragmentation effects. Because, within reasonable limits as explained in 3.2, no particular Ecut is preferable, this dependence is taken into the systematic error for the model. Other estimations of systematic uncertainty were made as detailed in 4.1, the attendant plots are presented in Fig 4.19 (Acceptance cuts), Fig 4.20 (O-q) and Fig 4.21 (min E1/counter). The determination is quite stable against variation of all three. The main single source of systematic error comes from the discrepancy between the two fragmentation models. Taking the simple average of the results from the two models, and increasing the systematic error to include the discrepancy, the strong coupling constant evaluated at \/% = 34.6 GeV is 0.018 (syst.). determined using the EECA to be: c,(\5 0 ; A) = 0.114 0.005 (stat .) 28 4.4 Comparison In the preceeding sections determinations of the strong coupling constant, a,, as a func- tion of center of mass energy have been presented using three very different methods. The variation of the coupling strength with energy, as predicted by perturbative QCD to complete second order, has been used to simultaneously fit the data at all energies. Before directly addressing the running of the coupling constant and whether the dependence is that given by Eqn (3-7), a set of a, values must be selected. Therefore, the relative merits of the three methods will be discussed. Comparing the results given in Tables 4.1, 4.4 and 4.5, the three methods show a systematic offset between the results when using the Ali and Lund models, the with Lund model are ~ 14% higher in a values obtained a, than those determined with the Ali model. Because the two models represent extreme viewpoints on how fragmentation proceeds, the results are averaged and the offset taken into the error. More difficult to estimate, especially as a function of energy, are method related systematics. Because the three methods used are well defined at the level of observation and at the level of partons, it is reasonable to look for systematic variations at the parton level. These include the effects of omitting higher order terms and imposing resolution criteria in the perturbative calculation. The method of jet multiplicity has a large systematic uncertainty associated with the resolution criteria (e.g. see Fig 4.10). The number of jets is a discreet measure so that parton configurations on the resolution boundary make a discontinuous contribution as the criteria are varied. This could be avoided by setting the algorithm parameters well above (parton level resolution)((fragmentation)((detection). From Eqn (3-3) hard 3-jet events are rare, so the result of such a hard criteria is that the number of three jets tends to statistical insignificance. In addition the method is dependent on low energy hits as shown by comparison with the MC calculation (Fig 4.13). The utility of the method was that it is a straight forward attempt to tag gluon bremsstrahlung. In conclusion, it qualitatively demonstrates QCD, but is not sufficiently sensitive. Use of the Energy Flow variables, specifically the rate of events with large 0 B, yields results which are stable with respect to a variation of the parton level resolution criteria over the experimentally relevant region. In particular the results were found to be insensitive to finite size Yeat, as given in Table 4.1. However, an estimate for the limit Ycut --+ 0 showed a change in the relative rate of large 0 . events leading to an order of magnitude 20% reduction in a,. This could indicate a sensitivity to higher order contributions when the cut is set to small values, and these have not been calculated. The third method does not rely on the specific event topology. The results from fit- ting the EEC-Asymmetry function have been shown to be relatively free of detector related 29 systematic uncertainties and from fragmentation effects. Additionally, the parton level prediction agrees well with the observed distribution and the effect of imposing the parton level resolution criteria is less than 2% on the parton level (again, see [3-12c],[4-6]), although the implementation of the method using the Ali model showed a larger variation. So this method is free of the problems found in the other two methods. Returning to Table 4.5, the chi-square per degree of freedom is ~ 1.5 for the fits using both the Ali model and the Lund model and at both of the high statistics points, (0) = 35, 44 GeV. Comparing the data taken at the same - but at different times ( 1 year) yields a X 2 /d.f at or below unity. The fit residuals versus cos X and versus -,F do not indicate any systematic trend. To account for the high X 2 the quoted statistical errors in Table 4.5 have been increased by 25%( = [(X2/d.f.)]I - 1). Because the use of either model yields fits of equal quality, the a, values are averaged. An independent analysis of the data and MC, internal to MARK-J but using separate calibration, selection and reconstruction procedures to those presented in 2.2.2, obtained a, values that were consistent with those presented in Table 4.5. The two sets of results are compared for the different energies in Table 4.6. For each model at each energy the statistical error on the averaged result is taken to be the larger of the two errors. The difference is accounted for in the systematic error. The combined result from the two models,the bottom row in Table 4.6, is taken as the most accurate and precise determination of the strong coupling constant at each of the five center of mass energies. The error has been increased to include the uncertainty due to fragmentation as reflected by the two models. A similar comparison can be made for the results obtained from fitting the MC predictions to the data over the entire energy range, with the running of the coupling constant given by Eqn (3-7). The results from the independent analysis are consistent (for the Ali model ci,(W 0) = 0.108 0.005 from Table 4.5 compared with 0.114 analysis; for the Lund model a,(Vs-) = 0.121 0.004 from the independent 0.006 compared with 0.131 0.006). Averaging these values yields the strong coupling strength to complete second order in QCD: ci(,(F= 34.6 GeV) = 0.119 0.005 (stat.) t 0.020 (syst.) where the uncertainty from the fragmentation models has been placed in the systematic error along with 0.008 to account for the difference of the two analyses. Evaluating this result for the QCD scale parameter yields: A = 78 i2 (stat.) i12 (syst.) MeV from Eqn (3-7), that is, in the MS renormalization scheme. This value of the scale parameter in agreement with the earlier analysis of the MARK-J data [4-7], which also used the second order calculation outlined in y3.2. 30 In the three years that second order calculations have been available, many other groups have presented determinations of the strong coupling constant using multihadron data from e+e- annihilation and several topological measures, including the EECA. Unfortunately for the field, the handiest second order calculation was that of [3-12b]. As mentioned, this calculation was too approximate and the a, values reported have been shown to be 20-30% too large. This has been understood and the increased values are reproducible using their method. However, since this was incorrect, it precludes showing these determinations for comparison [4-8]. Recently, determinations of the scale parameter using an accurate second order calculation have been published by the PLUTO collaboration [4-9] and reported for a combination of the PLUTO and TASSO data [4-10]. Both of these determinations used the EECA, and both find a A value in agreement with that given above, the PLUTO analysis finding A between 100 and 300 MeV. The QCD scale parameter can also be determined from deep inelastic scattering, where the running of the coupling constant appears as a logarithmic dependence of the structure functions on q 2 , the momentum transferred by the lepton probe squared. Several difficulties with the determination are discussed in [4-11]. A recent summary of the results is A between 100 and 500 MeV at a mean q 2 of 100 GeV2 [4-12]. The measurement of the strong coupling constant from onia decay is discussed in the following section. 