Freiburg 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

A: Fachverband Atomphysik

A 33: Poster VI

A 33.11: Poster

Thursday, March 14, 2024, 17:00–19:00, Tent A

Towards machine learning optimized time-averaged potentials to generate a Bose-Einstein condensate — •Max Schlösinger¹, Victoria Henderson¹, Simon Kanthak¹, Oliver Anton¹, Elisa Da Ros¹, and Markus Krutzik^{1, 2} — ¹Humboldt-Universität zu Berlin, Institute of Physics & IRIS Adlershof, Newtonstraße 15, 12489 Berlin — ²Ferdinand-Braun-Institut gGmbH, Leibniz-Institut für Höchstfrequenztechnik, Gustav-Kirchhoff-Strasse 4, 12489 Berlin

Time-averaged potentials (TAPs) are a versatile tool for the generation and manipulation of ultracold atom clouds. In order to fully take advantage of this techniques, we investigate machine learning routines with a setup based on acousto-optic deflectors.

Our aim is to mitigate non-linearities in the electro-optical system and effects due to frequency modulation which restricts predictability of shape and smoothness as well as to counteract temporal and spatial instabilities. In particular we focus on identifying the most suitable fitness function associated with the optimisation of different optical potential geometries using images based on a CCD camera.

In the future, we would like to rely on TAP’s capabilities to improve the evaporative cooling routine and to enhance the efficiency of a ⁸⁷Rb Bose-Einstein condensate-based quantum memory.

This work is supported by the German Space Agency (DLR) with funds provided by the Federal Ministry for Economic Affairs and Climate Action (BMWK) under grant numbers No. 50WM2247.

Keywords: Time-averaged potentials; Machine learning