List of previous students

Nicola Orlando
Faten Hariri
Emma Kuwertz
Spyridon Argyropoulo
Sabrina Sacerdoti
Simone Amoroso
Jesper Roy Christiansen
Nathan Hartland
Christian Roehr
Benjamin Watt
Philip Ilten
Nishita Desai
Sercan Sen
Miroslav Myska
Sudha Ahuja
Holger Schulz
Avi Gershan
Aleksander Kusina
Magdalena Slawinska
Flavia Dias
Kenneth Wraight
Irais Bautista Guzman
Sparsh Navin
Paolo Francavilla
Riccardo Di Sipio
Seyi Latunde-Dada
Devdatta Majumder
Martijn Gosselink
Christopher Bignamini
Marek Schönherr
Michal Deak
Noam Hod
Florian Bechtel
Jonathan Ferland
Manuel Bähr
Alexander Flossdorf
Piergiulio Lenzi

Colour reconnection (CR) is expected to occur at a significant rate at the LHC, and to have a non-negligible impact on event properties. The reason is that the high density of particle production points to a high rate of multiparton interactions which, in combination with parton showers, gives a large number of perturbatively defined partons emerging from typical proton-proton collisions. These partons give many force fields - strings or clusters - that are stretched between matching colour and anticolour charges. The fields overlap spatially, and therefore are likely to interact in nontrivial ways.

These interactions lie beyond the domain of perturbation theory, or any other first-principles approach. Instead simple models have been developed and implemented in MC generators. Since the standard generators operate with an infinity of colours, rather than three, a reassignment among matching colour pairs could be expected already from colour ambiguities. However, the rising average transverse momentum as a function of multiplicity for charged particles points to a dynamical component of the reconnections, where the string length (or cluster mass) tends to be reduced.

Colour reconnection can also have other effects on event properties. Notably the issue has been raised that the precision of top mass measurements could be influenced. This should be viewed in the light of ever more precise top mass values being published, now with quoted total errors below 1 GeV, whereof 0.1 - 0.3 GeV is attributed to CR. It is not always clear how these values have been arrived at. One shortcoming is also that, while the old Pythia 6 generator contained a few CR options, the current Pythia 8 one only contained one model (with a tunable strength parameter, and a possibility to let top quarks decay happen before or after the CR step).

During the three months spent in Lund, several new CR models were developed and implemented in Pythia 8, and tested in various ways. The objective was not to find the best possible model, but rather to produce a wide spread of them, to study various worst-case scenarios. In total some ten new models were introduced. Some models offer a complete replacement of the old CR model, while others retain the old CR behaviour for the underlying event and only set up new rules for how the top decay products should interact with this underlying event.

These models where compared with various LHC data sets, making good use of the RIVET package developed within MCnet. Tests related to underlying-event properties in general, but also to more specific observables such as jet shapes in top events. Models with very high mass shifts from CR effects tend to give too broad jets, and therefore the CR strength parameter in these models had to be tuned down to acceptable levels.

The top mass itself was studied by writing a Rivet analysis that in a semi-realistic fashion implements the typical experimental cuts used at the LHC, but does not aim to reproduce the sophistication of a full-fledged experimental analysis. This should be good enough for an estimate of the relative top-mass shift between different CR models.

The studies do not support the lower range of top mass CR errors in the literature, but rather suggests uncertainties of the order of 0.5 GeV. The uncertainty band is rather asymmetric, however, it being easy to obtain lower top mass values than the default CR model, but difficult to obtain higher. This is related to the default CR string length already being fairly minimal and therefore difficult to reduce further, whereas it is much simpler to increase it.

The studies resulted in an article, which now has been published: Effects of color reconnection on ttbar final states at the LHC, S. Argyropoulos and T. Sjöstrand, JHEP 1411 (2014) 043 [arXiv:1407.6653 [hep-ph]]