Research¶
Overview¶
Our research focuses on computational methods for engineering design under complexity, uncertainty, and high computational cost.
We are particularly interested in methods that improve:
- design-space exploration
- optimization efficiency
- physical interpretability
- reproducibility of computational results
Main Themes¶
Design-space dimensionality reduction¶
We develop methods to reduce the dimensionality of high-dimensional parametric design spaces while preserving relevant geometric and physical information.
Representative topics include:
- principal component analysis and modal parameterization
- parametric model embedding (PME)
- physics-informed and physics-driven dimensionality reduction
- backmapping and parametric consistency
Simulation-based design optimization¶
We investigate optimization strategies for expensive numerical simulations, including:
- single- and multi-objective optimization
- surrogate-based optimization
- adaptive sampling
- many-fidelity and multi-fidelity strategies
- robust and uncertainty-aware optimization
Scientific machine learning¶
We study how machine learning can support engineering design through:
- surrogate modeling
- reduced-order modeling
- data-driven representation learning
- explainability for design decisions
- hybrid physics-data-driven approaches
Application domains¶
Typical application domains include:
- ship hull design
- marine propellers
- autonomous underwater vehicles
- bio-inspired marine systems
- airfoils and aerodynamic configurations
- multidisciplinary vehicle design
Philosophy¶
Our methodological approach emphasizes:
- physical consistency
- computational efficiency
- interpretability
- reproducibility
- open and reusable research software