Refine
Year of publication
Document Type
- Article (1871)
- Preprint (1250)
- Doctoral Thesis (593)
- Conference Proceeding (248)
- diplomthesis (101)
- Bachelor Thesis (75)
- Master's Thesis (61)
- Contribution to a Periodical (46)
- Book (33)
- Diploma Thesis (33)
Keywords
- Kollisionen schwerer Ionen (47)
- heavy ion collisions (44)
- LHC (25)
- Quark-Gluon-Plasma (25)
- Heavy Ion Experiments (20)
- equation of state (19)
- quark-gluon plasma (19)
- Relativistic heavy-ion collisions (16)
- QCD (15)
- QGP (15)
Institute
- Physik (4406) (remove)
Abstract: We study transverse expansion and directed flow in Au(11AGeV)Au reactions within a multi-fluid dynamical model. Although we do not employ an equation of state (EoS) with a first order phase transition, we find a slow increase of the transverse velocities of the nucleons with time. A similar behaviour can be observed for the directed nucleon flow. This is due to non-equilibrium e ects which also lead to less and slower conversion of longitudinal into transverse momentum. We also show that the proton rapidity distribution at CERN energies, as calculated within this model, agrees well with the preliminary NA44-data.
To describe ultrarelativistic heavy-ion collisions we construct a three-fluid hydrodynamical model. In contrast to one-fluid hydrodynamics, it accounts for the finite stopping power of nuclear matter, i.e. for nonequilibrium e ects in the early stage of the reaction. Within this model, we study baryon dynamics in the BNL-AGS energy range. For the system Au+Au we find that kinetic equilibrium between projectile and target nucleons is established only after a time teq CM H 5 fm/c C 2RAu/³CM. Observables which are sensitive to the early stage of the collision (like e.g. nucleon flow) therefore di er considerably from those calculated in the one-fluid model.
We examine the properties of both forms of strange matter, small lumps of strange quark matter (strangelets) and of strange hadronic matter (Metastable Exotic Multihypernuclear Objects: MEMOs) and their relevance for present and future heavy ion searches. The strong and weak decays are discussed separately to distinguish between long-lived and short-lived candidates where the former ones are detectable in present heavy ion experiments while the latter ones in future heavy ion experiments, respectively. We find some long-lived strangelet candidates which are highly negatively charged with a mass to charge ratio like a anti deuteron (M/Z 2) but masses of A=10 to 16. We predict also many short-lived candidates, both in quark and in hadronic form, which can be highly charged. Purely hyperonic nuclei like the (2 02 ) are bound and have a negative charge while carrying a positive baryon number. We demonstrate also that multiply charmed exotics (charmlets) might be bound and can be produced at future heavy ion colliders.
Measured hadron yields from relativistic nuclear collisions can be equally well understood in two physically distinct models, namely a static thermal hadronic source vs. a time-dependent, nonequilibrium hadronization o a quark-gluon plasma droplet. Due to the time-dependent particle evapora- tion o the hadronic surface in the latter approach the hadron ratios change (by factors of <H 5) in time. Final particle yields reflect time averages over the actual thermodynamic properties of the system at a certain stage of the evolution. Calculated hadron, strangelet and (anti-)cluster yields as well as freeze-out times are presented for di erent systems. Due to strangeness distillation the system moves rapidly out of the T, µq plane into the µs-sector. Classif.: 25.75.Dw, 12.38.Mh, 24.85.+p
We discuss the early evolution of ultrarelativistic heavy-ion collisions within a multi- fluid dynamical model. In particular, we show that due to the finite mean-free path of the particles compression shock waves are smeared out considerably as compared to the one-fluid limit. Also, the maximal energy density of the baryons is much lower. We discuss the time scale of kinetic equilibration of the baryons in the central region and its relevance for directed flow. Finally, thermal emission of direct photons from the fluid of produced particles is calculated within the three-fluid model and two other simple expansion models. It is shown that the transverse momentum and rapidity spectra of photons give clue to the cooling law and the early rapidity distribution of the photon source.
