Christoph Ames Successful PhD Defense
We congratulate Christoph for his successful thesis defense!
24.09.2024
We congratulate Christoph Ames for successfully passing his PhD exam on September 24th 2024.
Christoph investigated the Matrix Element Method (MEM) in combination with Neural Networks for the application in the search for Higgs self interaction processes. The MEM evaluates the four momenta of the final state particles according to the Hamilton operator of the expected process. The MEM can be interpreted as a probability that the observed final state conforms with the expected process. And since the Hamilton operator contains all physical details of the considered process the MEM can be used to identify the searched for process.
Since the calculation of the MEM is very computing intensive and since it needs to be repeated for every measured final state, Christoph trained a NN to predict the probability which the MEM assigns to the final state. Once the NN is trained, the evaluation of the MEM is much faster. In fact, Christoph could show that the calculation is sped up by a factor of over 700.
To establish and to challenge the concept, Christoph considered two competing processes with identical final states, namely H → HH → bb WW → bb qq lν against Z → ZH → bb WW → bb qq lν. These two processes have almost identical kinematics for the final state particles due to the very similar mass of the Z and the H bosons. These two processes are experimentally very difficult to separate by ordinary cut-based analyses. Using the MEM and the NN, however, Christoph could proove that a clear separation between these two processes is possible.
Christoph's results will be of high relevance for the very challenging measurement of the Higgs self-coupling parameter. Since the cross section for the Higs self-interaction processes is very tiny, this measurement is expected only at the high luminosity LHC (HL-LHC). However, the successful application of MEM and NN might boost the sensitivity such that some more stringent limits might be set already before the start of HL-LHC using the data of the current runs at LHC.