4.5 Measurement of the Running of the Strong Coupling Strength There is no existing model of a non-running strong coupling constant. A first attempt to test the validity of Eqn (3-7) was made with the ad hoc assumptions that both the fractional partial cross sections, F', and the coupling, call it aNR, are independent of center of mass energy. Following this assumption the F used for the non-running analysis were calculated without including the effects of photon radiation in the initial state. In all other respects the analysis was identical to that of 4.3. The resulting X 2 /d.f. using either fragmentation model was 10% larger than that obtained assuming Eqn (3-7), which does not favor or disfavor a running coupling constant. Over the energy range considered the expected variation in a, is simply not discernible, given the size of the statisticalesystematic errors. A detectable variation is provided by including the lower energy measurements from the decays of the T and T'. The QCD understanding is that the bbs bound states decay at rest into three gluons, as shown in Fig 4.22a. This is an as process and so allows a fairly sensitive measure of the coupling. As in the continuum analysis presented above, the difficulty in separating the non-perturbative and perturbative is overcome by factorization. The prediction for the hadronic decay width is a non-perturbative term including the wave function of the bound state times a perturbative term from the decay [4-13]. The leptonic decay width (Fig 4.22b) can be similarly factored. Forming the ratio of these two widths the 31 non-perturbative term cancels and the perturbative terms are calculable, indeed calculated P(T -+ hadrons) P(T-+ p) 10(7r 2 - 9) a(q) 817rQ2 a f1+ 2 2q 7r ' [4-14]: IM where Qb = 1/3 is the charge of the b-quark and bo is from Eqn (3-5) with Nf = 4. The choice of renormalization scale, q, is a bit more flexible than in the continuum analysis, where the scale was set by the energy of the virtual photon, q = \, in the calculation of the closed loop diagrams of Fig 3.4. This is despite the application of the MS scheme in both calculations. Usually the scale suggested in [4-14] is chosen, so that the term inside the square bracket equals zero, i.e. the dependence on the higher order, a4, term is minimized. Using this and the data from the CLEO collaboration [4-151 yields: a,(q from the T decay, and a.(q = 0.48MT,) = 0.151 +0.026 = 0.48 Mr) = 0.165 0:08 from the T' decay. Another determination is possible using the radiative decay width of the upsilon, Fig 4.22c. Again this is calculated and the scale determined in [4-14]: -+ + hadrons) r(T -hadrons) 36Qb 5 a a,(q) f a(q) [ 2 1 2 0 6 ] ' r(T where the scale is taken at q = 0.157 Mr. The data from the CUSB collaboration yield [4-16]: a,(q = 0.157Mr) = 0.226 0067 from the T decay, and a,(q = 0.157MT,) = 0.197+0123 from the T' decay. Again, other choices of scale are possible, but, practically, they yield the same result for A evaluated in the MS scheme. Two systematic uncertainties in these determinations must be mentioned. First, the a, values would increase if the T decayed into glueballs [4-11]. This has not been observed [4-17]. Second, the radiative decay width measured has a dependence on the model used to fit the photon spectrum [4-18]. Below neither of these effects are included in the error; the published numbers are taken at face value. One point remains to be clarified before testing the running of the coupling constant: how to compare measurements involving different numbers of active flavors, Nf. The strong coupling constant, a., is the value directly accessible by experiment and it should be contin- uous and monotonic. The QCD scale parameter, A, can only be derived from the data using formulas like Eqn (3-7), but it is the one free parameter of QCD and should be independent of Nf. Remembering that Eqn (3-7) is from a perturbation expansion which starts with massless quarks and is only valid for V = q > 2Mf, the mass of the heaviest active flavor, it should not be applied at and just above flavor thresholds. The measurements from the T decay have Nf = 4. The measurements using the EECA, with F the b-threshold and have Nf = 5 (threshold effects falling as ;> 22 GeV are well above 3f3). In either region the same value of A is used. Between the two regions the coupling is assumed to decrease smoothly, but, as no measurements from this region are considered, the functional form is unimportant. 32 To test the running of the coupling constant and determine the validity of Eqn (3-7), the combined a, values obtained with the EECA (Table 4.6) were used with the points from the T decays. The best fit to the 9 points using Eqn (3-7) was obtained at: A = 98i1 MeV, with a X2 = 5.2 for seven degrees of freedom. This fit and the a, points are plotted in Fig 4.23. Evaluating the x 2 , the fit corresponds to the 59% confidence level (CL). Fitting for a constant aNR, treating the EECA measurements as described above, yielded: 0.004 with a X2 = 44.5 for seven d.f., which has a CL below 10- 4 . a R = 0.151 Clearly the hypothesis of c, running according to Eqn (3-7) is strongly favored over the hypothesis of a constant a,. To quantify this statement, use is made of the F-test, which depends on the ratio of the two x 2 . This has the advantage of separating the statistical spread of the measured points from their deviations to the given hypothesis [4-19]. For seven degrees of freedom the evaluation is: the running hypothesis is favored over the non-running hypotheses at more than the 99% CL. This conclusion and the value of A determined are effected only slightly when dropping the "good" points at q = 0.48 MT and fi and running is favored over non-running at the 92% CL. 33 = q = 34.6 GeV, then A = 92 MeV Chapter 5 A Stringent Test of QCD Versus Energy The major difficulty in testing perturbative QCD though the analysis of hadronic final states is the connection of the calculable parton level predictions with the observed hadrons. The systematic offset in a, values obtained when using the Ali model and the Lund model reflects this difficulty (see Tables 4.1, 4.4 and 4.5 or 4.6). To arrive at a quantitative test perturbative QCD versus energy, the Vfs dependence of the event topology is examined with the use of the Energy-Energy Correlation Asymmetry. To first order in a, the parton level prediction for the EECA has been derived analytically from Eqn (3-3) [4-5]: - X f (C) = The a(C), 6Crl(1 - a8 a(C) = f (1 - C) - f (C), [2(3 - 1 = -(1 - cos X), 2 (5-1) 6C+ 2C2) ln(1 - C) + 3C(2 - 3C)] ( A(cos x) = F dependence enters only through a,. Extension to second order in a, and inclusion of the effects of the parton level resolution cuts (EcUt &ut) and quark masses has of necessity been done numerically [5-1] using the techniques and calculations of [3-12a] mentioned in 3.2. The result is expressed as: AQCD a( a R ,)+ where Ra is the second order correction to a and (5-2) f is the quark velocity. Over the PETRA energy range the dependence on the quark velocities is linear and small ( 8% for f = b, and so much less on average). Again, the primary i/ dependence of AQCD enters logarithmically through a,. To facilitate the comparison of prediction and observation the measured EECA distributions, Fig 4.15, were corrected for initial state radiation, selection criteria and detector acceptance: A(Perfect Detector) A(Data) = A(Measured) A(Full MCSmlto) A(Full MC Simulation) (5-3) The correction ratio on the right hand side of Eqn (5-3) was found to be ~ 0.7 for cos X > -0.80 and ~ 0.85 for -0.96 < cosX < -0.88. More important than the value, the ratio was found not to depend significantly on the fragmentation model, on Vs- for Ns > 30 GeV, or on the parton type (i.e. it does not depend on a,). To obtain equal statistical significance the bins in cos X are coarser than in 4.3. Avoiding the use of any fragmentation model, the perturbative prediction of Eqn (5-2) was fit directly to the corrected data for -0.72 < cosX < 0.0 and 22 < ,/F < 47 GeV. The best fit was obtained at A = 85+1 MeV with a X 2 /d.f. towards -1 = 1.1. If the cosX range is extended the quality of the fit decreases, reaching a X 2 /d.f. = 1.7 for -0.88 < cosX 34 0.0. The perturbative prediction consistently yields too large an asymmetry for cos X ! -0.60, that is, outside the range of hard three jet events. To probe whether the theory can match the observation over a wider range of events, the expected non-perturbative effects need to be considered, preferably avoiding the complicated fragmentation models of 3.3. For the EECA, using only the assumption of limited transverse momentum fragmentation yields a simple parameterization [4-6], [5-2]: A(QCD ® Fragmentation) = AQCD (1 + C( ) (5-4) Roughly C can be taken as a measure of the net fragmentation effect, so the exact functional dependence of C on cos X is, as yet, unknown. Eqn (5-4), if valid, displays the utility of testing perturbative QCD versus energy, the fragmentation effects having a different evolution with Vs than the existing perturbative calculation, namely 1/fl compared to 1/log F. C was estimated to be ~ -8 GeV for -0.9 < cos X by applying the Ali independent jet model to the second order calculation [4-6] or by extrapolating the application of the Lund color string model to the first order calculation [5-2]. The parameterization (5-4) does not include the non-perturbative contribution of q-q fragmentation to the EECA. This contribution, Agq, is largest for cos X significant for cos X > - cos AData(X, /)= 26 cut. -- -1 but is still Including this the full prediction is: AQCD(XlA)(1 + 2-) + F 2 (V, A)A q(X, where F 2 is the partial cross section for qq events as explained in 3.2. ) (5-5) In practice Agq estimated with either the Lund or Ali model agrees. The average is used. The data, corrected by (5-3) versus ,F for different cosX bins are plotted in Fig 5.1a-g. Also shown are the best fit values from Eqn (5-5) to the region -0.88 22 < < cos X < 0.0 and fV < 47 GeV. The values determined for C(cos X) are in Table 5.1, the QCD scale parameter was determined with this method to be: A = 114+" MeV, with a X 2 /d.f. = 0.81 Allowing a different C in each cos X bin while requiring one overall value for A to fit the entire distribution A(X, Vi) has two advantages. First, the data is well described up to the parton level resolution cut jcosXJ L I cos28eua| = 0.9; fitting for a C independent of cosX yielded twice the X 2 /d.f. Second, the error on A covers the uncertainty from fragmentation without resorting to complicated models. Fragmentation is accounted for in a general but sufficient fashion. The C values determined above are in agreement with expectation, they are large and negative for cos X near -1 and go to zero with increasing cos X. The , data in the bin closest to cos X = -1, dependence of the (Fig 5.1a), could not be described with Eqn (5-5). This was expected as this bin has a large qq contribution and straddles cos X = - cos 2Scut. 35 To summarize the results of this chapter, the fi dependence of event topologies is well described by perturbative QCD for 3-jet configurations and without fragmentation corrections. A similar conclusion was reached in [4-9,10]. This description is improved and can be extended nearly to the resolution boundary with the inclusion of a simple power correction. 36 Chapter 6 Conclusions In this study the strong coupling constant, ao, has been determined to complete second order over the center of mass energy range 22 < -,F < 47 GeV using the MARK-J data collected at PETRA and several measures of gluon bremsstrahlung. Present theoretical understanding prefers the determination using the Energy-Energy Correlation Asymmetry: C,(5 = 34.6 GeV) = 0.119 0.005 (stat.) 0.020 (syst.), where a 14% systematic uncertainty has been included to account for the discrepancy between the results from the two fragmentation models. Within the MS renormalization scheme this is equivalent to a value of the QCD scale parameter A = 78 t- (stat.) ii22 (syst.) MeV. A value of A has been extracted from the data over a restricted region but without any fragmentation model which supports the above result, A(frag. = 0) = 85 i" (statistical) MeV. Therefore, the relative effect of fragmentation models to pure QCD is small in this region. These data do not conclusively favor or disfavor the expected variation of the coupling. By including the a, measurements from the T and T' decay widths the strong coupling constant has been shown to run with the center of mass energy. The scale parameter determined in this way is A(V/8 dependence) = 98 t+ The . MeV, which confirms the value quota above. dependence of the topology of hadronic final states has been studied with the use of the EECA. Perturbative QCD to second order has been found to provide an adequate description of this dependence for 3-jet configurations, specifically for cos X > -0.60, without the inclusion of any hadronization effects. Other configurations, -0.90 ! cos X < -0.60, are well described by perturbative QCD plus a simple parameterization of the non-perturbative effects. These effects have been determined. They range, at Fs = 35 GeV, from about 30% for -0.88 < cos X < -0.80 to less than 10% for -0.60 < cos X. The A value which best describes the data in this fashion is A = 114 jr MeV, consistent with the values quoted. The error brackets possible non-perturbative corrections. Looking ahead, two ways to proceed in the study of the strong interaction at high energies suggest themselves. Extremely difficult but necessary is the extension of the calculations to higher orders of a,. More promising is the availability of data at higher energies, fs = 60 GeV at TRISTAN and . = 90 to 160 GeV at LEP. This study has attempted, in part, to show that, even at PETRA energies, the uncalculable effects can be simply described, and that the detailed predictions of the theory can be tested, provided proper variables are used. With the increased center of mass energy these effects should cease to inter pose themselves between observation and calculation. This should allow more rigorous tests, which will hopefully lead to a more quantitative understanding of the strong interactions. 37 References [1-1] M. Gell-Mann, Phys. Lett. 8 (1964) 214. [1-2] G. Zweig, CERN TH 401 (1964); TH 412 (1964). [1-3] Y. Ne'eman, Nucl. Phys. 26 (1961) 222; M. Gell-Mann, Phys. Rev. 125 (1962) 1067. [1-4] M. Gell-Mann, Proc. of the Int. Conf. on High Energy Nucl. Phys. (Geneva, 1962) p. 805. [1-5] V. E. Barnes, Phys. Rev. Lett. 12 (1964) 204. [1-6] 0. W. Greenberg, Phys. Rev. Lett. 13(1964)598; M. Y. Han and Y. Nambu, Phys. Rev. B139 (1965)1006; M. Gell-Mann, Acta Phys. Austr. Supp. IX (1972) 733. [1-7a] S. L. Adler, Lectures on Elem. Particles and Field Theory, Brandeis Summer Institute 1970 (MIT Press, 1971). b] T. Applequist and H. Georgi, Phys. Rev. D8 (1973) 4000; A. Zee, Phys. Rev. D8 (1973) 4083; H. D. Politzer, Phys. Rep. 14 (1974) 129. [1-8] E. D. Bloom et al., Phys. Rev. Lett. 23 (1969) 930; J. I. Friedman and H. W. Kendall, Ann. Rev. Nucl. Sci. 22 (1972) 203. [1-9] J. D. Bjorken, Proc. of the 1967 Int. Symp. on Electron and Photon Interactions at High Energies (SLAC, 1967) p. 109; R. P. Feynman, Photon-Hadron Interactions (Benjamin, Reading, Mass. 1972) p. 152; J. D. Bjorken and E. A. Pachos, Phys. Rev. 185 (1969)1975. [1-10] F. Eisele, Proc. XXI Int. Conf. on High Energy Physics, Journal de Physique 43, Colloque C-3, supplement to number 12, C3 (1982) 337. [1-11] D. H. Perkins, Proc. 161 Int. Conf. on High Energy Physics, Chicago-Batavia, (NAL, Batavia, 1972) Vol. 4, p. 189. [1-12] J. J. Aubert et al., Phys. Rev. Lett. 33 (1974) 1404; J. E. Augustin et al., Phys. Rev. Lett. 33 (1974) 1406; G. S. Abrams et al., Phys. Rev. Lett. 33 (1974) 1453. [1-13] T. Applequist and H. D. Politzer, Phys. Rev. Lett. 34 (1975) 43; A. DeRdjula and S. L. Glashow, Phys. Rev. Lett. 34 (1975)46; T. Applequist, A. DeRdjula, H. D. Politzer and S. L. Glashow, Phys. Rev. Lett. 34 (1975) 365. [1-14] An enormous amount of work has been done on QCD, a few key references: C. N. Yang and R. L. Mills, Phys. Rev. 96 (1954) 191; H. Fritzsch and M. Gell-Mann, Proc. 16- Int. Conf. on High Energy Physics, Chicago-Batavia (NAL, Batavia, 1972) Vol. 2, p. 135; H. Fritzsch et al., Phys. Lett. 47B (1973) 365; D. J. Gross and F. Wilczek, Phys. Rev. Lett. 30 (1973) 1343; Phys. Rev. Lett. D8 (1973) 3633; Phys. Rev. Lett. D9 (197) 980; H. D. Politzer, Phys. Rev. Lett. 30 (1973) 1346; H. Gorgi and H. D. Politzer, Phys. Rev. D9 (1974)416; J. B. Kogut and L. Susskind, Phys. Rev. D9(1974)697; [1-15] G. t'Hooft, Nucl. Phys. B33 (1971) 173; B35 (1971) 167. 