Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
HK: Fachverband Physik der Hadronen und Kerne
HK 32: Heavy-Ion Collisions and QCD Phases VI
HK 32.2: Vortrag
Dienstag, 12. März 2024, 18:00–18:15, HBR 62: EG 03
Spectral Reconstruction with Gaussian Process Regression — •Jonas Turnwald1, Julian M. Urban2,3, and Nicolas Wink1 — 1Institut für Kernphysik, Technische Universität Darmstadt — 2Center for Theoretical Physics, Massachusetts Institute of Technology — 3The NSF AI Institute for Artificial Intelligence and Fundamental Interactions
In order to understand the dynamics of QCD, knowledge about non-perturbative real-time correlation functions is necessary. These correlation functions can be accessed by reconstructing the spectral function from Euclidean data. Since this is an ill-conditioned problem, sophisticated techniques are necessary to ensure reliable results.
In this talk, we will present a method facilitating Gaussian Process regression for reconstructing spectral functions. Additionally, we will introduce a Python package that allows for an immediate implementation of this method for different linear inverse problems.
As an illustration, we present the reconstruction of different spectral functions and compute the thermal photon rate of the quark-gluon plasma.
Keywords: QCD; Spectral functions; Gaussian Process; Quark-gluon plasma; Software package