Do physicists stop searches too early? A remote-science, optimization landscape investigation.

The manuscript for the results of our Alice Challenge is now submitted for review and available on the arxiv.

Despite recent advances driven by machine learning algorithms, experts agree that such algorithms are still often unable to match the experience-based and intuitive problem solving skills of humans in highly complex settings. Recent studies have demonstrated how the intuition of lay people in citizen science games [1] and the experience of fusion-scientists [2] have assisted automated search algorithms by restricting the size of the active search space leading to optimized results. Humans, thus, have an uncanny ability to detect patterns and solution strategies based on observations, calculations, or physical insight. Here we explore the fundamental question: Are these strategies truly distinct or merely labels we attach to dierent points in a high dimensional continuum of solutions? In the latter case, our human desire to identify patterns may lead us to terminate search too early. We demonstrate that this is the case in a theoretical study of single atom transport in an optical tweezer, where more than 200,000 citizen scientists helped probe the Quantum Speed Limit [1]. With this insight, we develop a novel global entirely deterministic search methodology yielding dramatically improved results. We demonstrate that this \bridging" of solution strategies can also be applied to closed-loop optimization of the production of Bose-Einstein condensates. Here we nd improved solutions using two implementations of a novel remote interface. First, a team of theoretical optimal control researchers employ a Remote version of their dCRAB optimization algorithm (RedCRAB), and secondly a gamied interface allowed 600 citizen scientists from around the world to participate in the optimization. Finally, the \real world" nature of such problems allow for an entirely novel approach to the study of human problem solving, enabling us to run a hypothesis-driven social science experiment \in the wild". (09/2017)

Leaderboard Effects on Player Performance in a Citizen Science Game

Accepted for ECGBL 2017

Quantum Moves is a citizen science game that investigates the ability of humans to solve complex physics challenges that are intractable for computers. During the launch of Quantum Moves in April 2016 the game's leaderboard function broke down resulting in a "no leaderboard" game experience for some players for a couple of days (though their scores were still displayed). The subsequent quick fix of an all-time Top 5 leaderboard, and the following long-term implementation of a personalized relative-position (infinite) leaderboard provided us with a unique opportunity to compare and investigate the effect of different leaderboard implementations on player performance in a points-driven citizen science game.
All three conditions were live sequentially during the game's initial influx of more than 150.000 players that stemmed from global press attention on Quantum Moves due the publication of a Nature paper about the use of Quantum Moves in solving a specific quantum physics problem. Thus, it has been possible to compare the three conditions and their influence on the performance (defined as a player's quality of game play related to a high-score) of over 4500 new players. These 4500 odd players in our three leaderboard-conditions have a similar demographic background based upon the time-window over which the implementations occurred and controlled against Player ID tags. Our results placed Condition 1 experience over condition 3 and in some cases even over condition 2 which goes against the general assumption that leaderboards enhance gameplay and its subsequent overuse as a an oft-relied upon element that designers slap onto a game to enhance said appeal. Our study thus questions the use of leaderboards as general performance enhancers in gamification contexts and brings some empirical rigor to an often under-reported but overused phenomenon. (07/2017)    

Knowledge Formation and Inter-Game Transfer With Classical and Quantum Physics

Published as Work in Progress Paper at ECGBL ’16

In order to facilitate an intuitive understanding of classical physics concepts we have developed Potential Penguin - a game where players manipulate the landscape around a sliding penguin in order to control its movement. The learning goal of Potential Penguin is to familiarize players with kinetic energy and potential energy - the energies associated with movement and position in the landscape respectively. The game levels introduce the concepts one by one, as players are tasked with sliding the penguin through a landscape towards a specific location, while keeping the velocity under control. When the player manipulates the landscape, the potential energy of the penguin is changed, which determines the penguin's movement. To build a strong connection between theory and game the analytical expressions for kinetic and potential energy are displayed during play with font sizes continually growing and shrinking according to changes in each energy type. With Potential Penguin we hope to study whether visualizing the amount of kinetic and potential energy through visible mathematical expressions generates a connection between the intuitive actions taken in the game and the underlying physics concepts. The knowledge about kinetic and potential energy gained with Potential Penguin can also be used to understand most of the physics behind the citizen science game Quantum Moves, which has the goal of building a working quantum computer. The two games share the principle of the core interaction - manipulating the potential-energy landscape. We aim to investigate whether a proficiency and understanding of Potential Penguin predicts a better performance in Quantum Moves and a deeper understanding of the quantum physics behind that game.  (08/2016)

Measurement-enhanced determination of BEC phase diagrams

Is now submitted for review and available on the arxiv.

