PuSH - Publication Server of Helmholtz Zentrum München

Ma, Q.* ; Ma, R. ; Yu, H.* ; Jiang, X.* ; Zhou, Q.* ; Wang, X.* ; Liang, X.* ; Ni, K.*

Inverse design of mode-locked fiber lasers based on conditional generative adversarial network (cGAN).

Opt. Commun. 596:132519 (2025)
DOI
Open Access Green as soon as Postprint is submitted to ZB.
Mode-locked fiber lasers (MLFLs) are indispensable across scientific and industrial applications, but their conventional design paradigm suffers from computational inefficiency due to reliance on iterative numerical methods. To address this, we propose an inverse design method based on a conditional generative adversarial network (cGAN). This approach enables end-to-end reverse design and explore the diversity of the solution space. First, by establishing a forward model to learn the mapping between the cavity parameters and the laser output, we achieve more efficient pulse prediction. Subsequently, a generator network produces cavity parameter candidates, while a discriminator network evaluates their feasibility. The forward model's loss is integrated into generator's training to enhance prediction accuracy. This approach efficiently explores the parameter space, overcoming the many-to-one mapping challenge while significantly reducing computational costs. The proposed method enables rapid and precise MLFL design, advancing high-performance ultrafast laser systems.
Altmetric
Additional Metrics?
[➜Log in]
Edit extra informations Login
Publication type Article: Journal article
Document type Scientific Article
ISSN (print) / ISBN 0030-4018
Journal Optics Communications
Quellenangaben Volume: 596, Issue: , Pages: , Article Number: 132519 Supplement: ,
Publisher Elsevier
Publishing Place Amsterdam [u.a.]
Reviewing status Peer reviewed
Institute(s) Helmholtz Pioneer Campus (HPC)
Grants
Hunan Provincial Natural Science Foundation
Guangdong ST Programme