← Latest papers
⚛️ quantum physics

CSS Circuits: useful quantum circuits that can be simulated with O(1)O(1) cost per gate

This paper introduces an efficient O(1)O(1)-cost simulator for "CSS circuits" composed of Clifford gates and Pauli measurements, demonstrating their utility for protocols like teleportation and error correction while proving via Bell inequalities that such local hidden variable models cannot support non-Clifford gates like Hadamard, CZ, or S.

Original author: Craig Gidney, Cody Jones

Published 2026-11-13
📖 5 min read🧠 Deep dive

Original author: Craig Gidney, Cody Jones

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

In the realm of quantum computing, scientists are constantly trying to build machines that can solve problems too complex for ordinary computers. To understand how these machines work, researchers often rely on a mathematical framework called stabilizer circuits. These are specific arrangements of quantum operations that, while powerful enough to perform tasks like teleporting information or correcting errors, are simple enough that classical computers can predict their behavior without needing a quantum computer themselves. This ability to simulate quantum behavior on standard machines is crucial for testing and designing new quantum protocols before they are built. However, as these circuits grow larger, the time required for a classical computer to simulate them usually explodes, making it difficult to test systems with thousands of qubits or millions of operations.

A new study by researchers at Google Quantum AI introduces a specialized subset of these circuits that can be simulated with unprecedented speed. They call these "CSS circuits." While these circuits are built from a very limited set of tools—specifically, operations that flip bits, flip phases, measure in specific directions, and link two qubits together—they are surprisingly capable. The researchers demonstrate that these circuits can execute vital quantum protocols, including teleportation and error correction, yet they can be simulated by a classical computer in constant time, regardless of how many gates are involved. This means that as the circuit grows, the computational cost per operation does not increase, a feat that was previously thought impossible for such a broad range of quantum tasks.

The core of this discovery lies in the specific restrictions placed on the gates used to build the circuit. The researchers defined a CSS circuit as one constructed only from X gates, Z gates, CNOT gates, and measurements in the X or Z basis. These operations are highly structured; they never mix different types of quantum information in the way that more complex gates do. Because of this structure, the researchers found that the entire state of the system can be tracked using a simple method that requires only a tiny, fixed amount of work for every single gate added to the circuit. Instead of performing a massive calculation that scales with the size of the system, the simulator simply updates a few bits of information for each qubit. This approach allows for the simulation of circuits with hundreds of thousands of gates and thousands of qubits in a fraction of the time it would take using traditional methods.

This speed comes with a significant trade-off, which the researchers explored in detail. They proved that this efficient simulation method relies on the circuits behaving in a way that is consistent with a "local hidden variable" model. In plain terms, this means the simulation assumes that the properties of the qubits are determined by local factors that do not communicate faster than light. Because of this assumption, the simulator cannot handle certain other quantum gates, such as the Hadamard gate or the S gate, which are essential for creating the kind of deep entanglement that violates Bell inequalities. Bell inequalities are tests that show whether a system is behaving in a truly quantum way that cannot be explained by local rules. The researchers showed that if they added these forbidden gates to their CSS circuits, the circuits would be able to pass these tests, proving that their simple simulation method would fail. This establishes a clear boundary: the speed comes at the cost of being unable to simulate the most "quantum" aspects of these systems.

The practical value of this finding is particularly high for testing quantum error correction in specific types of hardware. In machines built with neutral atoms, a common problem is the loss of an atom, which causes any gate targeting that atom to simply disappear. Traditional simulation methods struggle with this because they must track complex changes to the system's state. The new CSS simulation method, however, handles this naturally. Since the method tracks whether qubits are flipped or not, it can easily account for missing gates without the computational cost spiraling out of control. This makes CSS circuits an ideal candidate for benchmarking how well error correction works in the presence of atom loss, a critical step toward building reliable quantum computers.

The researchers also addressed the limits of their discovery by showing what these circuits cannot do. While CSS circuits can perform teleportation and entanglement purification, they cannot generate the specific type of entanglement required to win a famous test known as the GHZ test unless additional, non-CSS gates are introduced. The team constructed examples showing that adding a Hadamard gate, a CZ gate, or an S gate to a CSS circuit allows it to pass this test, which a local hidden variable model cannot do. This confirms that the efficient simulation is not a universal solution for all quantum circuits but is instead a powerful tool for a specific, highly useful class of them.

In their conclusion, the authors suggest that while CSS circuits are simple, they are not entirely new; they appear to be a rediscovery of a concept that has likely existed in different forms, such as in phase-free diagrams used by other researchers. Nevertheless, the paper provides a clear, concrete demonstration that a restricted set of quantum operations can be simulated with constant effort per gate. This offers a new, efficient way to analyze and design quantum error correction protocols, particularly for hardware prone to atom loss, without needing the full computational power of a general-purpose quantum simulator. The work stands as a reminder that sometimes, by narrowing the scope of what a system can do, scientists can find a path to understanding it much faster.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →