Description
The aim of this work is to demonstrate the feasibility of a hybrid optimization approach that combines Genetic Algorithms (GA) with Monte Carlo (MC) radiation transport simulations to optimize the parameters of diagnostics for magnetic confinement fusion (MCF) plasmas. The design and engineering of such complex systems, which are essential for plasma control and stability, typically involve high-dimensional, multi-parameter optimization problems. Genetic Algorithms, as a class of evolutionary algorithms inspired by natural selection, have strong potential for efficiently exploring non-linear, multi-dimensional design spaces [1], enabling significant progress in tokamak diagnostic optimization and automated data analysis.
In this contribution, we present a design optimization procedure that integrates GA-based optimization with fast, in-house developed MC radiation transport calculations. As a case study, we consider a generic thin-foil proton recoil (TPR) neutron spectrometer. This choice is motivated by the need to accurately diagnose and control plasma behavior in tokamaks, where high-resolution neutron spectroscopy plays a crucial role for ion temperature and fuel ion ratio monitoring, fast ion physics studies and fusion power calibration [2].
The optimization focuses on several geometrical parameters of the TPR spectrometer to achieve an optimal trade-off between energy resolution and detection efficiency, while maintaining the background rejection rate below a predefined threshold. The overall methodology follows a standard GA framework coupled with a fast MC radiation transport code to keep a tractable numerical cost. In the first step, the MC code is validated for selected cases against the widely used and benchmarked Geant4 toolkit [3]. Subsequently, the combined GA–MC approach is applied to optimize key geometrical parameters—such as converter thickness, detector dimensions, and layout—within a multi-dimensional parameter space. Finally, the performance of selected GA–MC optimized configurations, in terms of efficiency, energy resolution, and average reconstructed proton energy, is independently verified using a detailed Geant4 model.
The proposed methodology is generic and not limited to the TPR diagnostic or to geometrical optimization alone. It can therefore be extended to the design optimization of other plasma diagnostic systems for MCF devices in future studies.
[1] Goldberg, D. E. (1989). Genetic algorithms in search, optimization, and machine learning. Addison-Wesley.
[2] Scholz, M., Hjalmarsson, A., Hajduk, L., et al. (2019). Conceptual design of the high resolution neutron spectrometer for ITER. Nuclear Fusion, 59(6), 066001.
[3] S. Agostinelli et al., Nucl. Instrum. Meth. A 506 (2003) 250-303.
Acknowledgements:
The project is co-financed by the Polish National Agency for Academic Exchange (POLONIUM programme, contract no. BPN/BFR/2024/1/00002/U/00001). We gratefully acknowledge Polish high-performance computing infrastructure PLGrid (HPC Centers: ACK Cyfronet AGH, CI TASK, WCSS) for providing computer facilities and support within computational grants no. PLG/2025/017971 and no. PLG/2026/019127. This work was partially funded by National Science Centre, Poland (NCN) grant OPUS 29 no. 2025/57/B/ST2/04168.