38 D9(1974)3391. [1-16] The three jet nature was established in: D. P. Barber et al., Phys. Rev. Lett. 43 (1979) 830; see also: R. Brandelik et al., Phys. Lett. 86B (1979) 243; Ch. Berger et al., Phys. Lett. 86B (1979) 418; W. Bartel et al., Phys. Lett. 91B (1980) 142. [1-17] J. Ellis, M. K. Gaillard and G. G. Ross, Nucl. Phys. B111 (1976) 253; (Erratum B130 (1977) 516); A. DeRdjula et al., Nucl. Phys. B138 (1978) 387; T. A. DeGrand, Y. J. Ng and S. H. Tye, Phys. Rev. D16 (1977) 3251. [2-1] PETRA Proposal (Updated Version), DESY (1976). [2-2] D. P. Barber et al., Phys. Rep. 63 (1980) 337; B. Adeva et al., Phys. Rep. 109 (1984) 131. [2-3] The specific phototubes used are: AMPERX XP2230 for the A and B counters, RCA 4525 for the C counters and Thorn/EMI 9814B for the outer calorimeter counters. [2-4] M.-C. Ho, Ph.D. Thesis, NIKHEF, Amsterdam (1983). [2-5] H.-G. Wu, Master Thesis, Hefei University (1984); H.-S. Chen, Ph.D. Thesis, MIT (1984). [3-1] My favorite: G. Kramer, Theory of Jets In Electron-Positron Annihilation (Springer, Hamburg, 1984). [3-2] F. A. Berends, K. F. J. Gaemers and R. Gastmans, Nucl. Phys. B63 (1973) 381; Nucl. Phys. B68 (1974) 541; F. A. Berends, R. Kleiss, Nucl. Phys. B177 (1981) 237; B178 (1981) 141. [3-3] S. S. A. S. L. Glashow, Nucl. Phys. 22 (1961) 579; Weinberg, Phys. Rev. Lett. 19(1967) 1264; Phys. Rev. D5 (1972) 1412; Salam, Elementary Particle Theory, ed. N. Svartholm (Stockholm, 1968) p. 361; L. Glashow, J. Ilipoulos and L. Maiani, Phys. Rev. D2 (1970) 1285; J. Ellis and M. K. Gaillard, CERN 76-1 (1976). [3-4] This formula was first derived by J. Ellis et al., [1-17]. [3-5] Evaluation of the diagrams in Fig 3.2c requires a regularization procedure, the most commonly used: G. t'Hooft and M. Veltman, Nucl. Phys. B44 (1972) 189. [3-6] This theorem, originally stated for QED processes, has two parts: (1) The cross section for e+e- -+ X, such that X is a sum over all indistinguishable configurations, is finite. (2) The fully inclusive (i.e. total) cross section is finite in the limit of vanishing quark masses. T. Kinoshita, J. Math. Phys. 3 (1960)650; T. D. Lee and M. Nauenberg, Phys. Rev. 133 (1966) 1594. [3-7] Calculations of the three-parton contribution are in [1-17], for two-partons see: a] G. Sterman and S. Weinberg, Phys. Rev. Lett. 39 (1977) 1436. b] P. Binetury and G. Giraldi, Phys. Lett. 83B (1979) 382; P. M. Stevenson, Phys. Lett. 78B (1978)451; B. G. Weeks, Phys. Lett. 81B (1979) 377. [3-8] W. A. Bardeen, A. J. Buras, D. W. Duke and T. Muta, Phys. I ev. D18 (1978) 3998; A. J. Buras, Rev. Mod. Phys. 52 (1980) 199; M. J. Marciano, Phys. Rev. D29 (1984) 580. 39 [3-9] M. Dine and J. Sapirstein, Phys. Rev. Lett. 43 (1979) 668; K. G. Chetyrkin, A. L. Kataev and F. V. Tkachov, Phys. Lett. 85B (1979) 277; W. Celmaster and R. J. Gonsalves, Phys. Rev. Lett. 44 (1979) 560; Phys. Rev. D21 (1979) 3112. [3-10] A partial calculation is given in: B. Lampe and G. Kramer, DESY 85-030 (1985). [3-11] A. Ali et al., Phys. Lett. 82B (1979) 285; Nucl. Phys. B167 (1980) 454; J. G. K6rner, G. Schierholz, J.Willrodt, Nucl. Phys. B165 (1981) 365; K. J. F. Gaemers and J. A. M. Vermaseren, Z. Physik C7 (1980) 235; 0. Nachtmann and A. Reiter, Z. Physik C14 (1982) 47; C16 (1980) 45. [3-12a] R. K. Ellis, D. A. Ross and A. G. Terrano, Phys. Rev. Lett. 45 (1980) 1226; Nucl. Phys. B178 (1981) 421; R. K. Ellis and D. A. Ross, Phys. Lett. 106B (1981) 88; A calculation shown to give similar results is: J. A. M. Vermaseren, K. J. F. Gaemers and S. J. Oldham, Nucl. Phys. B187 (1981) 301. b] K. Fabricius, I. Schmidt, G. Schierholz and G. Kramer (and permutations), Phys. Lett. 97B (1980) 431; Z. Physik C11 (1982) 315; B. Lampe and G. Kramer, DESY 82-025 (1982); F. Gutbrod, G. Kramer and G. Schierholz, Z. Physik C21 (1984) 235. c] R. Y. Zhu, Ph.D Thesis, MIT (1983). d] T. D. Gottschalk and M. P. Schatz, Cal. Tech. CALT-68-1173 (1985); CALT-68-1173 (1985); Phys. Lett. 150B (1985) 451. [3-13] T. D. Gottschalk, Phys. Lett. 109B (1982) 331. [3-14] A. Ali et al.,Z. Physik C1 (1979)203; C1 (1979) 269; Phys. Lett. 83B (1979) 375; Nucl. Phys. B168 (1980) 490 A. Ali et al., DESY 79-86 (1979); Phys. Lett. 93B (1980) 155. Important groundwork for this model was lain in: G. Altarelli and G. Parisi, Nucl. Phys. B126 (1979) 298; D. J. Gross, Phys. Rev. Lett. 32 (1979) 298. [3-15] R. D. Field and R. P. Feynman, Nucl. Phys. B136 (1978)1. [3-16] B. Andersson, G. Gustafson, and C. Peterson, Z. Physik CL (1979) 105; B. Andersson and G. Gustafson, Z. Physik C3 (1980) 223; B. Andersson, G. Gustafson, and T. Sj6strand, Z. Physik C6 (1980) 235; Nucl. Phys. B197 (1982)45; G. Gustafson, Z. Physik C15 (1982) 155; T. Sj6strand, Comput. Phys. Commun. 27 (1982)243; 28 (1983) 229. [3-17] T. D. Gottschalk, CERN TH 3810 (1984). [3-18] Though starting from different ideas, both models specify fragmentation via the same four quantities: a longitudinal fragmentation function, f(z), where z is the fraction of the momentum carried by the sea quark parallel to the fragrienting quark (in the Ali model) or fragmenting string (in the Lund model); the mean t ransverse momentum of the sea quark, Uq; the fraction of pseudoscaler mesons produced, r; and the sea quark flavor. The parameters which best describe the data used in thm study are: For the Ali Model: For u, d, s quarks: For c, b quarks: For all quarks: f(z) = 1 - a + 3a(1 - z) 2 , with a = 0.7, f(z) = 1, -q ~ 0.30 GeV. For the Lund Model: 40 all u, d s c b all quarks: quarks: quarks: quarks: quarks: quarks: f(z) a a a a Uq = = = = = ~ (1 + a)(1 - z)a, 0.50, 0.35, 0.15, 0.05, 0.42 GeV . For For For For For For For both models r is taken as I and sea quarks are given flavors in the ratio u:d:s = 2:2:1, based on inclusive particle spectra. The uncertainty on oq for either model is ; 12%, as determined using the methods of: R. B. Clare, Ph.D Thesis, MIT (1982). [3-19] W. Bartel et al., Z. Physik C21 (1980) 37; C25 (1984) 231. [3-20] P. Hoyer et al., Nucl. Phys. B161 (1979) 349. [3-21] R. Odorico, Z. Physik C4 (1980) 113; G. R. T. B. C. D. D. R. Fox and S. Wolfram, Nucl. Phys. B168 (1980) 285; Field and S. Wolfram, Nucl. Phys. B213 (1983)65; Gottschalk, Nucl. Phys. B214 (1983) 201; Webber, Nucl. Phys. B238 (1984) 492. [3-22] H. G. Sander, Diplomarbeit, Aachen, HEP 74/04 (1974). [3-23] T. A. Gabriel and R. L. Bishop, Nucl. Instrum. Meth. 155 (1970) 81, and references therein. [3-24] H. Fesefeldt, RWTH Aachen, Report PITHA 85/02 (1985). [4-1] T. Walsh, Proc. of the Int. Europhys. Conf. on High Energy Phys. (Brighton, UK, 1983) p. 245. [4-2] see A. DeRdjula et al., [1-17] and D. P. Barber et al., [1-16]. [4-3] Oblateness was suggested by H. Gorgi. [4-4] K. Lanius, DESY 80-36 (1980); J. Dorfan, Z. Physik C7 (1981) 349; H. J. Daum, H. Meyer and J. Bui'ger, Z. Physik C8 (1981) 167; K. Lanius, H. E. Roloff and H. Schiller, Z. Physik C8 (1981) 251. [4-5] L. S. Brown, C. L. Basham, S. D. Ellis and S. T. Love, Phys. Rev. Lett. 41 (1978) 1585, Phys. Rev. D21 (1979) 2018. [4-6] A. Ali and F. Barreiro, Phys. Lett. 118B (1982) 155; Nucl. Phys. B236 (1984) 269. [4-7] B. Adeva et al., Phys. Rev. Lett. 50 (1983) 2051; 54 (1985) 1750. [4-8] However, a recent summary of PEP and PETRA results on a. is given in: R.