Hannover 2020 – wissenschaftliches Programm
Die DPG-Frühjahrstagung in Hannover musste abgesagt werden! Lesen Sie mehr ...
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
Q: Fachverband Quantenoptik und Photonik
Q 8: Quantum Information (Concepts and Methods) II
Q 8.4: Vortrag
Montag, 9. März 2020, 14:45–15:00, e001
Representing an experimental two photon state with a neural network — •Marcel Neugebauer1, Martin Gärttner2, Laurin Fischer1, Alexander Jäger1, Selim Jochim1, and Matthias Weidemüller1 — 1Physikalisches Institut, Universität Heidelberg — 2Kirchhoff-Institut für Physik, Universität Heidelberg
Neural networks are mathematical models on which parameter optimization can be done in an efficient way, so that immense numbers of parameters can be handled. High dimensional optimization is also crucial to solve important problems in many body quantum physics. In particular quantum state tomography is a problem that scales exponentially in the measurement cost and in numerical optimization cost. New approaches to tackle this via the neural network representation of quantum states emerged recently. In this talk an ansatz is discussed in which a probability distribution over an informationally complete measurement is represented with a restricted Boltzmann machine. We represent the quantum state of a twin photon source with this technique and test its prediction capabilities against predictions of a maximum likelihood density matrix and actual measurement.