New optimization framework targets fiber laser power limits

Researchers at the National University of Defense Technology say a new multiphysics optimization framework could improve the design of high-power fiber lasers by combining physical modeling with machine learning. The work, published in Opto-Electronic Science, is presented with a high-power narrow-linewidth fiber laser demonstration that reached the highest output power publicly reported to date.

Why it matters: - High-power fiber lasers are central to advanced manufacturing and high-energy laser systems. - Further power scaling has been constrained by nonlinear effects, thermal effects, and beam-quality degradation that become more tightly coupled as power rises. - A framework that can optimize multiple interacting variables at once could shorten development cycles and improve performance.

What happened: - Professor Jinbao Chen and colleagues at the National University of Defense Technology developed a multiphysics optimization framework for high-power fiber lasers. - The work was published in Opto-Electronic Science on Sept. 23, 2026, under DOI 10.29026/oes.2026.260029. - The paper presents a model-algorithm fusion approach for the intelligent design of high-power fiber amplifiers and fiber lasers.

The details: - The framework is built around scientific machine learning, which combines data-driven optimization with physical constraints. - The authors frame the design problem as a constrained optimization task that can be solved in the computational domain. - Traditional approaches often rely on sequential parameter sweeps, empirical models and physical testing, which can miss the global optimum. - The new framework is designed to reduce dependence on manual iteration and improve both design accuracy and efficiency. - For the physical model, the paper introduces an "equivalent refractive index profile" in finite element method simulations. - That approach is meant to account for fabrication deviations, thermally induced waveguide distortions and dynamic coupling among effects such as stimulated Brillouin scattering, stimulated Raman scattering and transverse mode instability. - The optimization strategy is physics-informed and combines physical residuals, constraint penalties and objective terms in a composite loss function. - The paper says this reduces the need for massive high-fidelity datasets, which are costly to collect in fiber-laser development. - In the demonstration platform, the researchers used a high-power, narrow-linewidth fiber laser and achieved the highest output power publicly reported to date.

Between the lines: - The paper is aiming to move fiber-laser design from heuristic tuning to global co-design. - That shift matters because the hardest limits in this field are no longer isolated problems; they interact, which makes single-parameter optimization less effective. - The use of physics-informed machine learning suggests a broader trend in photonics: combining simulation, constraints and optimization to reduce reliance on trial-and-error hardware development.

What's next: - The framework could be applied to future high-power fiber laser and amplifier designs. - The paper suggests the approach may support more reliable prediction, better engineering feasibility and wider adoption of multi-dimensional design methods. - Opto-Electronic Science lists the article in its online archive and provides journal submission and contact information through its website.