Hadron and hadron cluster production in a hydrodynamical model including particle evaporation
(1997)
We discuss the evolution of the mixed phase at RHIC and SPS within boostinvariant hydrodynamics. In addition to the hydrodynamical expansion, we also consider evaporation of particles o the surface of the fluid. The back-reaction of this evaporation process on the dynamics of the fluid shortens the lifetime of the mixed phase. In our model this lifetime of the mixed phase is d 12 fm/c in Au + Au at RHIC and d 6.5 fm/c in Pb + Pb at SPS, even in the limit of vanishing transverse expansion velocity. Strong separation of strangeness occurs, especially in events (or at rapidities) with relatively high initial net baryon and strangeness number, enhancing the multiplicity of MEMOs (multiply strange nuclear clusters). If antiquarks and antibaryons reach saturation in the course of the pure QGP or mixed phase, we find that at RHIC the ratio of antideuterons to deuterons may exceed 0.3 and even 4He/4He > 0.1. In S + Au at SPS we find only N/N H 0.1. Due to fluctuations, at RHIC even negative baryon number at midrapidity is possible in individual events, so that the antibaryon and antibaryon-cluster yields exceed those of the corresponding baryons and clusters.
After a brief introduction on QCD and effective models in the first chapter, I analyze the dependence of the QCD transition temperature on the quark (or pion) mass in the second chapter. I found that a linear sigma model, which links the transition to chiral symmetry restoration, predicts a much stronger dependence of T_c on m_pi than seen in present lattice data for m_pi >~ 0.4 GeV. On the other hand, an effective Lagrangian for the Polyakov loop requires only small explicit symmetry breaking to describe T_c(m_pi) in the above mass range. In the third and fourth chapter, I study the linear sigma model with O(N) symmetry at nonzero temperature in the framework of the Cornwall-Jackiw-Tomboulis formalism. Extending the set of two-particle irreducible diagrams by adding sunset diagrams to the usual Hartree-Fock (or Hartree) contributions, I derive a new approximation scheme which extends the standard Hartree-Fock (or Hartree) approximation by the inclusion of nonzero decay widths.
A study of secondary Drell-Yan production in nuclear collisions is presented for SPS energies. In addition to the lepton pairs produced in the initial collisions of the projectile and target nucleons, we consider the potentially high dilepton yield from hard valence antiquarks in produced mesons and antibaryons. We calculate the secondary Drell-Yan contributions taking the collision spectrum of hadrons from the microscopic model URQMD. The con- tributions from meson-baryon interactions, small in hadron-nucleus interac- tions, are found to be substantial in nucleus-nucleus collisions at low dilepton masses. Preresonance collisions of partons may further increase the yields.
A generalized Lagrangian for the description of hadronic matter based on the linear SU(3)L × SU(3)R -model is proposed. Besides the baryon octet, the spin-0 and spin-1 nonets, a gluon condensate associated with broken scale invariance is incorporated. The observed values for the vacuum masses of the baryons and mesons are reproduced. In mean-field approximation, vector and scalar interactions yield a saturating nuclear equation of state. We discuss the di culties and possibilities to construct a chiral invariant baryon-meson interaction that leads to a realistic equation of state. It is found that a coupling of the strange condensate to nucleons is needed to describe the hyperon potentials correctly. The effective baryon masses and the appearance of an abnormal phase of nearly massless nucleons at high densities are examined. A nonlinear realization of chiral symmetry is considered, to retain a Yukawa-type baryon-meson interaction and to establish a connection to the Walecka-model.