-Y. Zhu, Proc. of the DPF Conf. (Portland,Oregon, 1985), to be published. [4-9] Ch. Berger et al., Z. Physik C28 (1985) 365. [4-10] F. Barreiro, DESY 85-086 (1985). [4-11] D. W. Duke and R. G. Roberts, Phys. Rep. 120 (1985) 257; and references therin. [4-12] G. Altarelli, Proc. of the Int. Europhys. Conf. on High Energy Phys. (Bari, Italy, 1985), to be published. [4-13] This factorization holds to at least the first non-leading order in a,. The subject is reviewed in: E. Remiddi, E. Fermi Int. School of Phys. (Vareniii, 1980). 41 [4-14] P. B. Mackenzie and G. P. Lepage, Phys. Rev. Lett. 47 (1981) 1244; S. J. Brodsky and G. P. Lepage, Phys. Rev. D28 (1983) 228. [4-15] P. Avery et al., Phys. Rev. Lett. 50 (1983) 807; and references therein. [4-16] R. D. Schamberger et al., Phys. Lett. 138B (1984) 225; [4-17] I. C. Brock, Proc. of the DPF Conf. (Sante Fe, NM, 1984) p. 214. [4-18] R. D. Field, Phys. Lett. 133B (1982) 248; [4-19] P. R. Bevington, Data Reduction and Error Analysis for the Phys. Sci. (McGraw-Hill, New York, 1969) p. 185. [5-1] D. Richards, W. Stirling and S. D. Ellis, Phys. Lett. 119B (1982) 193; and see [3-12c] and [4-6]. [5-2] L. S. Brown and S. D. Ellis, Phys. Rev. D24 (1981) 2383; S. D. Ellis, Phys. Lett. 117B (1982) 333. 42 Data Summary (F) f dt 14* 22 14.0 1.6 22.0 3.2 33-36 39-43 43-45 45-47 GeV 34.6 87.5 41.3 6.7 44.1 18.2 46.1 6.0 GeV pb-1 # thousand 1.1 3.5 1.4 22.7 2.0 2.3 Events Acceptancet 64. 73. 76. < 77. * uncertainties about b-quark production and decay limit the utility of the 14 GeV Data t model uncertainties lead to a systematic error of ~ 2% on the acceptance - -% Table 2.1 Summary of the event sample upon which this study is based. 43 Strong Coupling Strength From: Rate of Large vFs (GeV) 22 rate(OB ;> 0.3) .145 B Events 33-36 39-43 43-45 45-47 .130 .116 .120 .128 Model from: 0 as 22-47 as We) Ali .138 .140 .122 .122 .128 .134 stat. error .017 .005 .012 .007 .012 .004 syst. error .015 .010 .017 .017 .019 .012 YcUt .008 .006 .013 .0132 .016 OB cut acceptance .008 .007 .003 .003 .002 .007 - .008 .002 .003 Og - min(Ehit) - .006 .004 - .002 .006 .007 .007 .003 .007 - .006 .006 - as as(V%) Lund .162 .157 .145 .141 .141 .154 stat. error .02 .005 .013 .008 .013 .004 syst. error from: Y,,ut .019 .014 .011 .004 .013 .005 .015 .010 .016 .010 .011 .005 .003 .001 .002 .002 .000 .001 accept. cut .009 .007 - .006 .008 .009 Uq .003 .006 .003 .007 OB cut Combined - a5 (v/so) = .144 .004 (stat.) - .015 (syst.) Table 4.1 The strong coupling constant determined from the rate of large 0 B events at the five center of mass energies and determined over the entire energy range 22 <; %f < 47 GeV and evaluated at V/f = 34.6 GeV, and the estimates of systematic uncertainty. 44 Different Acceptance Cuts used in Fig 4.5, 4.11 and 4.19 Cut# E > Ev/i < EI/Ey < Ez/Ev < cos 0 < 1 .50 .60 .60 - 1 2 1 3 I .70 1.30 .30 .30 - .70 I - .50 .50 I - Acceptance at \is= using Ali using Lund .84 .86 1 1 .76 .78 .69 .71 1 1 6 7 8 .70 .50 .50 .85 .70 1.30 .30 .30 .80 1.20 .20 .20 .80 1.20 .20 .20 .85 - .85 .62 .63 .58 .59 .50 .51 .43 .43 4 5 .50 .60 .60 .85 35 GeV .66 .67 Standard Cut Table 4.2 Different acceptance criteria used to define the event sample and so to estimate the systematic uncertainty on a. and a ( so) from a variety of sources, and the resulting acceptance at Vs = 35 GeV. 45 Distribution of Jet Multiplicities From Each Parton Type 2 : 3 : 3': 4 : qq qqg qqg qq(gg or q'4') # Jets 00(a0) 0 (C,) 0 (a2) 2 3 4 .81 .36 .47 .12 .17 .50 .43 .52 .02 .12 .09 .31 >5 - Parton Type => .01 .01 .05 Table 4.3a Predicted percent contribution of a given parton type to each jet multiplicity (using the Ali model at fi = 35 GeV, with cluster algorithm parameters of (Ejet = 0.13, 6 jet = 19.40) and parton resolution cuts of (Ecut = 0.10, 6 cut = 12.90). For example, 50% of the first order qqg events are observed as 3-jets. Composition of Each Jet Multiplicity by Parton Type Parton Type 42 : 3 : 3': 4: q-j q~qg qqg q-q(gg or q'q') # Jets 0 (c,) 0 (a2) 0 (a2) 2 3 4 >5 .77 .18 .04 .01 .33 .50 .08 .09 .16 .54 .07 .24 .07 .54 .32 .07 Table 4.3b The estimated fractional compostition of a measured jet multiplicity by parton type (parameters as Table 4.3a). For example, 58% of the observed 3-jet events come from first and second order qqg events (50% and 8% respectively). 46 Strong Coupling Strength From: 3 Jet Multiplicity (Ejet = .13, 5 cut = VfI (GeV) 22 rate of 3 Jets .382 12.90), (Ejet = .13, bjet = 19.40) 33-36 39-43 43-45 45-47 .301 .296 .291 .304 Model as 22-47 as(WS-0) Ali .136 .116 .127 .125 .130 .123 stat. error .027 .004 .007 .005 .008 .003 syst. error .046 .041 .035 .021 .025 .035 5 .018 .008 .011 .008 .015 .014 acceptance .009 .004 .007 .002 .003 .003 Uq - .012 .039 (e, ')cut, (E, )jet} min(Eit) - .005 .019 - .010 .030 as Lund .134 .144 .146 .149 +.042 .015 .005 05 .010 .10 .006 06 .010 .10.003 syst. error .053 .033 .033 .034 .035 .035 {(E, 6 )cut, (E, O)jet} acceptance .020 .010 .008 .014 .016 .012 .038 .004 .007 .001 .003 .002 aq - .013 .008 - .008 stat. error _____________ Combined .090 as (VF') - as(Ne) = .133 .003 (stat.) .037 (syst.) Table 4.4 The strong coupling constant determined from counting the number of 3-jet events for the different center of mass energies and simultaneously from a fit to all the data 22 < F < 47 GeV (evaluated at -Fs=34.6 GeV), and the estimates of systematic uncertainty. 47 .143 Strong Coupling Strength From: Energy-Energy Correlation Asymmetry N/s (GeV) 22 33-36 Model Ali 39-43 43-45 45-47 as .068 22-47 as We) .119 .071 .094 .122 .108 .005 .017 .010 .015 .005 .018 .014 .018 .013 .013 .010 .002 .010 .006 .002 .003 .015 .003 .009 .006 .009 .005 .003 acceptance .014 .010 .008 .007 .005 .003 aq - .008 - .004 - .006 - .002 - .002 - .002 0.7 1.5 stat. error syst. error from: Ecut cosXO min(Ehit) x 2 /d.f. +.02 0.8 1.3 1.5 1.5 cis Lund .080 as(Vo) .133 .097 .106 .123 .121 stat. error +:050 .007 .020 .012 .019 .006 syst. error .036 .010 .023 .014 .018 .010 .022 .019 .020 .003 .004 .004 .002 .017 .014 .003 .012 .004 .002 .011 .013 .001 .004 .007 from: Ecut cosXO acceptance Og x 2 /d.f. Combined - .006 0.9 1.3 - .006 1.6 1.8 cs(F = 34.6 GeV) = .115 - .006 1.0 1.5 .006 (stat.) t .012 (syst.) Table 4.5 The strong coupling constant determined by fitting the EEC Asymmetry distribution using the X 2 defined in Eqn (4-5). The values are for the five center of mass energies and the fit to all the data 22 < F < 47 GeV, evaluated at \F = 34.6 GeV. The prediction is made with the full MC simulation, including the Ali or Lund fragmentation models. The statistical errors given have been scaled to reflect the mean X 2 /d.f. Nominally cos X > -0.72 and Ecut = 0.10 were used. The estimates of systematic u ncertainty the sources considered are also given. The systematic error on the combined value has been increased to reflect the uncertainty due to fragmentation. 48 Comparison and Summary of a. Results Determined with the EEC Asymmetry 39-43 33-36 22 ,Fs (G eV) 43-45 45-47 using the Ali model .119 .005 .071 .017 .094 .010 .122 .015 .119 .004 .090 .010 .105 .007 .115 .012 +.28 .119 .005 .081 .020 .100 .010 .119 .015 From Table 4.5 .080 +.050 .133 .007 .097 .020 .106 .012 .123 .019 Indep. Analysis .110 .040 .144 .006 .102 .012 .116 .008 .118 .015 Average 095 8 .139 .007 .100 .020 .111 .012 .120 .019 - .129 .012 .090 .023 .105 .013 .119 .019 From Table 4.5 .068 Indep. Analysis .085 Average .076 +.025 .030 using the Lund model 4 Combined (Ali.086 Table 4.6 Comparison of the results obtained in 4.3 ("Table 4.5") and an independent analysis of the data and MC, for a. at the five center of mass energies, as determined from fitting the EEC Asymmetry. The larger relative error was taken as the relative error on the simple average of the two sets of results. The combined value is the simple average of the results for the different models, its error has been increased to include the resulting uncertainty. The estimates of systematic uncertainty from Table 4.5 need to be increased to reflect the difference. 49 MeV +5' s114 Determination of AMg and Fragmentation Effect from Vs Dependence A(Dat) at Ns Bin < C(cos x) (GeV) -. 8 8 .304 .006 -. 88 < cos X < -. 80 .153 .003 -9.3 72 .0883 .0021 -6.5 +2.4 -. 72 < cos X -. 60 .0527 .0013 -5.1 +2.6 -. 60 < cos X -. 48 .0307 .0010 -0.5 -. 48 < cos X < -. 24 .0155 .0006 0.8 +3.4 +3. 