Milestone! Neutral atomic system moves towards scalable universal quantum computer

Neutral atomic systems have advantages in quantum simulations, but were previously thought to have no programming and general capabilities. In 2021, more and more research teams have demonstrated the programmability of neutral atomic systems. For example, Mikhail Lukin's team at Harvard University demonstrated a 256-atom programmable quantum simulator.

And on April 20, two papers simultaneously published in the journal Nature demonstrated the versatility and scalability of neutral atom quantum computers.

Demonstrate quantum algorithms
 

In the paper "Multi-qubit Entanglement and Algorithms on Neutral Atomic Quantum Computers" [1], a team led by Mark Saffman, professor of physics at the University of Wisconsin-Madison and chief quantum information scientist at ColdQuanta, became the world's first quantum computer A team demonstrating quantum algorithms on a neutral-atom quantum computer with a programming gate model.

Neutral atomic hyperfine qubits offer inherent scalability due to their identical properties, long coherence times and ability to be trapped in dense multidimensional arrays.

In this work, the architecture of the computer is based on the independent addressing of individual atoms, scanning a two-dimensional array of qubits using tightly focused beams. The team realized the preparation of 6-qubit entangled Greenberger-Horne-Zeilinger (GHZ) states, quantum phase estimation for chemical problems, and quantum approximate optimization algorithm (QAOA) for maximum cut graph problems.

Experimental quantum computing platform. a) Experimental layout for trapping and addressing atomic qubits. Atoms are trapped in a blue-detuned wire grid array imaged on the atom-trapping regions with a lens with NA = 0.7. Atomic occupancy was determined by collecting atomic fluorescence on opposite faces of the vacuum chamber and imaging the light to two separate regions of an EMCCD camera using a lens with NA = 0.7. A 1064 nm beam was used to rearrange the atoms into the desired positions for the line operation. The circuit is decomposed into a general set of gates, including microwave-driven global Rφ(θ) rotation around the x-y plane axis, local RZ(θ) rotation driven by the 459 nm beam, and pairing of atoms with 459 nm and 1040 nm beams The synchronous Rydberg excited CZ entanglement gate. b) Atomic energy level diagrams and wavelengths for cooling, trapping and qubit control. c) Average atomic fluorescence image of the 49-position array with a spacing of 3 µm. Each camera pixel is 0.6 × 0.6 µm at the atomic level. d) Global microwave Rabi rotation on a 9-qubit block at 76.5 kHz. The microwave phase, amplitude and frequency are controlled by an arbitrary waveform generator. e) Ramsey experiment using microwave π/2 pulses and a focused 459 nm beam to provide RZ(θ) rotation at a single location. A Stark shift of 600 kHz was used so that the 15 ns rise/fall time of the on/off acousto-optic modulator (not shown) can be ignored when calculating the pulse time of the RZ(θ) gate. f) Parity oscillation of the 2-qubit Bell state created using the CZ gate. One fact that measures the performance of an entanglement gate is its ability to generate Bell states. Measured and uncorrected Bell state fidelity of 92.7(1.3)% (State Preparation and Measurement (SPAM) fidelity 95.5%) for the optimized qubit pair, for all connected qubit pairs used in the circuit The average fidelity is 90% (SPAM fidelity 92.5%).

Preparation of the GHZ state. a) Parity oscillations of GHZ states with 2-6 qubits. The frequency of oscillation is linearly related to the number of qubits. b) Fidelity of the created GHZ states relative to the number of qubits. c) Measurement of the decoherence time of the GHZ state by Ramsey interferometry.

Quantum phase estimation using 3-qubit and 4-qubit. a) Phase estimation circuit using 4 qubits. b) Quantum phase estimation results using 3 qubits. c) Estimate the molecular energy of the H2 molecule using 4-qubit quantum phase estimation.

A QAOA algorithm for solving the maximum cut problem. a) 4-qubit QAOA max-cut circuit for a single p = 1 period. b) Decompose the Z interaction into two CNOT gates and one RZ rotation. c) Line results for p=1, 2, 3 with optimized γ and β values.

These results highlight the highly scalable capabilities of neutral atom qubit arrays for general-purpose programmable quantum computing as well as for nonclassical state fabrication for quantum-enhanced sensing.

