Theory Seminars WS 19/20


Thu 10.10.2019, 16.30 h

J. Thompson (Uni Heidelberg)

(Machine) Learning to be Unsure

Machine learning methods are becoming ubiquitous in particle physics. However, many of these techniques
suffer from overconfidence and a lack of associated uncertainties. In this talk, I will discuss how Bayesian neural networks can address these problems and help us to understand model uncertainties in a machine learning context. In order to do this, I will first explore how these uncertainties can be understood in a frequentist framework, before considering how the Bayesian neural network responds to statistical and systematic uncertainties.


Thu 17.10.2019, 16.30 h

S. Paßehr (RWTH Aachen)

Towards high-precision predictions for Higgs masses and decays in theories beyond the Standard Model

The particle with a mass of about 125GeV that has been discovered at the LHC has properties that, within the current theoretical and experimental uncertainties, leave plenty of room to interpret it as part of an extended Higgs sector in a theory beyond the Standard Model. Supersymmetric models are particularly interesting, since different problems of the Standard Model could be solved at the same time. In such theories, the Higgs mass is a predicted quantity, and in the most popular supersymmetric model, the MSSM, large radiative corrections occur. In order to be competitive with the experimental uncertainty, high-precision predictions are necessary that entail higher-order calculations. In view of feature collider experiments, the endeavor for high-precision predictions should also be extended to the decay widths of Higgs bosons. Apart from the MSSM, only a few other models have received attention. However, especially due to the absence of signs for new physics at the LHC, a model-agnostic approach can be pursued. This leads to generic calculations in general renormalizable theories.
I will discuss recent progress for the prediction of the Higgs-boson spectrum. At first, I will focus on the MSSM, and later I will show the possibilities of generic calculations. Finally, I want to address some properties of higher-order corrections to the decay widths of Higgs bosons.


Thu 24.10.2019, 16.30 h

K. Ng (GRAPPA, Amsterdam)

High-energy Gamma Rays and Neutrinos from the Sun

I will discuss recent results on gamma-ray observations of the Sun with Fermi, which revealed many interesting and surprising features. These gamma rays are expected to be produced by hadronic interactions between cosmic rays and the solar atmosphere. The high flux of gamma rays observed from the Sun requires a large boost of gamma-ray production by some mechanism, which is likely related to solar magnetic fields. Our new results include the first resolved image of the Sun and a mysterious dip in the spectrum between 30-50 GeV. In particular, we also find that the solar gamma-ray spectrum during solar minimum is hard (~E−2.1) and reaches at least 200 GeV. This suggests that ground-based experiments like HAWC and LHAASO will be important for probing the Sun at the TeV regime. Understanding solar gamma rays is crucial for predicting the solar atmospheric neutrino flux, which can potentially be detected by IceCube/KM3NeT, and is important for many aspects of solar dark matter searches.


Thu 31.10.2019, 16.30 h

M. Mühlleitner (KIT)

The Higgs Boson - A Dissection Tool for New Physics

Despite obvious flaws of the Standard Model of particle physics no direct sign of New Physics has been discovered so far. On the contrary, the discovered Higgs boson behaves very Standard-Model-like. With the increasing mass scale of the exclusion limits on New Physics, the Higgs boson itself becomes increasingly important as tool in the search for physics beyond the Standard Model. I will discuss several examples where New Physics might reveal itself indirectly in the Higgs sector.


Thu 14.11.2019, 16.30 h

L. Magnea (U. Torino)

Threshold logarithms beyond leading power
Universality and resummation

Resummations of threshold logarithms at leading power in the threshold variable have a long history and a vast range of phenomenological applications. In recent years, several groups have studied the possibility of extending the resummation technology to next-to-leading power (NLP) in the threshold expansion. I will review the theoretical motivations that suggest this generalisation, discuss some of the technical difficulties, and present recent results on the resummation of leading logarithms at NLP for processes with colour-singlet final states.


Thu 21.11.2019, 16.30 h

M. Bustamante (NBI, Copenhagen)



Thu 28.11.2019, 16.30 h

C. Garcia-Cely (DESY Hamburg)



Thu 05.12.2019, 16.30 h

D.J.E. Marsh (Uni Göttingen)



Thu 12.12.2019, 16.30 h

M. Wiesemann (CERN und MPI München)



Thu 19.12.2019, 16.30 h

R. Teyssier (Uni Zürich)



Thu 09.01.2020, 16.30 h

A. Fialkov (University of Sussex)



Thu 16.01.2020, 16.30 h

M. Diehl (DESY)



Thu 23.01.2020, 16.30 h

A. Font-Ribera (UC London)



Thu 30.01.2020, 16.30 h

C. Sturm (U. Würzburg)