Fast algorithms for classical specifications of stabiliser states and Clifford gates
Authors:
Nadish de Silva,
Wilfred Salmon,
Ming Yin
Abstract:
The stabiliser formalism plays a central role in quantum computing, error correction, and fault-tolerance. Stabiliser states are used to encode computational basis states. Clifford gates are those which can be easily performed fault-tolerantly in the most common error correction schemes. Their mathematical properties are the subject of significant research interest.
Conversions between and verif…
▽ More
The stabiliser formalism plays a central role in quantum computing, error correction, and fault-tolerance. Stabiliser states are used to encode computational basis states. Clifford gates are those which can be easily performed fault-tolerantly in the most common error correction schemes. Their mathematical properties are the subject of significant research interest.
Conversions between and verifications of different specifications of stabiliser states and Clifford gates are important components of many classical algorithms in quantum information, e.g. for gate synthesis, circuit optimisation, and for simulating quantum circuits. These core functions are also used in the numerical experiments critical to formulating and testing mathematical conjectures on the stabiliser formalism.
We develop novel mathematical insights concerning stabiliser states and Clifford gates that significantly clarify their descriptions. We then utilise these to provide ten new fast algorithms which offer asymptotic advantages over any existing implementations. We show how to rapidly verify that a vector is a stabiliser state, and interconvert between its specification as amplitudes, a quadratic form, and a check matrix. These methods are leveraged to rapidly check if a given unitary matrix is a Clifford gate and to interconvert between the matrix of a Clifford gate and its compact specification as a stabiliser tableau.
For example, we extract the stabiliser tableau of a Clifford gate matrix with $N^2$ entries in $O(N \log N)$ time. Remarkably, it is not necessary to read all the elements of a Clifford matrix to extract its stabiliser tableau. This is an asymptotic speedup over the best-known method that is superexponential in the number of qubits.
We provide example implementations of our algorithms in Python.
△ Less
Submitted 26 May, 2024; v1 submitted 17 November, 2023;
originally announced November 2023.
