The Signal
Pasqal Finds Three Relaxation Regimes in a 256-Qubit Quantum Magnet
Pasqal reports three relaxation patterns in a 256-atom Rydberg simulation of a two-dimensional quantum magnet. The study finds a slow-relaxation regime where classical tensor-network calculations lose agreement at late times.
Quantum.world Team · Written with AI · fact-checked ·
On Oct. 27, Pasqal described a 256-atom experiment that found three distinct ways a two-dimensional quantum magnet relaxes after a sudden change in its governing field. The result includes a slow-relaxation regime that the company says appears where leading classical tensor-network simulations cease to agree at late times. The underlying preprint, submitted to arXiv on Aug. 7, studies a transverse-field Ising model using a programmable array of rubidium atoms.
A 16-by-16 array, suddenly quenched
The experiment arranged 256 rubidium atoms in a square lattice, with each atom’s ground and Rydberg states representing the two states of a spin. Laser coupling supplied the model’s transverse field, while interactions between excited atoms supplied ferromagnetic coupling. The team first prepared the array in a fully polarized state, then abruptly changed the transverse field and tracked magnetization and correlations.
By varying the ratio of field strength to interaction strength, the researchers sampled different points across the model’s phase diagram. Pasqal reports measurements for ratios from 2.0 to 6.0. The researchers examined correlations as well as magnetization because the two observables reveal different aspects of the system’s relaxation.
Fast thermalization gives way to a slowdown
At lower field strengths, near a ratio of 2.5, correlations approach their thermal values on a timescale of roughly 1/J, where J is the interaction scale. At an intermediate ratio of 3.5, they move toward equilibrium much more slowly and remain far from the thermal value beyond that characteristic timescale. At the higher ratio of 6.0, correlations settle into a long-lived prethermal plateau rather than quickly reaching thermal equilibrium.
“The second, and most unexpected, is a crossover regime characterized by a slowdown in relaxation,” write Tiago Mendes-Santos et al., in their arXiv preprint. The paper links the high-field plateau to an effective XY-type model with a U(1) symmetry. The authors expect that plateau eventually to melt into thermal behavior, but say the timescale grows exponentially with the ratio of field to interaction strength.
Where classical calculations stop agreeing
Tensor-network simulations agree quantitatively with the 16-by-16 processor data at early times, up to about tJ = 1. Later, calculations using different tensor-network approaches begin to diverge: around tJ greater than 1.5 for an 8-by-8 array and greater than 1 for a 16-by-16 array. In the slow regime, the spread among those calculations becomes comparable to or larger than their deviation from the processor measurements, leaving no controlled classical prediction of the later dynamics.
The comparison does not establish a general quantum-computing advantage. Pasqal explicitly says the study does not compare computational runtimes or costs. Its narrower result is that the measured processor dynamics remain consistent across system sizes in a time range where the classical methods tested no longer provide a reliable common account.
The slowdown’s cause is still open
The paper’s authors propose two possible contributors: the high-field effective model may break down as the field weakens, while critical slowing near a thermal transition may also affect relaxation. Neither explanation is established by the experiment. The team’s 5-by-5 state-vector simulations reproduce the three qualitative regimes even without noise, while its analysis also finds that noise and system size can change the measured decay rate.
Pasqal says separating those effects will require measurements at larger sizes and longer evolution times, alongside lower-noise hardware and improved classical calculations. The present data map the regimes and their crossover; they do not yet pin down a quantitative account of why the intermediate-field correlations decay so slowly.