Abstract. A generalized Lagrangian for the description of hadronic matter based on the linear SU(3)L × SU(3)R -model is proposed. Besides the baryon octet, the spin-0 and spin-1 nonets, a gluon condensate associated with broken scale invariance is incorporated. The observed values for the vacuum masses of the baryons and mesons are reproduced. In mean-field approximation, vector and scalar interactions yield a saturating nuclear equation of state. Finite nuclei can be reasonably described, too. The condensates and the e ective baryon masses at finite baryon density and temperature are discussed.
Untersuchungen von evolutionären Algorithmen zum Training neuronaler Netze in der Sprachverarbeitung
(1997)
Im Rahmen der vorliegenden Diplomarbeit wurde die Leistungsfähigkeit von evolutionären Algorithmen zum Training von RNN untersucht und mit gradientenbasierten Trainingsalgorithmen verglichen. Die Zielsetzung war dabei im besonderen die Prüfung der Verwendbarkeit in der Sprachverarbeitung, speziell der Spracherkennung. Zunächst wurde anhand eines Prädiktionsproblems die prinzipielle Leistungsfähigkeit von EA untersucht, indem ein MLP mit unterschiedlichen evolutionären Algorithmen trainiert wurde. Verschiedene Varianten von GA und ES sind an diesem Beispiel getestet und miteinander verglichen worden. Im Rahmen der Untersuchungen an GA stellte sich heraus, daß eine Mindestgenauigkeit der Quantisierung zur Lösung erforderlich ist. Es zeigt sich, daß die Genauigkeit der Approximation mit abnehmendem Quantisierungsfehler besser wird. Damit ist eine Behandlung dieses Problems mit grob quantisierten Gewichten nachteilig. Demgegenüber profitiert ES sowohl in der Approximationsgenauigkeit, als auch in der Konvergenzgeschwindigkeit von der direkten Darstellung der Objektvariablen als reelle Zahlen. Weiterhin zeigte sich bei ES, daß die Genauigkeit einer Lösung auch von der Populationsgröße abhängig ist, da mit wachsender Populationsgröße der Parameterraum besser abgetastet werden kann. Im Vergleich mit ES benötigten GA längere Konvergenzzeiten und bedingten zudem aufgrund der Codierung und Decodierung einen höheren Rechenaufwand als ES, so daß die Untersuchungen an RNN nur mit ES durchgeführt wurden. Zunächst wurde mit dem Latching-Problem eine, in der Komplexität eng begrenzte, Klassifikationsaufgabe mit Zeitabhängigkeiten untersucht. Die zur Verfügung gestellte Information war bei diesem Beispiel sehr gering, da der Fehler nur am Ende einer Mustersequenz berechnet wurde. Es stellte sich heraus, daß selbst bei dieser sehr einfachen Aufgabenstellung die gradientenbasierten Verfahren nach dem Überschreiten einer maximalen Mustersequenzlänge T keine Lösung finden konnten. Im Gegensatz dazu war ES in der Lage, das Problem für alle gemessenen Variationen des Parameters T zu lösen. Erst wenn während des Trainings dem Gradientenverfahren zusätzliche Informationen durch Fehlereinspeisung zur Verfügung gestellt wurde, hatte der BPTT-Algorithmus die selbe Leistungsfähigkeit. Als weiteres Experiment mit Zeitabhängigkeiten wurde das Automaton-Problem un- tersucht, welches mittels eines RNN gelöst werden sollte. Bei diesem Problem wurde besonderer Wert auf die Untersuchung des Konvergenzverhaltens bei Änderungen der Parameter von ES gelegt. Die Untersuchungen ergaben, daß die einzelnen Parameter in komplexer Weise miteinander interagieren und nur eine gute Abstimmung aller Parameter aufeinander eine befriedigende Leistung in Bezug auf Konvergenzgeschwindigkeit und Klassifikationsergebnis erbringt. Wie bei dem Latching-Problem wurde der Fehler nur am Ende einer Mustersequenz berechnet. Dies bewirkt, daß der BPTT-Algorithmus bereits bei Sequenzlängen von T = 27 nicht mehr in der Lage ist, die Zeitabhängigkeiten in dem Gradienten zu repräsentieren. Mit ES dagegen konnten RNN trainiert werden, die in der Lage sind, Sequenzlängen bis zu T = 41 richtig zu klassifizieren. Die Untersuchungen bestätigen, daß der beschränkende Faktor in erster Linie der Trainingsalgorithmus und nicht das Netzwerksparadigma ist. Die