Quantum Information & Computation

Without entanglement, MBQC cannot work. But with little entanglement, it can [3]. At the same time, too much entanglement can be equally detrimental as no entanglement [4]. So, what does constitute a good resource for MBQC? This is a rather vexing question, and there is still no complete answer. However, over the years, we have made leaps and bounds towards a unifying answer [5, 6]. In particular, we have furthered a fascinating connection to condensed matter physics: certain phases of quantum matter can be used as universal resources for MBQC [7, 8, 9]. Specifically, we found that one unifying link are is provided by quantum cellular automata, which further a completely new perspective on the almost 20 year old MBQC. We are excited to explore this connection with many of our collaborators all over the world, from Vancouver to Berlin.

MBQC is a fascinating framework where a quantum computation is realized by simply performing single-qubit measurements on an initially entangled state. This new approach makes MBQC particularly suitable for certain technological platforms such as photons [10, 11] or trapped ions [12]. Considering the ongoing research on resources for MBQC, one may even find certain phases of matter [8, 9] which occur naturally in certain experiments or materials to facilitate MBQC. We are always happy to collaborate with experimentalists and discuss contemporary research and novel implementations.

The cluster state [2] and, more generally, graph states [13], form the basis of standard MBQC. Graph states have surprising and useful properties and provide a playground for the study of entanglement [14]. This line of research has had a remarkable impact on quantum communication, and facilitated significant improvements in various protocols such as entanglement purification and quantum repeaters. For instance, we demonstrated a significant increase in the robustness of these protocols against typical sources of noise. Our goal is not only to develop (special purpose) small-scale quantum machines for practical application in quantum communication [15] but also to facilitate the development of large-scale quantum networks [16].

For more details, see [Quantum Networks].

Resources for MBQC such as the cluster state can be defined for arbitrary graphs. Interestingly, the specific layout of the underlying graph carries meaning and facilitates different computations or communications scenarios [17]. Analogously, Projective Simulation is a model for artificial intelligence [18] where decision making is modelled by a memory-network described by a weighted graph. We are working towards an integration and further development of the two paradigms to design a quantum AI based on MBQC.

Machine learning has proven to be a powerful tool that already permeates our every-day life. We want to leverage this technology for quantum information science. Indeed, due to its flexibility and broad applicability, MBQC may serve as an optimal playground to test the capabilities of artificial intelligence in quantum physics. In this endeavor, we have already shown that reinforcement learning can help develop quantum communication protocols and we are excited to see where else this development will lead us!

For more details, see [Artificial Intelligence and Science].