Topological quantum compilation of metaplectic anyons based on the genetic optimized algorithms

TL;DR

This paper develops topological quantum gate compilation methods for metaplectic anyons using genetic algorithms, achieving efficient approximation of quantum gates with potential for fault-tolerant quantum computing.

Contribution

It introduces a novel approach combining unconventional encoding and genetic algorithms to compile quantum gates in metaplectic anyon models, extending to multi-qubit systems.

Findings

Successfully constructed H and T gates with high accuracy.

Achieved smaller approximation errors than Fibonacci anyon models.

Demonstrated effective two-qubit gate approximation with braidwords.

Abstract

Topological quantum computing holding global anti-interference ability is realized by braiding some anyons, such as well-known Fibonacci anyons. Here, based on $SO(3)_2 $ theory we obtain a total of 6 anyon models utilizing \textit{F}-matrices, \textit{R}-symbols, and fusion rules of metaplectic anyon.We obtain the elementary braiding matrices (EBMs) by means of unconventional encoding. After braiding \textit{X} and $X^\prime$, we insert a pair of \textit{Z} anyons into them to ensure that the initial order of anyons remains unchanged. In this process only fusion is required, and measurement is not necessary. Three of them $\{V^{113}_3,V^{131}_3,V^{133}_1\}$ are studied in detail. We study systematically the compilation of these three models through EBMs obtained analytically. For one-qubit case, the classical \textit{H}- and \textit{T}-gate can be well constructed using the genetic algorithm enhanced Solovay-Kitaev algorithm (GA-enhanced SKA) by $\{V^{113}_3,V^{131}_3,V^{133}_1\}$. The obtained accuracy of the \textit{H}/\textit{T}-gate by $\{V^{113}_3,V^{133}_1\}$ is slightly inferior to the corresponding gates of the Fibonacci anyon model, but it also can meet the requirements of fault-tolerant quantum computing, $V^{131}_3$ giving the best performance of these four models. For the two-qubit case, we use the exhaustive method for short lengths and the GA for long lengths to obtain braidword for $\{V^{113}_3,V^{131}_3,V^{133}_1\}$ models. The resulting matrices can well approximate the local equivalence class of the CNOT-gate, while demonstrating a much smaller error than the Fibonacci model, especially for the $V^{113}_3$.The braiding processes of conventional encoding (using identical anyons) and unconventional encoding (using distinct anyons) are compared. Finally, we attempt to generalize the model to the \textit{N}-qubit case.