Skip to main content
European Commission logo print header

Measurement-based dynamic control of mesoscopic many-body systems

Periodic Reporting for period 4 - MECTRL (Measurement-based dynamic control of mesoscopic many-body systems)

Reporting period: 2019-11-01 to 2020-04-30

Quantum control is an ambitious framework for steering dynamics from initial states to arbitrary desired final states. It has over the past decade been used extensively and with immense success for control of low- dimensional systems in as varied fields as molecular dynamics and quantum computation. Only recently have efforts been initiated to extend this to higher-dimensional many-body systems. Most generic quantum control schemes to date, however, put quite heavy requirements on the controllability of either the system Hamiltonian or a set of measurement operators. This will in many realistic scenarios prohibit an efficient realization.
Within this project, we will develop a new quantum control scheme, which is minimalistic on system requirements and therefore ideally suited for the efficient and reliable optimization of many-body control problems. The fundamentally new ingredient is the total quantum evolution dictated by a combination of fixed many-body time evolution and the precise knowledge of the quantum back-action due to repeated quantum non-destruction (QND) measurements of a single projection operator.
The scheme will be applied to the control of bosons in optical lattices and subsequently experimentally realized. Recent experimental advances in single site manipulation in optical lattices had enabled the high fidelity preparation of mesoscopic samples of atoms (5-50) and using high resolution holographic control of the local light potential this subsystem can be isolated from the remaining lattice. This forms an ideal starting point for many-body quantum control, and we will i.a. demonstrate engineering of quantum phase transitions and preparation of highly non-classical Schödinger cat states.
Finally, using the results from an online graphical interface allowing users of the internet to solve quantum problems we will attempt to build next-generation optimization computer algorithms with a higher level of cognition built in.
Work performed in the project has been separated on four main areas, two experimental and two theoretical.
1: Theoretical work on comparing the results of normal users of the internet playing a quantum game on the platform www.scienceathome.org with state of the art optimization algorithms. Results from 10,000 online players were used in the comparison. The analysis clearly demonstrated that humans exhibit a superior ability to find promising solutions to the complex optimization challenge relative to current algorithms. This finding was published in Nature in April 2016 and was subsequently featured in more than 100 media outlets around the world. Since then more than 150,000 additional citizens have played the games and contributed a wealth of new data currently being analyzed. The goal is to extract the methodologies of the players and implement them into novel optimization algorithm. Initial results are very promising and will be published soon.
2: Theoretical work to investigate the degree to which non-destructive measurements of complex interacting many-body systems can be used for control of these systems. Initial work in the group focused on demonstrating the use of combining projective measurements with natural time evolution of the system for general control. Since then work has focused on generalizing this to continuous weak measurements which should be much more realistic to implement experimentally. We are currently investigating the quantum phase transition between a superfluid and a Mott insulating state in optical lattices and the dynamics induced by adding continuous probing.
3: Experimental work on implementing non-destructive measurements for the probing and control of large ensembles of ultra-cold atoms. Within the first period we have achieved two main results. First we have employed a weak non-destructive probe in the process of preparing the ultracold cloud and combined it with active feedback to achieve a suppression of number fluctuations by two orders of magnitude. This extremely well controlled source of ultracold atoms will serve as ideal starting condition for precision measurements in the coming years. Secondly, we have implemented multiple non-destructive images as the phase transition to a Bose-Einstein condensate. With this “movie” of a phase transition crossing we can start asking fundamental questions on the effect of the knowledge we gain onto the phase transition structure. This work lays the foundation for similar investigations to be implemented later in the project on phase transitions in optical lattices.
4: Experimental progress towards single site detection and manipulation in optical lattices. Within this subproject the high-resolution imaging system has been successfully tested and is currently being built into the apparatus. Work has also focused on generating a setup for arbitrary control of the potential landscape of atoms using so-called Digital Micromirror Devices (DMD).
The project aims to push the boundaries of control of quantum systems in two ways. First we explore the potential of quantum measurements for control. As detailed above, both initial experimental and theoretical work clearly validate the potential of this approach. Current work focusses primarily on extended this to the experimental verification in optical lattices which due to the potential for extremely versatile quantum simulation will have great impact.
The second approach is to investigate the inclusion of ordinary citizens in the solution of quantum problems. Our work on quantum games has opened up a new field of research and has already spurred immense interest in the scientific community to extend this approach to other challenges in the field. The impact on the field of quantum optimization and optimization in general will, however, be even greater if we succeed in our current endeavor to extract the learning strategies of the players and implement them in novel autonomous algorithms. This could potentially provide the basis for novel forms of artificial intelligence much more closely mimicking the human ability to solve complex problems using little rather than Big Data. Given the current societal impact of Big Data and AI such a finding could of course have immense impact.