I am the senior postdoc in Prof. Weber's working group, currently working on my habilitation. I support all activities of the group, but am most active on the T2K and DUNE experiments. My main research interest is in statistical methods for neutrino physics experiments and related fields
Contact
Staudingerweg 7, 04-432
Tel.: +49 6131 39-28615
lukas.koch@uni-mainz.de
Research
I develop methods for the solution of "inverse problems", mostly in the context of neutrino cross-section experiments. In general, in inverse problems one tries to learn something about the "cause" of some data, while only the "effect" is visible in the detector. The two are related through detector smearing and efficiency losses. "Undoing" these effects can lead to an inflation of statistical or systematic uncertainties, and generally introduces correlations in the results. These effects need to be handled carefully to avoid introducing biases in the analysis.
I am also interested in the way we deal with systematic uncertainties in our analyses. Systematic uncertainties enter whenever there is some unknown element in our physics modelling, detector performance, or analysis methods. These uncertainties cannot be reduced by simply taking more data, and require a case-by-case treatment specific for the circumstances.
On the experimental side, I am convener of the Near Detector Neutron Interaction working group of T2K. With the recently upgraded ND280 near detector, we are developing analyses to detect and resolve neutrons originating from neutrino interactions in the detector. This will allow us investigate nuclear processes in the interactions in ways that have not been possible before. These processes are the cause of leading systematic uncertainties in neutrino oscillation experiments, so improved measurements are absolutely crucial for the success of future high-precision oscillation experiments like DUNE and Hyper-Kamiokande.
Within DUNE I am working at innovative ways to use the near detector data provide by the PRISM approach. In this approach, the near detector is made movable, allowing us to take data at different angles w.r.t. the beam axis, and thus in different neutrino energy spectra. Using the same detector in different beams allows us to disentangle the effects of beam uncertainties and interactions uncertainties, which usually are very hard to distinguish in a single detector in a single beam. By building linear combinations of the data taking in different fluxes, one can create "virtual measurements" in virtual fluxes that correspond to the same linear combination. These can then be used "simulate" the oscillated flux at the far detector without having to rely on an interaction model, since all recorded events are actual data. Or they can be combined to simulate a narrow energy distribution to measure cross sections at well-defined neutrino energies instead of the usual wide distributions of the beams.
Some publications
- Lukas Koch. Covering unknown correlations in Bayesian priors by inflating uncertainties. Journal of Instrumentation, 21(01):P01040, January 2026. arXiv:2509.11821, doi:10.1088/1748-0221/21/01/p01040.
- Lukas Koch. Hypothesis tests and model parameter estimation on datasets with missing correlation information. Physical Review D, 111(3):033002, February 2025. arXiv:2410.22333, doi:10.1103/physrevd.111.033002.
- Lukas Koch. Post-hoc regularisation of unfolded cross-section measurements. Journal of Instrumentation, 17(10):P10021, October 2022. [Erratum: JINST 20, E05001 (2025)]. arXiv:2207.02125, doi:10.1088/1748-0221/17/10/p10021.
- Lukas Koch and Stephen Dolan. Treatment of flux shape uncertainties in unfolded, flux-averaged neutrino cross-section measurements. Physical Review D, 102(11):113012, dec 2020. arXiv:2009.00552, doi:10.1103/physrevd.102.113012.
- L. Koch. A response-matrix-centred approach to presenting cross-section measurements. Journal of Instrumentation, 14(09):P09013--P09013, sep 2019. arXiv:1903.06568, doi:10.1088/1748-0221/14/09/p09013.
Full lists: ORCID:0000-0002-2966-7461