This project focuses on the development of efficient numerical methods for the design and optimization of acoustic metamaterials, with the aim of enabling their practical application in noise mitigation for engineering systems. By combining reduced-order modeling with advanced optimization strategies, the proposed methods significantly decrease the computational cost typically associated with metamaterial design, allowing for the rapid exploration of complex unit-cell geometries and material configurations tailored to specific noise-attenuation requirements in real-world engineering applications.
Noise pollution due to transport or industrial activities is one of largest threats for the environment and may cause severe health problems. Therefore, noise mitigation plays an essential role for engineering design. In the current engineering practice, this is usually achieved by adding damping treatments in a late phase of the design process or even after manufacturing has begun. The resulting increase in mass can be significant and deteriorates the ecological footprint of the product. In recent years, acoustic metamaterials have attracted scientific attention and are considered a promising solution for finding the delicate balance between lightweight design and quietness. The overarching goal of this project is to develop efficient numerical methods for design optimization of elastic and acoustic metamaterials.
This goal will be achieved by combining expertise in acoustic engineering and numerical linear algebra. As acoustic metamaterials consist of periodic arrangements of unit cells (so-called metaatoms), the periodicity is also present in the parameter-dependent linear systems of equations arising in the corresponding FEM-BEM models. In this project, efficient structure-exploiting numerical methods for their solution will be developed. This will be a great benefit for model order reduction and also enables computational design optimization. One aim is to optimize the resonance frequencies of the metamaterial by nonlinear eigenvalue optimization. Additionally, we will consider the actual load cases that result in H∞ optimization problems. The new computational techniques will be tested on relevant industrial benchmark models and then be put into engineering practice.
Book Chapters
- Trust-region methods with low-fidelity objective models, A. Angino, M. Aurina, A. Kopaničáková, M. Voigt, M. Donatelli, and R. Krause, Domain Decomposition Methods in Science and Engineering XXIX, Springer, December 2025, Accepted for publication, https://doi.org/10.48550/arXiv.2511.00434
- A non-monotone preconditioned trust-region method for neural network training, A. Angino, B. Çapriqi, S. Likaj, K. Trotti, and R. Krause, Domain Decomposition Methods in Science and Engineering XXIX, Springer, 2025, Accepted for publication, https://doi.org/10.48550/arXiv.2605.14860
Conference Papers
- Second-order optimization of variable projection SVM models and road abnormality detection, A. Angino, M. Voigt, R. Krause, and T. Dózsa, ICASSP 2026 – 2026 IEEE International Conference on Acoustics, Speech and Signal Processing, Barcelona, Spain, 2026, https://doi.org/10.1109/ICASSP55912.2026.11463454
Preprints / Submitted Papers
- Adaptive kernel methods, T. Dózsa, A. Angino, Z. Szabó, J. Bokor, and M. Voigt, arXiv preprint arXiv:2601.21707, Jan. 2026, Submitted for publication, https://doi.org/10.48550/arXiv.2601.21707
- H2-optimal model order reduction using hyperbolic geometry, A. Angino, T. Dózsa, and M. Voigt, Mar. 2026, Submitted for publication
