Optimization complexity and resource minimization of emitter-based photonic graph state generation protocols

TL;DR

This paper develops graph theory-based heuristics to optimize emitter-based photonic graph state generation, significantly reducing entangling gates and resource costs, with implications for scalable quantum computing and networking.

Contribution

It introduces novel heuristics for minimizing entangling gates in photonic graph state protocols, achieving substantial resource reductions despite the NP-hard nature of the optimization.

Findings

Up to 75% reduction in entangling gates for random graphs.

Heuristics enable processing large graphs with up to 66% fewer CNOTs.

Global minimization of resources achieved for repeater graph states of any size.

Abstract

Photonic graph states are important for measurement- and fusion-based quantum computing, quantum networks, and sensing. They can in principle be generated deterministically by using emitters to create the requisite entanglement. Finding ways to minimize the number of entangling gates between emitters and understanding the overall optimization complexity of such protocols is crucial for practical implementations. Here, we address these issues using graph theory concepts. We develop optimizers that minimize the number of entangling gates, reducing them by up to 75$\%$ compared to naive schemes for moderately sized random graphs. While the complexity of optimizing emitter-emitter CNOT counts is likely NP-hard, we are able to develop heuristics based on strong connections between graph transformations and the optimization of stabilizer circuits. These patterns allow us to process large graphs and still achieve a reduction of up to $66\%$ in emitter CNOTs, without relying on subtle metrics such as edge density. We find the optimal emission orderings and circuits to prepare unencoded and encoded repeater graph states of any size, achieving global minimization of emitter and CNOT resources despite the average NP-hardness of both optimization problems. We further study the locally equivalent orbit of graphs. Although enumerating orbits is $\#$P complete for arbitrary graphs, we analytically calculate the size of the orbit of repeater graphs and find a procedure to generate the orbit for any repeater size. Finally, we inspect the entangling gate cost of preparing any graph from a given orbit and show that we can achieve the same optimal CNOT count across the orbit.