Simulationsexperimente mit zeitnormierten Sprachdaten zeigen, daß mit ES prinzipiell höhere Erkennungsleistungen als mit dem gradientenbasierten Algorithmus des BPTT erzielt werden können. Jedoch nimmt schon bei der Klassifikation der Zahlwörter Zwei und Drei die Klassifikationsleistung mit zunehmender Sequenzlänge ab. Es erfordert eine drastische Vergrößerung der Populationsgröße, um zumindest gleich gute Ergebnisse zu erzielen. Zusätzliche Tests am Automaton-Problem stützen diese Aussage. Jedoch steigt der Rechenaufwand durch Vergrößerung der Populationsgröße so stark an, daß bei nicht zeitnormierten Sprachdaten ES mit adäquater Populationsgröße nicht mehr simulierbar waren. In den Untersuchungen an dem Vokabular mit sechs Wörtern wurde der Fehler für jeden anliegenden Merkmalsvektor berechnet und im Gradienten bzw. zur Bewertung bei ES im Training verwendet. In diesen Messungen erbringen beide Algorithmen nahezu identische Klassifikationsergebnisse. Insgesamt verhindert der drastisch ansteigende Rechenaufwand bei den Sprachdaten die Verarbeitung von größeren Vokabularien und langen Wörtern durch ES. Aus der Beschränkung der Populationsgröße durch die vorhandene Rechnerkapazität resultierte eine nichtoptimale Anpassung von Selektionsdruck, Mutationsrate und Populationsverteilung im Suchraum. Insbesondere erweist sich die globale Anpassung der Strategieparameter bei den vergrößerten Populationen als problematisch. Weitere Untersuchungen an ES mit Strategien zur Selbstadaption dieser Parameter bieten sich daher für zukünftige Forschung an.
The deconfinement transition region between hadronic matter and quark-gluon plasma is studied for finite volumes. Assuming simple model equations of state and a first order phase transition, we find that fluctuations in finite volumes hinder a sharp separation between the two phases around the critical temperature, leading to a rounding of the phase transition. For reaction volumes expected in heavy ion experiments, the softening of the equation of state is reduced considerably. This is especially true when the requirement of exact color-singletness is included in the QGP equation of state.
A self-consistent relativistic integral-di erential equation of the Boltzmann- Uehling-Uhlenbeck-type for the N*(1440) resonance is developed based on an effective Lagrangian of baryons interacting through mesons. The closed time-path Green s function technique and semi-classical, quasi-particle and Born approxima- tions are employed in the derivation. The non-equilibrium RBUU-type equation for the N*(1440) is consistent with that of nucleon s and delta s which we derived before. Thus, we obtain a set of coupled equations for the N,Delta and N*(1440) distribution functions. All the N (1440)-relevant in-medium two-body scattering cross sections within the N,Delta and N*(1440) system are derived from the same effective Lagrangian in addition to the mean field and presented analytically, which can be directly used in the study of relativistic heavy-ion collisions. The theoreticalprediction of the free pp - pp* (1440) cross section is in good agreement with the experimental data. We calculate the in-medium N+N - N+N* , N* +N - N+N and N*+N - N* +N cross sections in cold nuclear matter up to twice the nuclear matter density. The influence of different choices of the N* N* coupling strengths, which can not be obtained through fitting certain experimental data, are discussed. The results show that the density dependence of predicted in-medium cross sections are sensitive to the N* N* coupling strengths used. An evident density dependence will appear when a large scalar coupling strength of g^(sigma) N*N* is assumed. PACS number(s): 24.10.Cn; 25.70.-z; 21.65.+f