"There is a race to make a useful quantum computer, and several different approaches are being developed for this," Saffman said. "Neutral atom qubits are one of five actively being developed. This paper demonstrates for the first time the use of neutral atom qubits. The ability to run quantum circuits and quantum algorithms.”

Core components of a quantum computer in the Saffman lab

ColdQuanta will soon launch the Hilbert, a 100-qubit-scale computer that builds on the groundbreaking work of this research. Take advantage of the natural scalability of neutral atom methods to provide Hilbert systems with high connectivity, high fidelity, miniaturization, and room temperature operation.

Demonstrate scalability
 

Another Nature paper, "Quantum Processors Based on Coherent Transport of Arrays of Entangled Atoms" [2] from Mikhai Lukin's group at Harvard University, demonstrated the potential of using neutral atoms to realize scalable quantum processors.

In quantum processors, the ability to engineer parallel, programmable operations between required qubits is key to building scalable quantum information systems. In most state-of-the-art methods, qubits interact locally, limited by the connectivity associated with their fixed spatial layout. To this end, the team demonstrated a quantum processor with dynamic, non-local connections, in which entangled qubits are coherently transported in two spatial dimensions in a highly parallel manner between single-qubit and two-qubit operational layers .

Quantum information architecture through coherent transport of neutral atoms. a) In this method, qubits are transported to perform entanglement gates with remote qubits, enabling programmable and non-local connections. Atom shuttling using optical tweezers is highly parallel between two dimensions and multiple domains, allowing selective manipulation. b) Atomic image of coherent transport of entangled qubits. c) The odd-even oscillations show that motion does not significantly affect entanglement or coherence. d) In the 110 μm range, the measured Bell state fidelity is a function of separation speed, indicating that fidelity is not affected at moving speeds below 200 μs (average separation speed of 0.55 μm/μs).

The team's method utilizes arrays of neutral atoms trapped and transported by optical tweezers; hyperfine states for robust quantum information storage and excitations as Rydberg states for entanglement generation. Use this architecture to enable programmable generation of entangled graph states, such as cluster states and 7-qubit Steane code states.

1D and 2D graphs of transport using dynamic entanglement. a) Generation of a 12-atom 1D cluster state graph, created by initializing all qubits (vertices) in |+⟩ and applying CZ gates on the links (edges) between qubits. b) Quantum circuit representation of one-dimensional cluster state fabrication and measurement. Preparation and measurement of quantum circuit representations. Dynamic decoupling applies to all quantum circuits (methods). c) The original measurement stabilizer of the resulting 1D cluster state. d) Graph representation of the 7-qubit Steane code. e) The circuit for preparing the logic state |+⟩L of the Steane code is executed in four parallel gate layers. f) Stabilizers and logical operators measured after preparing |+⟩L.

In addition, the researchers shuttled the entangled auxiliary array to achieve a surface code state with 13 data and 6 auxiliary qubits and a toric code state with 16 data and 8 auxiliary qubits.

Topological surface code and toric code states using mobile-assisted qubit arrays. a) Surface code for graph implementation. b) X-plaquette (lattice) and Z-star (star) stabilizers of the measured surface codes, and logical operators with and without error detection (implemented in post-selection). c) Implementation of toric codes. d) The X-plaquette and Z-star stabilizers of the measured toric codes, and the logical operators of the two logical qubits with and without error detection (implemented in post-selection).

Finally, the researchers used this architecture to achieve a hybrid analog-digital evolution and used it to measure entanglement entropy in quantum simulations, experimentally observing non-monotonicity associated with quantum many-body scars Entanglement dynamics.

Dynamic reconfigurability of hybrid analog-digital quantum simulations. a) Hybrid quantum circuits combining coherent atomic transport and simulated Hamiltonian evolution with digital quantum gates. b) Measurement of entanglement entropy in many-body Rydberg systems using two-copy interferometry. c) Measurement of the half-chain Renyi entanglement entropy of two 8-atom systems after many-body kinetic quenching. d) Measurement of mutual information during 0.5 μs quenching time reveals volume-law scaling for thermalized |gggg...⟩ states and area-law scaling for scar |rgrg...⟩ states.

These results achieve a long-term goal of providing a path towards scalable quantum processing and enabling applications from simulation to metrology.

Link:

[1] https://www.nature.com/articles/s41586-022-04603-6

[2] https://www.nature.com/articles/s41586-022-04592-6