2.9 .0006 -2.6 -. 80 < cos X <; -. -. 24 < cos X -. 00 A__ .0044 =_114 _MeV-4.4 no fit + -. 96 < cos X = Table 5.1 Results from fitting the EEC Asymmetry as a function of center of mass energy directly to the QCD prediction with the possible fragmentation effects in each bin accounted for by C(cos X)/ i, as given in Eqn (5-5). The error given on Afs includes the uncertainty from the C parameters. The EECA measured at Ve= 34.6 GeV and corrected as in Eqn (5-3) is also given. 50 Figure Captions 2.1 Layout of the PETRA storage ring and experiments. 2.2 The integrated luminosity collected with the MARK J detector over the last six years versus the center of mass energy, Vs. 2.3 The MARK-J Detector (a) end view, (b) side view, (c) cross section at constant q. 2.4 Energy deposited by hadronic events in the inner calorimeter (A, B and C in Fig 2.3) divided by .F at (a) 35 GeV and (b) 44 GeV. The remaining energy is deposited in the outer calorimeter (K). In this figure and in Fig 2.5-2.7, 2.10 and 2.11 the solid symbols are the measured values and the connected open symbols or histograms refer to the two different detector simulations discussed in 3.4 as noted. For this figure the GHEISHA values are smoothed owing to poor statistics. 2.5 Energy weighted distribution of the inner calorimeter hits in cos 0 at (a) 35 GeV and (b) 44 GeV. The rectangles at the bottom of the figure indicate the radial and longitudinal extent of the different counter layers. 2.6 As 2.5, except in the outer calorimeter. The rectangles indicate the extent of the different layers and the amount of iron in between them, showing, despite adjustment with the counter track 0, the decreased angular acceptance compared with the inner calorimeter. 2.7 Energy weighted distribution of the counter tracks in 4 at (a) 35 GeV and (b) 44 GeV, showing that the corners at 45*,135*,..., are correctly modeled. 2.8 Schematic production diagrams of the signal (a) and the backgrounds (b-f) -+ hadrons (a) One photon multihadron production, e+e- -+ (b) Tau pair production, e+e- -+ (r -* hadrons + v)(r -+ hadrons or e + v) (c) Two photon hadron production, e+e- -' Y -4 e+e- + hadrons (d) Bhabha scattering, e+e- -- e+e- or e+e-(e) Electron-Beam gas scattering, eA -+ junk (f) Cosmic Rays, pA - junk. 2.9 Computer reconstruction of a "typical" one photon multihadron event (compare with Fig 2.3): (a) The entire detector: views from the end, side, top and an expanded view of the A and B layers. The heavy black lines indicate the counter tracks constructed from energy depositions. (b) top and side view of hits and tracks fit in the vertex detector. Typically, when events need to be scanned, this display and (a) suffice for classification. (c) Expanded end view giving the energy deposited (in GeV) in each of the calorimeter elements and the hits in the S and T drift chambers. 51 2.10 The fractional visible energy for hadronic events at (a) 35 GeV and (b) 44 GeV. The smooth dot-dash curve is a gaussian fit to the data Ev/s > 0.70 with a width of - ~ 18%. 2.11 Mean values of detector related quantities versus Vs-: (a) the fractional visible energy; (b) the fractional energy imbalance perpendicular to the e+e- beams; (c) the fractional energy imbalance parallel to the beams; (d) the fractional energy measured in each of the calorimeter layers. 2.12 The distribution of the 35 GeV data in AE., AE 1 , showing the final acceptance cuts (the area of each box is proportional to the number of events). 2.13 The number of accepted hadronic events in 3.1 Conceptual view of the process e+e- -+ F which constitute the basis of this study. hadrons. 3.2 Feynman diagrams for the order a, 1 calculation: (a) zeroth order pair production; (b) single gluon bremsstrahlung; (c) second order, virtual correction, 2-parton, which interferes with (a). 3.3 Feynman diagrams for the order a, 2 calculation: (a) tree level 4-parton; (b) third order, virtual correction 3-parton, which interferes with 3.2b; (c) uncalculated, fourth order, virtual correction, 2-parton. 3.4 Diagrams involved in calculation of the total cross section to order a,2. The scale, q, is taken as the energy of the incoming = outgoing photon in the MS renormalization scheme, q = V. 3.5 Schematic showing the application of the fragmentation models to a of a qqg event. In the Ali independent jet model, (a), each parton (q or g) is fragmented into a jet of hadrons with out considering the energy and direction of the other partons. In the Lund color string model, (b), color strings which run between the partons are fragmented, as the strings are assumed to have a ~ constant energy density per unit length, this correlates the results of fragmentation i.e. the hadron energies and directions, with the overall configuration of the initial partons. 4.1 Schematic showing the coordinate system of the Energy Flow Variables for a 3 jet event. 4.2 Energy Flow variables versus c-,Fs) F for the data (dark circles), the MC predictions with = 0.144 using the Ali model (open squares connected by dashed lines), and the Lund model (open diamonds connected by solid lines). The dot-dash curve is the qq contribution estimated using either model. (a) the mean value of 1 - Thrust, (b) the mean value of Major, (c) the mean value of minor, (d) the rate of accepted events with Thrust < 0.8, (e) the rate of accepted events with OB > 0.3. 4.3 The values of the strong coupling constant determined using the rate of large 0 events versus Vi using the Ali model (open squares) and the Lund model (open diamonds). Also plotted are the results of the simultaneous fits for ca(V/s?0 ) or the QCD scale parameter, 52 I < 47 GeV using the Ali model (dashed curve) and 4.4 (a) Effect of varying the 0. cut on the a (VO) determination from the data 22 < 5F < 47 A, to the entire energy region 22 < the Lund model (solid curve). GeV for the Ali model (open squares) and the Lund model (open diamonds). (b) The measured variation of the rate of accepted events passing the 4.5 Effect on the as(F0; 0 B cut at 35 GeV. OB) of eight different sets of cuts used to define the event sample (see Table 4.2) versus the estimated acceptance at 35 GeV, using the Ali model (squares) and Lund model (diamonds). The statistical error is ~ the symbol size, not including the point-to-point correlations. 4.6 The a,(O.) results obtained with different Uq values for the Ali and Lund models at F = 35 GeV (dark symbols) and \/s = 44 GeV (open symbols). From [3-18] a, is reliably determined to be one of the 3 central points for each model, excluding the leftmost point which was determined using the Ali model with Uq = 0. 4.7 Effects of imposing a cut on the minimum energy deposition per counter: (a) on the mean value of Ey at 5F = 35 GeV (the dark circles are the data and the solid bar graph is the MC prediction) and at , = 44 GeV (open circles, dashed bar graph). (b) on the a (0,) determination. Also plotted are the results using the GHEISHA detector simulation and using only the inner calorimeter (ABC). 4.8 The measured (dark symbols) and predicted (Ali model: open squares connected by dashed lines, Lund model: open diamonds connected by solid lines) jet multiplicities versus F for (jet,6 jet) = (0.13, 19.40) (circles), (a) 2-jets, (b) 3-jets, (c) 4-jets. The MC prediction is for (Ecut = 0 10 6 . , cut = 12.90) with a,(,F/o) = 0.133. 4.9 The strong coupling strength and QCD scale parameter determined from the rate of 3-jet events, plotted as in Fig 4.3. 4.10 The a, results using the rate of 3-jet events for different sets of (Cjet,,jet) and (cut,6cut). The horizontal scale is the fraction of matched 3-jets, i.e. the fraction of 3-jet events inferred to result from 3-parton events. For the nominal cuts, (et ,,6 et) = (0.13, 19.40) and (ecut = 0 . 1 0 , 6 cut = 12.90), from Table 4.3b, this was found to be 58%. The points plotted as crosses have a large discrepancy between the predicted and measured 2- and 4-jet multiplicity. 4.11 The effect of varying the acceptance cuts on a,(\F/o) determined using the rate of 3-jet events, as Fig 4.5. 