Solving perfect matchings by frequency-grouped multi-photon events using a silicon chip
Year: 2025
Type: article
Abstract: Computing the number of perfect matchings of a graph is a famous #P-complete problem. In this work, taking the outstanding advantages of the frequency dimension of photon, we propose and implement a photonic perfect matching solver, by combining two key techniques, frequency grouping and multi-photon counting. Based on a broadband photon-pair source from a silicon quantum chip and a wavelength-selective switch, we configured graphs up to sixteen vertices and calculated the perfect matchings of subgraphs up to six vertices. The experimental fidelities are more than 90% for all the graphs. The developed photonic system can be applied in solving nondeterministic-polynomial-time(NP)-complete problems, and we experimentally demonstrate the samples can enhance the random search for solving of the Boolean satisfiability problem and the finding of densest subgraph. Our perfect matching solver avoids destructive interference and does not need to consider the indistinguishability of photons. Moreover, it is simple and convenient to transform or scale up the graph by programming a wavelength-selective switch without changing the photonic source. Our work suggests an excellent platform for graph problem solving.
Source: Nature Communications
Authors: Pingyu Zhu, Qilin Zheng, Kun Wang, Miaomiao Yu, Gongyu Xia, Jiacheng Liu, Yong Liu, Zhihong Zhu, Ping Xu
Institution: National University of Defense Technology
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $6,990
OpenAlex ID: W4409681229