There are always several topics for Bachelor and Master theses available. A few examples are given below. In case of interest, we will discuss the detailed topics with the candidates.
Please contact the team members for more information.
This project aims at realizing and testing a multi-cell quantum memory in which
several optical pulses on the single photon level can be stored simultaneously. Such
memories can be used in advanced experiments on neuro-inspired photonic
computers , photonic quantum computation/simulation , quantum
Optical memories convert optical pulses traveling at the speed of light into long-lived
stationary excitations in an optical active medium, so called spin waves . Ideally,
this process is reversible and spin waves can be coherently back-converted into
optical pulses. Various schemes for realizing optical memories have been introduced
in the context of quantum communication. Most prominent are the schemes based on
electromagnetically introduced transparency (EIT) or Raman transitions and photon
echo schemes. Both can be implemented in a great variety of physical systems,
where rare-earth doped crystals and warm or cold atomic vapors are most advanced.
For applications alkaline vapor cells at room temperature are favorable, as they
require lower experimental complexity compared to the other systems. See Fig. 1 for
more details. Coherence times of 10 milliseconds up to one minute were observed in
vapor cells with suitable anti-relaxation coatings. A recent experiment with a vapor
cell memory reached a storage time of τ = 1 s. While quantum applications usually
require memories with extreme low readout noise, the figures for this project are
bandwidth and memory capacity. Furthermore, the capability of manipulating optical
waveform shapes, in particular to perform a partial readout, where several optical
pulses can be generated from a single stored spin wave is required. This has not yet
been investigated experimentally in vapor cell memories.
 D. Brunner, et al., Parallel photonic information processing at gigabyte per second
data rates using transient states, Nat. Commun. 4, 1 (2013).
 J. Nunn et al., Phys. Rev. Lett. 110, 133601 (2013).
 N. Sangouard et al., Quantum repeaters based on atomic ensembles and linear
optics, Rev. Mod. Phys. 83, 33 (2011).
 M. Fleischhauer, et al., Dark-State Polaritons in Electromagnetically Induced
Transparency, Phys. Rev. Lett. 84, 5094 (2000).
This project aims at realizing and testing a reconfigurable optical neural network.
Coupling between individual network nodes is realized by spatial light modulators
(SLMs), while an optical non-linearity is realized in form of a Cesium vapor cell
representing a saturable absorber. Using computer controlled feed forward optical
deep neural networks will be realized and applied, e.g. to image recognition.
Classical digital computer architectures are visibly approaching their technological
and physical limits. Thus, there is a growing interest in developing post-digital
computing approaches to overcome these limitations. Besides quantum computers,
approaches that emulate neuromorphic processes represent a very promising
alternative because they mimic the massively parallel, energy-efficient computations
carried out by the human brain. Such computations constitute the building blocks of
the pattern recognition algorithms underpinning the success of machine learning and
artificial intelligence (AI). Optically integrated systems promise 2–3 orders of
magnitude higher energy efficiency compared to today's electronic approaches.
We will realize machine learning with optical neural networks (Fig. 1), in a free-space
bulk optics approach. That is, we want to use light to power machine learning. The
starting point is the realization of a reconfigurable neural network in a tabletop free-
space optical experiment. To this end, the methodology from modern optics is
applied. Spatial light modulars (SLMs) are used to realize linear mappings between
an input and output data (e.g. images). In addition, an atomic Cesium vapor severs
as optical non-linearity. In combination of these, optical neural networks (ONN) will
be realized. The performance of the realized ONN will be first demonstrated on data
classification, i.e. image recognition. Building on the gathered knowledge, a strategy
for all-optical multilayer neural networks shall be developed.
This challenging interdisciplinary project requires a highly motivated student with
background in optical physics, as well as with affinity to computer science.