4.12 The effect of varying Uq on a,(%,5 0 ; JM) values determined at %Is= 35 and 44 GeV, as Fig 4.6. 4.13 The effect of imposing a cut on the minimum energy deposition per counter, as Fig 4.7b. 4.14 The measured (dark circles) and predicted Energy-Energy Correlation using the Ali Model (dashed histogram) and the Lund model (solid histogram). The plots at different 53 energies are offset by a factor of 10'/2 starting at the bottom, e.g. the (Vs) = 41 GeV plots are multiplied by a factor of 10. The MC histograms were all produced with ia,(V0) = 0.114, not the best fit to these distributions. 4.15 The measured and predicted Energy-Energy Correlation Asymmetry as in Fig 4.14, except the MC prediction is the average of the best fit to the two models a,(v') = 0.114, and, as the vertical scale indicates, the plots for each energy are offset by 1} decades. 4.16 The strong coupling constant determined by fitting the EECA with the full covariance matrix defined in Eqn (4-5) using the Ali model (open squares) and the Lund model (open diamonds). Also plotted are the results of the simultaneous fits for a,(/Fs) using the Ali model (dashed curve) and the Lund model (solid curve). 4.17 The x 2 contribution from each bin, as defined in Eqn (4-5), for the best fit at 35 GeV to the Ali model with cosXO = -0.72, (a) the positive contributions, (b) the negative contributions. The area of each box is proportional to the magnitude of the contribution. 4.18 The effect of different (Ecut, cos Xo) on (a) the X 2 /d.f. and (b) c,50; A) using the Ali (x's) and Lund (diamonds) models. 4.19 The variation of a,Fs; 4.20 ci,(A) at V A) when varying the acceptance criteria, plotted, as in Fig 4.5. = 35 and 44 GeV for different Oq values used in the fragmentation models, as Fig 4.6. 4.21 The effect of imposing a cut on the minimum energy deposition per counter on a,(A) at N/i= 35 and 44 GeV, as Fig 4.7b. 4.22 Lowest order diagrams important in this study for the decay of the T, (a)-+ ggg, (b)-. pp, (c)--+ Ygg. 4.23 Result of the fitting for "running" coupling constant, 4.4. The two lowest energy points hadrons)/(T, T' -+ -y + (squares) are from the ratio of the decay widths for (T, T' -- hadrons), [4-10]. The next two highest points (diamonds) are from the ratio of the decay widths of (T, T' -+ hadrons)/(T, T' -+ pp), [4-9]. For all four of these points the scale V, or more precisely, q, used is that suggested in [4-8]. The five highest energy points (circles) are the a. points determined using the EECA from Table 4.6. 5.1 The EECA data, corrected with Eqn (5-3), versus F for different cos X bins (open circles) compared to the best fit from Eqn (5-5) to the region -0.88 < cos x 5 0 (dotted line), and its component terms: the pure perturbative QCD prerliction, Eqn (5-2), (solid line); the qq contribution (dashed line); and the perturbative predIction times the simple fragmentation correction, AQCD(1 + C(X)//s), Eqn (5-3), (doi dash line). 54 r Exp.JADE 10 8 m HF-Hallen N PLU> Halle NO Halle NW DORIS cJ1 (Ji [INACH E Halle W Halle 0 DESY H NACTI Figure 2.1 Halle SW Exp. MAF KIJ -Halle SO HF-Hallen 4-108m- S syExp.TASSO Luminosity 102 I I I I I E(fLdt) = I I I I I I I I I 132.4 pb -1 7 -4 Q. 4 . . 101 CA1 4~) r.O 100 '-4 t * -..-...* 101' I 10 ..-.-.- I I I I - -.. I 20 30 Vs Figure 2.2 - 40 (GeV) 50 MARK J DETECTOR (Cross Section) r 7 a (D (1 ( D no s m is t PARTICIPANTS: RWTH - Aachen SHOWER COUNTERS DESY -Hamburg MIT - Cambridge TRIGGER COUNTERS DRIFT TUBES BEAM PIPE Amsterdam NIKNEFMAGNETIRON CHAMBERS, MEDIAN ORIF( CHABER, RIF MEIANHEM (I) - RING At OUTER DRIFT CHAMBERS, MULTIPLIERS DRIFT CHAMBERS, INNER -Peking JEN-Madrid CALTECH - Pasadena Figure 2.3a 57 WEIGHT Itotal : -. O0t MAGNETIC FIELD: II T "EN -Madrid MARK J- DETECTOR ..-.......I N, .1Xx5 - A.1 SHOWER COUNTERS TRIGGER COUNTERS 0k E M MINI-BETA QIADRUPOLES K CALORIMETER COUNTERS OT DRIFT TUBES S.TU.V DRIFT CHAMBERS.INNER 4 DRIFT CHAMBERSMEDIAN P.R DRIFT CHAMBERS,OUTER AL-RING 2 MAGNETIROC 3 BEAM PIPE 4 ROTATIONAL SUPPORT COILS--- -- 6 VACUUM PUMP 00 ---mbridge -EY-Hmbr i Figure 2.3b * WEIGHT:~4Oit MAGNETICFIELO:.8 PARTICIPANTS. RW TH -Aachen OESY- Hamburg MIT -Cambridge NIKHEF -Amsterdam EPI -Peking J ?I. DESV 6181 10 PLanes Driftchambei ii 17 KGx M ve-.v~v~v ...... ~ ....... - 45cm Fe ,ounter Ir tchamber 15 cm Fe - 10 cm Fe . y.....v . ............ .. .. .. ... 2.5cm Fe 12 Planes Icm thick calorimeter counters, K S,)T Driftchamber C..-12 Xo mm Pb B: 3 Xo + 5mm A:= 3 X scintiIlator drift tubes e+ em Figure 2.30 59 Inner Cal. Energy I I I - 2.5 11111111 ' I I I I I F1 ld N Nd(C) I 0 Data --- Newman--- Gheisha- 33 -36 GeV 2.0 r 1.5 1.0 (a) 1< - TCYT t 0.5 0.0 2.5 SData - - - Newman- 43 -45 GeV Nd(C) 1.5 Gheisha 1.0 (b) -L -r 0.5 - 2.0 - ld N L II I 7- ~~ I 0.0 0 0.25 0.5 EABC Figure 2.4j 60 I i 1 0.75 S 1.25 1.5 Inner Cal. I I I I I 1 1 1- -33 -36 GeV : -- 1 1 Data Newmar - I 0.8 1 d EAB N d cosO 0.6 0.4 0.2 *- - (a) -- I - 0.0 Data -Newman 143 -45 GeV 0.8 E EABC N d cosO 0.6 0.4 OC- (b) 0.0 -- - 0.2 -1 0 -0.5 Cos 0 Figure 2.5 61 0.5 1 Outer Cal. - I I I I I I I I I II 33 -36 GeV I -N I I I ama 0.8 1 dE1 /EK NdcosO 0.6 0.4 m -Ne - 0.2 - (a) Fe 0.0 0.8 -- 1 dE,/EK e Fe NdcosO 0.6 0.4 (b) 0.2 Fe 0.0 -1 -0.5 0 cos 0 Figure 2.6 62 0.5 1 Counter Track < -33 -36 GeV 0.004 1 dE/Ev Nd 5 0.003 0.002 (a) 0.001 Data Newman I II II I II I II II I I 0.000 I II I I II II I I I II 1 1 1 I 1 1 -43-45 GeV 0.004 1 dEct/Ev Nd k 0.003 - -K 0.002 (b) 0.001 0.000 Data Newman I 0 I I IllIllIll I I I III 180 90 Figure 2.7 63 I I III 270 360 Signal (a) One Photon Backgrounds hadrons e L e+ e+ (btau pair T ~" or r e- e e e+ (c) Two Photon Multi hadrons hadrons e e- e +e (d) Bhbh e+ Scat ering e e- e+ e+ e e e- +- hadronic junk - (e) Electron Beam gas (undisturbed) YZ hadronic junk (f)Cosmic Rays (undisturbed) e- Figure 2.8 64 - ha drons s hUd rn" - Multi hadrons I I 7 A 2 - .. 7_________ - mJ I. I J IT I I1-II nniiii H .. 11111wi I I I I Figure 2.9 a - 0 01 0 0.73 9 ETOT 41. 2 EV 6.0 4.2 -0. EK 9.2 5 EBM 22.05 5 R-E 6973/ 47316 ORTE 1/ 7/84 + - 0 . THRUST W OL WIs*- I I - BH T 4736 NOTHIT 152 NVERT 11 ZVERTEX ...... . .. ... .. .. .... ... ... .. ..Ile.. .. ... .... ... .. .... ... .. .... .. .. .... ... .... .. ... ... ... .. ... ..... .. ... .. ... ... .. ........ .. ... .. .... ... .. ... ... i, .. .. ... .... .. ...... .. .. .... ... ......... ... ..... ... .. I... .. ... .I.. .. ... ... .. .... . ... ... .. .. ..... .. .... .. .... ... ... .. .......... . . .. . . . .. . . . . . . . . . . .. .. . . . . .. ... . . . . . . I . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i . . . ...................... ....................... ... ........................9 ............................... 1.32 . R-E 6973/ y ll pz xr . .... ... ..... ..... .49 . ... .......... .. ... .... ..... .. ... ... .. .... .. .. ..... .... ...... ... .... ..... .. .. ........ ...... ... ... ... .. 0... ... .. ... .. ... . . . . . . .. ..1 ... ..... ........... %. . ............... ....... ..... ... ... Figure 2.9b .. ........................................................................................................... .. ......... if ........ . .. ......................... ............................................... I I1f ~~~~- I / - 7 0.2 0. 4 L _ 1 1 6 I 7' 0" I. I I L I I H I I I I - I I II T __ I I I I Figure 2.9c 67 I , I -6--E-- - I -~ I I 43 -45 GeV ld N ---- 2.5 Cut Newmarr- Gheisha - - g=---.=187 ) .0 - Nd( Data O - Visible Energy 1.5 1.0 1*1 0.5 I 0.0 I I 43 -45 GeV 2.5 JI - NewmarrGheisha a=. 187 - -- 2 .0 ) Nd( ----- Id N Data * - (a) j 1.0 0.5 CP 0.0 0 1 0.5 EyVs Figure 2.10 68 1.5 2 I EV/s - I i I I I I I I --G--Gheisha I I I I I - (a) 1.0 El- ErI I 0.9 0.15 I I I I E, /EV I _ I _I I I I 0.10 0.15 1 I I I I I I I I I I I I (b) Ms- -~E I~ I I I 1 I I I I I 7j (c) E0z /E0 M IM I 10 I I 20 30 s (GeV) Figure 2.11 -0-Newman - 1.1 0 Data Mean Visible Energy/ s Mean Energy Imbalance/Ey I I I I 40 50 Data C) - I I I - E-Newman ----Gheisha I I i - Mean Energy per Layer/VI I I I -II I I 0.4 EA/S 5- - - ii 07 -~ - 0.3 ~_ E c/s 0.2 - ~ -~--- - - -~ EB/VY 0.3 -0- D--o- - 0.1 II II EK/VS 0.2 I 14 0 I I I 20 I I I I I I___I I I 30 S (GeV) Figure 2.11d II II 40 I I I I 50 0.6 . * a.. . . . . . . . .a . a .. . . . .. . . - - - . Vs = . 