Speakers
Description
The design of fourth-generation synchrotron light sources based on Hybrid Multi-Bend Achromat (H-MBA) structures faces significant challenges due to the high dimensionality of design variables and the strong nonlinear effects induced by strong focusing forces. The traditional paradigm of manual matching followed by stepwise fine-tuning'' encounters bottlenecks in optimization efficiency and physical interpretability. This paper proposes a modular optimization framework that fuses physics priors with statistical learning to achieve synergistic optimization of linear optics and nonlinear dynamics. The framework uses Twiss parameter evolution as an intermediate physical representation, and a physics-prior screening mechanism driven by linear transport is combined with the Covariance Matrix Adaptation Evolution Strategy (CMA-ES). This approach identifies stable periodic solutions with reduced natural emittance within minutes and shows reproducible efficiency gains over manual initializations. Machine-learning classifiers trained on the generated dataset perform high-confidence pruning of the solution space and retain high-quality solutions. A local trust region constructed around thesepromising solutions'' introduces the Sequential Model-based Algorithm Configuration (SMAC) strategy based on Random Forests for refined iteration. This method provides an efficient and intelligent pathway for complex, high-dimensional lattice design.
| In which format do you inted to submit your paper? | LaTeX |
|---|---|
| Preprint marking on your proceeding paper | I wish my paper to be marked as preprint. |
| I no longer wish to present this contribution, please withdraw it. | Keep my contribution |