We demonstrate how dispersive atom number measurements during evaporative cooling can be used for enhanced determination of the non-linear parameter dependence of the transition to a Bose-Einstein condensate (BEC). Our analysis demonstrates that conventional averaging of shot-to-shot fluctuations introduces systematic errors and reduces precision in comparison with our method. We furthermore compare in-situ images from dispersive probing of a BEC with corresponding absorption images in time-of-flight. This allows for the determination of the transition point in a single experimental realization by applying multiple dispersive measurements. Finally, we explore the continuous probing of several consecutive phase transition crossings using the periodic addition of a focused "dimple" potential.  (07/2016)

Manipulating matter waves in an optical superlattice

Published in Phys. Rev. A with the group of Gabriele De Chiara, Belfast

We investigate the potential for controlling a noninteracting Bose-Einstein condensate loaded into a one-dimensional optical superlattice. Our control strategy combines Bloch oscillations, originating from accelerating the lattice, with time-dependent control of the superlattice parameters. We investigate two experimentally viable scenarios, very low and very high potential depths, in order to gain a better understanding of matter wave control available within the system. Multiple lattice parameters and a versatile energy band structure allow us to obtain a wide range of control over energy band populations. Finally, we consider several examples of quantum state preparation in the superlattice structure that may be difficult to achieve in a regular lattice. (12/2016)

Preparation of ultracold atom clouds at the shot noise level

Published in Phys. Rev. Lett. with the group of Jan Arlt

We prepare number stabilized ultracold atom clouds through the real-time analysis of nondestructive images and the application of feedback. In our experiments, the atom number N∼106 is determined by high precision Faraday imaging with uncertainty ΔN below the shot noise level, i.e., ΔN<√N. Based on this measurement, feedback is applied to reduce the atom number to a user-defined target, whereupon a second imaging series probes the number stabilized cloud. By this method, we show that the atom number in ultracold clouds can be prepared below the shot noise level. (08/2016)

Universal three-body physics in ultracold KRb mixtures

Published in Phys. Rev. Lett. with the group of Jan Arlt (Editor’s suggestion) See also Physics Focus story

Ultracold atomic gases have recently become a driving force in few-body physics due to the observation of the Efimov effect. While initially observed in equal mass systems, one expects even richer few-body physics in the heteronuclear case. In previous experiments with ultracold mixtures of potassium and rubidium, an unexpected nonuniversal behavior of Efimov resonances was observed. In contrast, we measure the scattering length dependent three-body recombination coefficient in ultracold heteronuclear mixtures of 39K−87Rb and 41K−87Rb and do not observe any signatures of Efimov resonances for accessible scattering lengths in either mixture. Our results show good agreement with our theoretical model for the scattering dependent three-body recombination coefficient and reestablish universality across isotopic mixtures. (10/2016)

Inferring causality from noisy time series data

Published in COMPLEXIS 2016

Convergent Cross-Mapping (CCM) has shown high potential to perform causal inference in the absence of models. We assess the strengths and weaknesses of the method by varying coupling strength and noise levels in coupled logistic maps. We find that CCM fails to infer accurate coupling strength and even causality direction in synchronized time-series and in the presence of intermediate coupling. We find that the presence of noise deterministically reduces the level of cross-mapping fidelity, while the convergence rate exhibits higher levels of robustness. Finally, we propose that controlled noise injections in intermediate-to-strongly coupled systems could enable more accurate causal inferences. Given the inherent noisy nature of real-world systems, our findings enable a more accurate evaluation of CCM applicability and advance suggestions on how to overcome its weaknesses.  (03/2016)

Manipulation of collective quantum states in Bose-Einstein condensates by continuous imaging

Published in Phys. Rev. A with the group of Klaus Mølmer

We develop a Gaussian state treatment that allows a transparent quantum description of the continuous, nondestructive imaging of and feedback on a Bose-Einstein condensate. We have previously demonstrated [A. C. J. Wade et al., Phys. Rev. Lett. 115, 060401 (2015)] that the measurement backaction of stroboscopic imaging leads to selective squeezing and entanglement of quantized density oscillations. Here, we investigate how the squeezing and entanglement are affected by the finite spatial resolution and geometry of the probe laser beam and of the detector and how they can be optimized. (02/2016)

