Short answer: Ultracold fermionic atoms in optical lattices can realise Hubbard models, which are central to condensed-matter physics, and are used as analog quantum simulators. A 2025 Nature paper reports a several-fold reduction in the temperature reached in such a simulator. Theory work proposes using the same platforms as programmable simulators for ground-state preparation. These are laboratory experiments and proposals.
What an analog simulator does
A Hubbard model describes particles hopping on a lattice with interactions between them. The 2025 paper notes that ultracold fermionic atoms in optical lattices give pristine realisations of these models, which matter for modern condensed-matter physics. In an analog simulator the atoms themselves play the part of the electrons and the lattice is set up to match the model. A 2017 Science review describes ultracold atoms in optical lattices as an ideal platform for many-body simulations, with quantum gas microscopes allowing observation and manipulation of single atoms in large samples. It says such simulators have been used to probe quantum magnetism, to realise and detect topological matter, and to study systems with controlled long-range interactions, including out of equilibrium.
Reaching lower temperatures
The 2025 paper says the temperatures accessible in these optical lattice analogues have been too high to address many open problems. The authors report a several-fold reduction in temperature, by transforming a low-entropy product state into strongly correlated states through dynamic control of the model parameters, which they note is extremely challenging to simulate classically. At half-filling they report long-range antiferromagnetic order close to saturation, with a temperature given in units of the tunnelling energy of about 0.05 (with an uncertainty range reported in the paper), from comparison with numerically exact simulations. Away from half-filling, they note that accurate and predictive numerical simulation is exceedingly challenging.
Programmable operation
A separate paper in Physical Review describes how recent advances, including time-dependent local control of chemical potentials and tunnelings, let the platform operate digitally as a programmable simulator. The authors develop ground-state preparation algorithms for different fermionic models using single-particle and many-body fermionic gates provided by the native Hubbard Hamiltonian. They design variational precompiled circuits to prepare the ground state of the native model, with significant speedups relative to competing adiabatic protocols, and say the approach can also target extended models with terms not natively realised on the platform. They benchmark for ladder geometries and say the protocols can be applied to two-dimensional setups.
Why simulate at all
The 2017 review and the 2025 paper both frame the motivation the same way: strongly correlated many-body states are hard for classical computers, and a controllable quantum system can reach regimes that numerical methods struggle with. The simulator is also a test bench, since its output can be compared with numerically exact calculations where those exist, as the 2025 authors do at half-filling. That comparison is how the temperature is inferred, and it is why the figure carries an uncertainty.
Limits
The temperature figure depends on comparison with numerical simulation, and the programmable protocols are proposals benchmarked on ladders. Neither paper claims a practical advantage for any industrial problem.
FAQ
What is analog quantum simulation?
Using a controllable quantum system, here ultracold atoms in a lattice, to imitate the physics of another.
What did the 2025 experiment report?
A several-fold reduction in temperature and antiferromagnetic order close to saturation at half-filling.
Why is doping hard to simulate classically?
The 2025 paper notes that away from half-filling, accurate predictive numerical simulation is exceedingly challenging.
Can the platform be programmed?
A Physical Review paper proposes variational circuits for ground-state preparation using the native gates.
Is this useful commercially?
The papers do not claim that. They are basic research on many-body physics.
All field notes