In 1963, a Tanzanian schoolboy named Erasto Mpemba noticed that hot ice cream freezes faster than cold. That paradox, named after him, long remained a subject of debate in classical thermodynamics. Now the quantum world offers its own version: a state with more strongly broken symmetry can recover it faster than a less broken one. This is the quantum Mpemba effect, a race of symmetries in which the one that deviates more crosses the finish line first.

On a 16-qubit superconducting processor with a ring topology, the researchers staged this race under controlled conditions. 14 qubits formed a chain, with nearest-neighbor couplings set by capacitive elements and long-range couplings mediated by a central tunable resonator. The ratio of nearest-neighbor to long-range coupling, r, varied from 10 to 1, allowing switching between regimes of strong localization and thermalization. The initial states, tilted Néel and ferromagnetic product states with tilt angles π/4 and π/2, broke symmetry to different degrees. To measure entanglement asymmetry (a measure of deviation from a symmetric state), they performed quantum tomography of a three-qubit subsystem: 27 measurement bases, density matrix reconstruction via least squares with physical constraints, then projection onto charge sectors and computation of the difference in von Neumann entropies.

The results showed that the effect is not constant but depends on conditions. In the regime of strong nearest-neighbor couplings (r≈10), the entanglement asymmetry curves for states with angles π/4 and π/2 cross: the state with greater symmetry breaking recovers faster, a clear sign of the effect. In the intermediate regime (r≈1), the crossing disappears: the asymmetry decays monotonically, and the entanglement entropy reaches the Page value, indicating suppression of the effect upon thermalization. However, the effect could be brought back in two ways: by introducing a linear potential on the sites (W=6 MHz), which suppresses ergodicity more strongly for the smaller tilt angle, and by starting from tilted ferromagnetic states, where the dynamics reduces to the Lipkin–Meshkov–Glick model. In the second case, the crossing persists even under strong on-site disorder (uniform distribution in the range ±14g), unlike in previous ion-trap experiments.

This work is not just a demonstration of exotic behavior. It shows that the quantum Mpemba effect is a controllable property of nonequilibrium dynamics. Unlike neighboring studies where the effect was observed in a single configuration, here a multidimensional modulation is demonstrated: changing the coupling range, local potentials, and initial states. This provides universal knobs for controlling relaxation and links symmetry recovery to thermalization and ergodicity. Practical benefits may appear in faster qubit reset and quantum state preparation, which are critical operations for quantum computers.

The prospects are broad. Possible directions include studying the effect in the prethermalization regime, under other symmetries (Z₂, SU(2)), and with Hamiltonians that break symmetry. It would be interesting to explore the connection with accelerated qubit reset and state preparation. The race of symmetries is just beginning, and now we have a track where we can change the rules.