A coin in your hand always lands either heads or tails: that's how the familiar world works. But a cloud of 370 ultracold atoms can behave differently: it can freeze into a state akin to a coin that is both heads and tails at the same time. A new study shows how to test this experimentally.
The idea rests on comparing two scenarios. In one, the quantum scenario, the system can 'tunnel' between two states, like a ball that passes through a hill rather than rolling over it. In the other (the classical one, like a coin) there are no transitions between states; the system simply freezes. The authors calculated that with a slow change of the external field, the quantum system will make errors less and less often, while the classical one will maintain a constant error rate. The difference between them is huge and measurable.
To test this, you need a setup with a Bose–Einstein condensate — a gas cooled to nearly absolute zero. Calculations show that with currently achievable parameters, the effect survives even with noise: the safety margin is 8.8 times larger than the minimum required. The key secret is the system's symmetry, which suppresses the main source of disturbance.
Such an experiment would serve as a bridge between a world where a coin always lands on one side and a world where it can hover in midair, choosing neither.
as written by the authors arXiv:2604.18638
We derive strict quantitative conditions under which a collective quantum system of N~370 spins exhibits classically forbidden temporal correlations in a spinor Bose-Einstein condensate (BEC). The Lipkin-Meshkov-Glick (LMG) model near its Z_2-breaking quantum critical point supports a mesoscopic superposition a|P> + b|R> of two macroscopic ordered phases (|P>: m_z ~ +m_*; |R>: m_z ~ -m_*) at the Goldilocks crossover N ~ N_c, where the tunnel splitting equals k_B T and macroscopic susceptibility chi ~ N coexists with finite quantum coherence. We establish two quantum-discriminating predictions. P4 (Landau-Zener crossover, proposed discriminator): quantum tunnelling drives P_error -> 0 exponentially with quench time, while the classically non-ergodic system remains kinetically frozen at P_error -> 1 - a parametrically large, computable separation that rests on a specific kinetic foil. P5 (Leggett--Garg inequality violation, strictly model-independent): K_3 > 1 is forbidden by macrorealism for all classical models satisfying non-invasive measurability. A five-level Lindblad simulation yields K_3 ~ 1.32 and dephasing threshold gamma_phi < 0.289 s^{-1} at the BEC target (gamma_phi = 0.05 s^{-1}, N = 370, Gamma/J = 0.95) - a margin of 8.8x above the optimal measurement interval, well within current experimental reach. The threshold has a precise physical origin in the emergent collective Z_2 symmetry: exact parity eliminates the dominant dephasing cross-term and renders the LGI correlator immune to T_1 population mixing without requiring ground-state preparation (2.35x improvement over the naive mean-field estimate); constructive dynamical phase alignment of higher odd-parity states at the benchmark parameters contributes a further 2.47x. All results are reproducible from the provided self-tested Python code.
A cloud of 370 atoms at a temperature of billionths of a degree above absolute zero is like all the cars in a city simultaneously freezing at a traffic light, yet each one knowing it is both moving and standing still.