Fault-tolerant quantum computation with a neutral atom processor

TL;DR

Fault-tolerant quantum computation on a quantum processor with 256 qubits, each an individual neutral Ytterbium atom, and the Bernstein-Vazirani algorithm with up to 28 logical qubits encoded into 112 atoms, showing better-than-physical error rates are demonstrated.

Abstract

Quantum computing experiments are transitioning from running on physical qubits to using encoded, logical qubits. Fault-tolerant computation can identify and correct errors, and has the potential to enable the dramatically reduced logical error rates required for valuable algorithms. However, it requires flexible control of high-fidelity operations performed on large numbers of qubits. We demonstrate fault-tolerant quantum computation on a quantum processor with 256 qubits, each an individual neutral Ytterbium atom. The operations are designed so that key error sources convert to atom loss, which can be detected by imaging. Full connectivity is enabled by atom movement. We demonstrate the entanglement of 24 logical qubits encoded into 48 atoms, at once catching errors and correcting for, on average 1.8, lost atoms. We also implement the Bernstein-Vazirani algorithm with up to 28 logical qubits encoded into 112 atoms, showing better-than-physical error rates. In both cases, "erasure conversion," changing errors into a form that can be detected independently from qubit state, improves circuit performance. These results begin to clear a path for achieving scientific quantum advantage with a programmable neutral atom quantum processor.

Figures (17)

• Figure 1: Experimental architecture. a, b, $^{171}$Yb atoms, each giving a physical qubit in the ground-state nuclear spin, can occupy tweezer traps within either a register array or an interaction zone (IZ). Up to 8 pairs of atoms interact at a time in the IZ, driven by two-qubit (2Q) gate lasers to implement controlled-phase (CZ) gates. These lasers perform coherent sequential excitation to a metastable clock state ($^{3}$P${_0}$) and the Rydberg state (65$^{3}$S${_1}$). Arbitrary single-qubit (1Q) operations can be applied in parallel to the atoms within a register row using Raman transitions detuned from the $^{3}$P${_1}$ manifold. Moving atoms between the zones enables arbitrary 2Q gate connectivity. c, Circuit-running sequence. Dissipative operations (left column) are performed with atoms held in a cavity-enhanced optical lattice. Three images (orange boxes) allow us to determine initial occupancy of register sites, as well as the final state and presence of atoms. Coherent operations (right column) are performed with atoms held in tweezers.
• Figure 2: Fidelity of prepared cat states versus qubit number $n$. Red points give fidelities for trials with no lost atoms. Cat states with up to $40$ qubits exceed the entanglement threshold. Gray points include trials in which all atoms are present initially but atoms lost during the circuit are replaced by the maximally mixed state, and black points include all trials.
• Figure 3: Encoded $24$-qubit cat state results. Shown is the error for measurements in the $X$ basis, the error for measurements in the $Z$ basis, and their sum, for three cases. The baseline is an unencoded $24$-qubit cat state, based on trials with no lost atoms. For the encoded cat state data, in one case we reject trials with a detected error, but try to correct for qubits lost during the circuit. In the other case, we reject trials with any lost qubit or detected error.
• Figure 4: Results for the Bernstein-Vazirani algorithm with an $n$-bit secret string. The string is always $1^n$, as the corresponding oracle is the most difficult to implement. The success probability is the probability of measuring $1^n$, conditioned on acceptance. For the unencoded algorithm (black), we accept when the $n$ measured atoms are present. For the encoded algorithm (red), we accept when all of the $n$ measured code blocks are decodable; in particular, it can detect faults and correct for lost atoms. The lighter points are based only on those trials with no initial atom loss.
• Figure 5: a, Logical circuit to prepare a $24$-qubit cat state. The qubits are arranged in $12$ pairs that will each be encoded into a ${[\![} 4,2,2{]\!]}$ code block. b, The encoded circuit involves $59$ atoms and $101$ physical CZ gates. All but the first code block use extra ancilla qubits, shown in gray, to prepare encoded ${|00\rangle}$ fault tolerantly; each ancilla should be measured to be $0$. In the diagram, a blue wire is a qubit in the Hadamard basis. Not shown are the final transversal $X$ or $Z$ measurements. c, Schedule for the atom movements and gates.