0.4 . Cuts . 0.5 - : : : 33-36 GeV 22.7k Events U ...... . U .......... 0.3 . ... ... .EE. . n .. a. . - u.EEEE......u..-. * 500 Events . 100 Events . 0.2 --.-.-. 1 Event 0.1 I* 0.0 0 0.1 0.2 0.3 Ez Ey Figure 2.12 0.4 0.5 LI IW 0.6 Hadronic Events I I I I I I I - I I I I I I I [I I I I I 35884 Events 10000 5000 K!) t'3 1000 r-O 500 4. 10 I I I 20 H VS Figure 2.13 -::.:::::::: - 30 40 (GeV) -. . -L1 50 e +c y aa G a Y zo 1 e""Z* REGIME ELECTROWEAK QEDr I PERTURBATIVE OCC L-1OB SERVATIONS FR AGMENTATION MODELS(f,c-g---) E 2BEAM xE 2BEAM Figure 3.1 73 2 f2 M; toM IT Yq e- Figure 3.2a 74 Figure 3.2b Figure 3.2c 75 (0 co ;:: 76 > Figure 3:.3b] 77 Figure 3.3c 78 Figure 3.4 79 g (a) q s (b) Figure 3.5 80 MAJOR Axis = m q4 THRUST Axis q. .9 MINOR Axis = THRUST Axis - ~ p - pr q the Energy Flow Variables Figure 4.1 81 0.20 I I I I j I I Data -b- - Lund -&-- Ali qq only- 0.15 <I-T> 00 0.10 0.05 I 0.00 10 I I I I I I 1 30 20 S Figure 4.2a I - -] 1'. 1 40 (GeV) 1 1 I I I I 50 - I I 0.35 I II I I I I I I I Data Lund Al -qq - -- Ei.- only- 0.30 N <Major> N 00 CA4 0.25 N N 0.20 10 I I 20 30 _ NS Figure 4.2b ~ 40 (GeV) 50 0.25 I I I I I I I I I I I Data 0 Lund Ali -8-~ 0.20 qq only <minor> 00 0.15 MM MMM -M 0.10 H- I u.u0 10 I I I 20 I I 30 II 40 (GeV) Figure 4.2c I I I 50 I - IrData 0 -H-- Ali 0.3 0.2 00 C" Lund qq only + V - - 0.4 0 0.1 4-- 4-1 0.0 10 NS Figure 4.2d 40 30 20 (GeV) 50 0.4 -i I I I j I I I I I I I Data 0.3 1- Lund Ali qq only - Al - - -0- -E U) RI 0 0. 2 H'IN 0 -- 1 0 .4) 4-) 0.1 0.0 I 10 I I I I I 20 S Figure 4.2e ~-~- - 00 I I 30 40 (GeV) I 50 Strong Coupling Strength 0B from: Rate of Large 0.20 -- Events < OLund~ A 341. 0.15 A =166- 0cxs 00 0.10 F- 0.05 F 0.00 1 10 I I I I I 20 I I I I I I I 30 F rS Figure 4.3 I I I 40 (GeV) 50 O( ( from Rate(OB O cut) vs OB cut Fit: 22 Vs <47 GeV 0.20 ~0 Lund ~ D Ali SX(\s) (a) 0.15 0.10 00 00 Fraction of Events B 0 0.2 (b) B CUt 0.1 I I I I I 0.3 0.35 I - - 0 Data at -Vs=34.6 0.0 0.2 0.25 OB cut Figure 4.4 0.4 ) from Rate(OB s a 0.3) vs Acceptance Cut Fit: 225 Vs<47 GeV 0.20 - I I I I I I -- 1 - Lund L u n d cu t n Ali I 0.15 0x 00 R5 k P6 L971 0.10 0.05 Standard Cut -I 0.00 . I 0.4 I I I I I 0.6 I I I- 0.8 Acceptance at 35 GeV Figure 4.5 1 -0.3)vs from Rate(OB at V CCc 0.20 c = 35, 44 GeV - K ' I I I I I I I I I J I -1 illEl + 0.15 cX S ~ 0.10 0 LuLfld Ali 0.05 S(V) -1 0.00 I1 .0 I I 250 ac I I I (\I) = 34.6 GeV = 44.1 GeV I I I 300 350 350 q Figure 4.6 <& 400 (MeV) I I 450 I 500 . Fractional Ey vs Ej cut 1.1 I 10I I I F Data 0 0 MC s) = 44.1: 0 1.0 r 0 0 U 35, 44 GeV at V= I 0.9 )= 34.6: j ~-01 0.8 -L All 0 0- -K U - 0.6 --( 0 L 5 I 0 -. - - - - - - - 0.5 i II I - 1 E cut (GeV) Figure 4.7a 0 1.5 - 0.7 from Rate(OB at V= 0.20- 0.3) vs E cut 35, 44 GeV 0.15 0.10 I (V) 0 0.5 34.6: 1 E cut (GeV) Figure 4.7b - 44.1:LII - s) u.uu 1.5 Rate of 2-Jet Events 1.0 I I I I -e- - Data - Lund Ali ---0.8 0.6 CD3 B- 0.4 b 0.2 I 0.0 I 10 20 I I I I I I I I- I 30 I (GeV) Figure 4.8a I I 40 I i I I 50 Rate of 3-Jet Events I I I I I I I I Data Lund -c- - -8E - 6- - 0.5 Ai -__ 0.4 4p 0 0.3 CYD) 0.2 b 0.1 I 0.0 10 I I I I I I 40 30 20 NS Figure 4.8b I I (GeV) I 50 Rate of 4-Jet Events 0.25 I I I I I I I I I Data I Lund Ali -8-0.20 41 0 0.15 cJ1 0.10 0.05 I 0.00 10 I I I I 20 I I I 40 30 FS Figure 4.8c I (GeV) I I I I 50 Strong Coupling Strength from: 3-Jet Multiplicity 0.20 I ~ - -] - Lund~ - Ali 0.15 0 s A=97 1 0.10 0.05 K- - I 0.00 10 I I I I I I I 30 20 NS Figure 4.9 I I II 40 (GeV) I I I I I I I I 50 afrom 3-Jet Rate vsE, 0.20 0.15 + +++ 0.10 0.05 X4 2.- X 24> 0.00 I 0 *I I I 0.2 0.4 L II 0.6 I 0.8 Fraction(3 Jets From 3 Partons) Figure 4.10 1 from 3-Jet Rate vs A cceptance Cut (s\/ X Fit: 22 v!s<47 GeV 0.20 I I - I I I I I I I I Lund F A ~ cutn- 0.15 Co 00 0.10 4In3 F7 [-6L -w F21 0.05 Standard Cut 0.00 I- I 0.4 I I I I 0.6 I I 0.8 Acceptance at 35 GeV Figure 4.11 1 aS from 3-Jet Rate vs uq at Vs 0.20 35, 44 GeV - I I T\T I 1- 0.15 xs 0.10 Lund Ali 0.05 (s) =34.6 GeV (s) = 44.1 GeV < E I 0.00 .0 250 300 I I 1 1 II 1 cuJ 9Cl Figure 4.12 l II i 350 350 - - -- 400 (MeV) 450 500 as from 3-Jet Rate vs Ei cut 35, 44 GeV at V' 1 - 0.20 I I I I---- Is I - I Lr CO 4j. 4-) Z 0.15 M ifi as 4-0 0 rl 0.10 C C ID I I - - K 0.05 cn ) =44.1: EZ (K) = 34.6: M 0 (9 0.00 m 0 - I I I I I 0 II I I 0.5 I I I I I 1 Ei cut (GeV) Figure 4.13 I I I 1.5 EEC Data Lund Ali 102 Is> idEc 0- c dcosx 22.0 .0 0 - 00--j - -0 0- 101 . 34.6 000 0 00 00 00 0 100 -- a a 0 - a0 0 0 44 1 aA! 0 000 0 0 a 6 44.1 ..-- 4 6 .1 ... 0a a . i o0 a a a- 10-1 ---- -1 -0.5 Cos Figure 4.14 101 0 X 0 - 0 0.5 1 EEC Asymmetry (v' ) 10-1 22 GeV -- 10- 1 10 -2 34.6 -- 10-1 A(cosX)y ol 11 10 -1I 41.3 10-1 44.1- 10 - 1 46.1 e a 10 -2 - - Data Lund -A i 10 -3 -1 -0.8 -0.4 cos X 0.6 Figure 4.15 102 -0.2 0 Strong Coupling Strength from: EEC Asymmetry 0.20 -E] Lund -- Ali 0.15 88 a S 0.10 - 0.05 =400 E-A I 0.00 I I I 10 I I 30 20 Vs Figure 4.16 I I ' 0 (GeV) I I I 40 I I I I 50 0.0 .. U *. . . * i n i -- * Contribution -0.2 . . - - U * . 2 - . * * . -0.4 m Xkl> - -- U 0 Em 0 U) 0 C) E(xk > - 0) 33.3 U -0.6 * U * . 0 2 X from to X =+2.0 flu... 2 -0.8 -0 .8 I I I I -0.6 I -- I -0.4 COS Xk Figure 4.17a SXkl- +0. -0.2 0 2 0.0 U I * I*I . * . * I*I . . . . . -0.2 I . . U U I I U Contribution U U U to x 0 - rom 2 Xk1<l -U 0 -0.4 cf~ 0 C) 01 ik< Z(x -9.2 0) -0.6 x2 * Xkl= 9_ -2.0 -0.8 -0.8 I I -0.6 I I -0.4 COS Xk Figure 4.17b I I -0.2 - 0 X kl . 2 = I I - . C x2 df from EEC Asy Fit vs (ccut, Cos XO) Fit: 22 5 s<47 GeV 2.5 II I I I I III I I I I I - F I I I I I Lund Ali 2.0 X'/df r> 1.5 XWX 0 1.0 - {.07, .10, .13, .15 0.5 0.0 -1 -0.9 -0.8 -0.7 Cos Figure 4.18a X0 -0.6 -0.5 -0.4 s ( s)from EEC Asy vs (cect, cos Xo) Fit: 225 s<47 GeV I I I I I I I I I Lund Ali - 0.20 0.15 0 -1 0.10 0.05 .07, .10, .13, .15- e 0.00 -1 -0.9 -0.8 -0.7 Cos Xo Figure 4.18b -0.6 -0.5 -0-4 a( (\s) V from EEC Asy vs Acceptance Cut Fit: 22v':s<47 GeV 1 0.20 II n Lund Au n cut n [n Ali 0.15 0 (sV 8 5 6 [7 - ( 0.10 00 0.05 I Standard Cut 0.4 II I __ I I 0.6 I I I 0.8 Acceptance at 35 GeV Figure 4.19 I I - II 0.00 1 as from EEC Asy vs 0.20 - cc s at cIT oq = 35, 44 GeV N F II Lund 0.15 0.10 Lund All 0.05 E(V) $ I( 34.6 GeV <> 44.1 GeV o ) 0.00 I .0 250 300 350 350 0- Figure 4.20 q 400 (MeV) I I I I II I 450 1 - - - 500 aS from EEC Asy vs Ej cut at 35, 44 GeV s 0.20 0.15 - ~ -- 0.10 0 (Ks) 0.05 - 0.00 0 0.5 44.1: ED 1 E cut (GeV) Figure 4.21 0 s) = 34.6: 1.5 as (a) as alas 6a (b) e (c) 111 Strong Coupling Strength 0.4 I I I I I I I ~ I I - This Experiment F(T,T'-- ggg) 0.3 - F(T,T'-> tp) +13 A 0.2 98 -11 - cs - F(T,T'- ggg) F(T,T'-> ygg) MeV -zzzzzzzra - 0.1 a 0 I I I I I ' ' 0.0 10 30 20 Vs Figure 4.23 (GeV) I I I I I 40 I I I 50 <5-.88) EECA( -. 96< cos QCD Corrected - - - - - - Data qq only 0.6 0.5 I 00 I I I I I 0.4 A 0.3 0.2 0.1 0.0 10 Figure 5.1a 40 30 20 Vs (GeV) 50 Corrected Data -. 80 5 - QCD(1+C/j )+qq - -- -- - QCD(1+C/v-) ) EECA( -. 88 < cos X QCD - - - --- - qq only 0.25 0.20 A ........... 0.15 ......... 0.10 r - 0.00 10 Figure 5.1b 20 40 30 Vs (GeV) - 0.05 50 EECA( -. Corrected Data - -- - -- - - < 80 cosy X< -. 72) QCD QCD(1+C/ -)+qq QCD(1+C / -)- --- -- - - - qq only 0.150 0.125 0.100 A 01 0.075 I 0.050 I I I I I I I LF~---[--~--kl- 0.025 0.000 10 Figure 5.l c 30 20 Vs 40 (GeV) 50 ) EECA( -. 72< cos Xi5 -. 60 QCD ------------ NQCD(1+C/V-)+qq Corrected Data - -- I I - QCD(1+C/--) - - -- - - - qq only 0.08 0.06 A 0.04 0.02 I I I-L ----- 0.00 10 Figure 5.1d 30 20 NS 40 (GeV) 50 EECA( -. 60 < cos X:!; -. 48) Corrected QCD QCD(1+C/-)+qq - Data - -- - -- QCD(1+-C --qq only 0.05 0.04 A - I I 0.03 0.02 0.01 0.00 10 Figure 5.1e 40 30 20 Vs (GeV) 50 EECA( -. 48 < cos X < -. 24) QCD QCD(1+C/)+qq Corrected -- - - - Data QCD(1+C/--- - -- - - - qq only 0.030 0.025 00 A 0.020 0.015 0.010 I mC Li r[ -vJ R I~~~ 0.005 0.000 10 Figure 5.lf 40 30 20 s (GeV) 50 - Data -. 00 QCD QCD(1+C/V )+qq - Corrected 5 ) EECA( -. 24 < cos y -- - - - QCD(1+C/ )- - - - - - - qq only 0.010 0.008 I A I~- -a .-. - - I 0.006 0.004 0.002 0.000 10 Figure 5.1g 20 30 Vs 40 (GeV) 50