Virtual learning environment for interactive engagement with advanced quantum mechanics

Published in Phys. Rev. Phys. Education

A virtual learning environment can engage university students in the learning process in ways that the traditional lectures and lab formats cannot. We present our virtual learning environment StudentResearcher, which incorporates simulations, multiple-choice quizzes, video lectures, and gamification into a learning path for quantum mechanics at the advanced university level. StudentResearcher is built upon the experiences gathered from workshops with the citizen science game Quantum Moves at the high-school and university level, where the games were used extensively to illustrate the basic concepts of quantum mechanics. The first test of this new virtual learning environment was a 2014 course in advanced quantum mechanics at Aarhus University with 47 enrolled students. We found increased learning for the students who were more active on the platform independent of their previous performances. (04/2016)

Play or science?: a study of learning and framing in crowdscience games

Published in Well Played 4(2), 30 (2015)

Crowdscience games may hold unique potentials as learning opportunities compared to games made for fun or education. They are part of an actual science problem solving process: By playing, players help scientists, and thereby interact with real continuous research processes. This mixes the two worlds of play and science in new ways. During usability testing we discovered that users of the crowdscience game Quantum Dreams tended to answer questions in game terms, even when directed explicitly to give science explanations.We then examined these competing frames of understanding through a mixed correlational and grounded theory analysis. This essay presents the core ideas of crowdscience games as learning opportunities, and reports how a group of players used "game", "science" and "conceptual" frames to interpret their experience. Our results suggest that oscillating between the frames instead of sticking to just one led to the largest number of correct science interpretations, as players could participate legitimately and autonomously at multiple levels of understanding. (10/2015)

Non-Gaussian distribution of collective operators in quantum spin chains

Published in New Journ. Phys.

We numerically analyse the behavior of the full distribution of collective observables in quantum spin chains. While most of previous studies of quantum critical phenomena are limited to the first moments, here we demonstrate how quantum fluctuations at criticality lead to highly non-Gaussian distributions. Interestingly, we show that the distributions for different system sizes collapse on the same curve after scaling for a wide range of transitions: first and second order quantum transitions and transitions of the Berezinskii–Kosterlitz–Thouless type. We propose and analyse the feasibility of an experimental reconstruction of the distribution using light–matter interfaces for atoms in optical lattices or in optical resonators. (10/2016)

Exploring the Quantum Speed Limit with Computer Games

Published in Nature See also Nature News and Views

Humans routinely solve problems of immense computational complexity by intuitively forming simple, low-dimensional heuristic strategies1, 2. Citizen science (or crowd sourcing) is a way of exploiting this ability by presenting scientific research problems to non-experts. ‘Gamification’—the application of game elements in a non-game context—is an effective tool with which to enable citizen scientists to provide solutions to research problems. The citizen science games Foldit3, EteRNA4 and EyeWire5 have been used successfully to study protein and RNA folding and neuron mapping, but so far gamification has not been applied to problems in quantum physics. Here we report on Quantum Moves, an online platform gamifying optimization problems in quantum physics. We show that human players are able to find solutions to difficult problems associated with the task of quantum computing6. Players succeed where purely numerical optimization fails, and analyses of their solutions provide insights into the problem of optimization of a more profound and general nature. Using player strategies, we have thus developed a few-parameter heuristic optimization method that efficiently outperforms the most prominent established numerical methods. The numerical complexity associated with time-optimal solutions increases for shorter process durations. To understand this better, we produced a low-dimensional rendering of the optimization landscape. This rendering reveals why traditional optimization methods fail near the quantum speed limit (that is, the shortest process duration with perfect fidelity)7, 8, 9. Combined analyses of optimization landscapes and heuristic solution strategies may benefit wider classes of optimization problems in quantum physics and beyond. (04/2016)    

Time-limited optimal dynamics beyond the quantum speed limit
Gajdacz, M., Das, K. K., Arlt, J., Sherson, J. F., & Opatrný, T. 
Physical review

Squeezing and entanglement of density oscillations in a Bose-Einstein condensate
A. Wade, J. F. Sherson, and K. Mølmer
Physical Review Letters

Getting humans to do quantum optimization: User acquisition, engagement and early results from the citizen cyberscience game quantum moves
Lieberoth, A., Pedersen, M. K., Marin, A. C., Planke, T., & Sherson, J. F. 
Human Computation