Quantum Contextuality Measured Across 105 Qubits by Google

Google Willow processor used to test quantum contextuality across 105 qubits

Google Quantum AI has measured quantum contextuality on its superconducting processors and shown it holds up in systems as large as 105 qubits. The results, published in Science, beat classical limits in several tasks and double as a new way to catch hardware errors.

What contextuality means

Contextuality is the experimentally testable claim that you can’t give fixed, pre-existing values to everything you might measure. What a measurement returns can depend on which compatible measurements are made alongside it, which is its context.

Entanglement usually gets the credit for quantum computing’s power. A growing body of theory points to contextuality as the deeper resource behind certain algorithms that beat classical ones. Until now it had been shown almost entirely in systems of a few particles. Whether it survives in the many-qubit machines that real computers use was open.

The magic square game

The first test is a puzzle. Fill a 3×3 grid with +1 and -1 so each row multiplies to +1 and each column to -1. It can’t be done, because the rows force an even count of -1 entries and the columns force an odd count.

Turn it into a game. Alice gets a row, Bob gets a column, and they can’t talk. They win if their answers agree in the shared cell. With any plan fixed in advance, they win at most 8 rounds out of 9, or 88.89%.

On the Willow processor, each player held two qubits from two shared entangled pairs. Across all nine versions of the game, played 100,000 times each, the qubits won 98.30% of rounds (0.9830 plus or minus 0.0001). They lost roughly one round in sixty.

A state-independent score

The team then measured how contextual the machine is, using a Bell-Kochen-Specker inequality that bundles the row and column products into one number, chi. Any predetermined-outcome model gives chi of 4 or less. Quantum mechanics allows up to 6.

The test doesn’t depend on how the qubits are prepared, so it describes the machine rather than a chosen input. Measuring it required reading every row and column on the same two qubits without disturbing them, which the team did with a helper qubit. They ran 540 measurements in a row, repeated 1,000 times, and got chi of 5.590 plus or minus 0.008. That is about 7% from the quantum limit.

Scaling quantum contextuality to many qubits

Noise grows with every added qubit, so the harder question was scale. Two further experiments addressed it.

In the GHZ parity game, N players share one entangled state and each answers a one-bit question without communicating. Random guessing wins half the time. The best classical plan wins 75% for three players and drifts toward 50% as players are added. A perfect GHZ state wins every round. Google played with up to 71 players, one per qubit. Noise kept the win rate under 100% and the lead narrowed with each added qubit, but results stayed above the classical bound. The team also verified genuine entanglement across 45 qubits.

The second experiment ran a 2018 theorem’s problem of finding a hidden linear function. A constant-depth quantum circuit can solve it while no constant-depth classical circuit can, because each classical output only sees a small neighborhood of the input. Google ran instances on grids of up to 105 qubits, with success probabilities above one-half, using the shallow-circuit recipe the theory calls for. It hadn’t been run at this scale on real hardware before.

A benchmark that finds hidden errors

These tasks need many qubits to agree at once and exactly. Running them exposed subtle coherent errors, the kind that rotate a quantum state slightly off target. Standard random-circuit benchmarks tend to average those errors out of sight.

The tasks also give a fixed ceiling set by physics: chi of 6, a win probability of 1, a perfect contextual fraction. The gap between what a chip achieves and that ceiling is a score that doesn’t get easier as classical methods improve. Google says it matters for applications such as molecular simulation, where structured algorithms leave no place for small errors to hide.

Links to error correction and what comes next

Theory ties contextuality to the magic states that error-corrected quantum computers need to run full algorithms. Google’s team recently prepared such states using a technique called magic state cultivation. The researchers say measuring contextuality across many qubits lets them probe on hardware what gives quantum computers their edge. The practical next step is using these tasks as a regular yardstick as processors improve, since the distance to the physical ceiling can be read directly.

Source: research.google

Written by

Saksham Bhargava

Saksham Bhargava runs Alteroid and has been writing about consumer technology for over five years, covering iPhone, Android, Windows, AI tools and gaming. With an MCA and six years of hands-on web development experience, he tests the features he writes about on his own devices and focuses on clear, practical guides that solve real problems. Away from the screen, he enjoys travelling to places most tourists skip.