Research areas

Entanglement, which Erwin Schrödinger coined the characteristic treat of quantum mechanics, is the resource that renders many quantum technologies superior to their classical counterparts. At the same time entanglement is at the heart of many physical phenomena. For example, it explains why interacting quantum systems, even when perfectly isolated from their environment, can relax to thermal equilibrium. Thus, techniques for detecting and quantifying entanglement based on experimental data are direly needed. We develop such techniques taking into account the concrete measurement capabilities of quantum simulation platforms including cold atoms and photonic systems.

In physics, the art of simulating can be thought of as the capability of imitating - from Latin simulare - the behaviour of a given physical system by using an ad-hoc device.

It is remarkable how one can in principle tackle a physical problem by changing the language in which it is naturally expressed.

For instance, it is possible to rephrase the time evolution of a quantum field just in terms of optical elements! This is a major research aspect within our project.

Our aim is to better comprehend the structure of different quantum field theories by means of photonic platforms and to use the latter ones to explore fundamental aspects, such as the behaviour of the entanglement entropy. At the same time, we use Machine Learning techniques to develop device-independent methods that allow computing the fingerprint of a state, for instance its non-classicality, a feature lying at the heart of quantum information theory.

Quantum measurements are distinctly different from classical ones in that they inevitably affect the state of the system to some degree. This leads to subtleties when trying to characterize quantum states from experimental data, a highly relevant task for quantum technologies. We design measurement strategies that allow characterizing quantum state in a maximally sample efficient way using a feedback loop. We also investigate how to correct experimental errors occurring during the readout process though post-processing of the measurement data.

Disorder is present in many natural systems from glasses and amorphous solids to social networks with seemingly random connections. Such systems often show surprising emergent dynamical effects. In the realm of quantum many-body systems disorder can lead to hierarchical relaxation and glassy behavior. We study relaxation dynamics and transport in quantum spin systems where the inter-particle interactions are to some degree random. Such systems can be realized by cold atoms excited to Rydberg states. In these highly excited states the atoms interact via strong dipole-dipole interactions. We model these systems numerically and try to come up with new ways for experimentally probing their properties.

Artificial neural networks have proven extremely successful for machine learning tasks such as computer vision and speech recognition. Specifically, generative models can be trained to approximate probability distributions based on data samples. Quantum states are represented by high-dimensional probability distributions, inviting the use of generative models for finding efficient state representations. We use this approach to develop numerical tools for calculating the time evolution of quantum many-body states and to do quantum state tomography. Furthermore, we use supervised learning to efficiently predict entropic quantities from measured data.