Preliminary experimental data for particle number ratios in the collisions of Au+Au at the BNL AGS (11A GeV/c) and Pb+Pb at the CERN SPS (160A GeV/c) are analyzed in a thermodynamically consistent hadron gas model with excluded volume. Large values of temperature, T = 140 185 MeV, and baryonic chemical potential, µb = 590 270 MeV, close to the boundary of the quark-gluon plasma phase are found from fitting the data. This seems to indicate that the energy density at the chemical freezeout is tremendous which would be indeed the case for the point-like hadrons. However, a self-consistent treatment of the van der Waals excluded volume reveals much smaller energy densities which are very far below a lowest limit estimate of the quark-gluon plasma energy density. PACS number(s): 25.75.-q, 24.10.Pa
We study the bremsstrahlung of virtual omega mesons due to the collective deceleration of nuclei at the initial stage of an ultrarelativistic heavy ion collision. It is shown that electromagnetic decays of these mesons may give an important contribution to the observed yields of dileptons. Mass spectra of e+e and µ+µ pairs produced in central Au+Au collisions are calculated under some simplifying assumptions on the space time variation of the baryonic current in a nuclear collision process. Comparison with the CERES data for 160 AGev Pb+Au collisions shows that the proposed mechanism gives a noticeable fraction of the observed e+e pairs in the intermediate region of invariant masses. Sensi tivity of the dilepton yield to the in medium modification of masses and widths of vector mesons is demonstrated.
Collective bremsstrahlung of vector meson fields in relativistic nuclear collisions is studied within the time dependent Walecka model. Mutual deceleration of the colliding nuclei is described by introducing the e ective stopping time and average rapidity loss of baryons. It is shown that electromagnetic decays of virtual ω mesons produced by bremsstrahlung mechanism can provide a substantial contribution to the soft dilepton yield at the SPS bombarding energies. In particular, it may be responsible for the dilepton enhancement observed in 160 AGev central Pb+Au collisions. Suggestions for future experiments to estimate the relative contribution of the collective mechanism are given.
Dielectron mass spectra are examined for various nuclear reactions recently measured by the DLS collaboration. A detailed description is given of all dilepton channels included in the transport model UrQMD 1.0, i.e. Dalitz decays of π, η, ω, ή mesons and of the (1232) resonance, direct decays of vector mesons and pn bremsstrahlung. The microscopic calculations reproduce data for light systems fairly well, but tend to underestimate the data in pp at high energies and in pd at low energies. These conventional sources, however, cannot explain the recently reported enhancement for nucleus-nucleus collisions in the mass region 0.15GeV ≤ Me+e- ≤ 0.6GeV. Chiral scaling and ω meson broadening in the medium are investigated as a source of this mass excess. They also cannot explain the recent DLS data.
The transverse momentum distribution of prompt photons coming from the very early phase of ultrarelativistic heavy ion collisions for the RHIC and LHC energies is calculated by means of perturbative QCD. We calculate the single photon cross section (A + B -> gamma + X) by taking into account the partonic sub processes q + q -> gamma + g and q + g -> gamma + q as well as the Bremsstrahlung corrections to those processes. We choose a lower momentum cut-off k0 = 2 GeV separating the soft physics from perturbative QCD. We compare the results for those primary collisions with the photons produced in reactions of the thermalized secondary particles, which are calculated within scaling hydrodynamics. The QCD processes are taken in leading order. Nuclear shadowing corrections, which alter the involved nuclear structure functions are explicitly taken into account and compared to unshadowed results. Employing the GRV parton distribution parametrizations we find that at RHIC prompt QCD-photons dominate over the thermal radiation down to transverse momenta kT ≈ 2 GeV. At LHC, however, thermal radiation from the QGP dominates for photon transverse momenta kT ≤ 5 GeV, if nuclear shadowing effects on prompt photon production are taken into account.