Selected work

Representative Papers

Neural Operators for Immersed-Boundary Soft Swimmers Locomotion

Aug 07, 2026

This work addresses the high computational cost of high-fidelity immersed boundary simulations for fluid–structure interaction in soft-bodied swimmers, which hinders repeated evaluations required for design optimization and control. To overcome this limitation, the authors propose a neural operator-based surrogate model trained on regular-grid data generated from adaptive fluid–structure interaction simulations. Conditioned on swimmer geometry and Reynolds number, the model enables temporally resolved prediction of unsteady flow fields around two- and three-dimensional eel-like swimmers. This study presents the first neural surrogate capable of full-field, multi-physics prediction—including velocity, vorticity, and pressure—for moving-boundary swimmers, demonstrating strong generalization across geometries and parameters. The two-dimensional model achieves a mean relative L² error of 3.51% on five extrapolated high-Reynolds-number trajectories, while the three-dimensional model yields errors of 3.44% (velocity), 5.58% (vorticity), and 19.2% (pressure) on interpolated trajectories.

0 citationsRead paper

The role of physical models in the validation and calibration of numerical models -- The example of the Lillebælt Bridge

Apr 28, 2026

This study addresses the growing marginalization of physical models by reasserting their value in engineering research and education, using a scaled model of the Lillebet Bridge to tackle challenges in validating and calibrating complex numerical models. Through operational modal analysis, the experimentally determined natural frequencies and damping ratios yielded high-fidelity dynamic characteristics. These data effectively supported the calibration and validation of the corresponding numerical model, underscoring the critical role of physical models as a bridge between theory and experimentation. The findings provide a reliable benchmark for hybrid simulation methodologies, reaffirming the indispensable contribution of physical modeling to structural dynamics research and pedagogy.

0 citationsRead paper

Data Driven Calibration of Analytical Concrete Creep Models Considering Preloading Effects Using Gaussian Processes

Apr 28, 2026

Traditional concrete creep models struggle to accurately capture the influence of preloading on both the magnitude and variability of creep, leading to insufficient predictive accuracy. This study proposes a data-driven calibration approach based on Gaussian process regression that, for the first time, systematically incorporates preloading intensity, application timing, and concrete age into an analytical creep model. The proposed method not only significantly enhances prediction accuracy but also enables quantification of predictive uncertainty and supports optimal experimental design. By doing so, it provides a robust theoretical foundation for the sustainable design of concrete structures.

0 citationsRead paper

Assessing the dynamic response of long-span bridges under simultaneous wind and traffic loads

Apr 27, 2026

This study addresses the limitations of conventional approaches that treat wind and traffic loads in isolation, thereby neglecting their nonlinear coupling effects on the dynamic response of long-span bridges. A comprehensive vehicle–wind–bridge coupled dynamic model is developed, integrating Kármán wind spectra with Davenport coherence for turbulent wind fields, ISO road roughness profiles, three-dimensional vehicle dynamics, and quasi-steady aerodynamic theory to perform time-history analyses. Moving beyond the assumption of linear superposition, the research uncovers the nonlinear interaction mechanisms under multi-source loading and accurately quantifies bridge response characteristics. The findings establish a new paradigm and provide a scientific foundation for evaluating service performance, formulating serviceability criteria, and optimizing structural design of long-span bridges.

0 citationsRead paper

K$α$LOS finds Consensus: A Meta-Algorithm for Evaluating Inter-Annotator Agreement in Complex Vision Tasks

Mar 28, 2026

This work addresses the limitations of current visual task benchmarks, which are often hindered by annotation noise and unable to disentangle genuine model improvements from inconsistencies in labeling. Conventional evaluation metrics further fail to account for spatial correspondences among instances. To overcome these challenges, the authors propose KαLOS, a meta-algorithm grounded in a “localize-then-evaluate” principle that reformulates complex vision tasks into nominal reliability matrices. KαLOS employs a data-driven approach to calibrate localization parameters tailored to diverse tasks, enabling the first standardized and interpretable assessment of annotation consistency in complex visual settings. The method supports fine-grained diagnostics—such as annotator vigor and collaborative clustering—without relying on heuristic assumptions or circular validation. Experiments demonstrate that KαLOS robustly distinguishes signal from noise across multiple vision tasks, establishing a reliable standard for data quality evaluation in modern computer vision benchmarks.

0 citationsRead paper

Recent publications

Latest Papers

Neural Operators for Immersed-Boundary Soft Swimmers Locomotion

Aug 07, 2026

This work addresses the high computational cost of high-fidelity immersed boundary simulations for fluid–structure interaction in soft-bodied swimmers, which hinders repeated evaluations required for design optimization and control. To overcome this limitation, the authors propose a neural operator-based surrogate model trained on regular-grid data generated from adaptive fluid–structure interaction simulations. Conditioned on swimmer geometry and Reynolds number, the model enables temporally resolved prediction of unsteady flow fields around two- and three-dimensional eel-like swimmers. This study presents the first neural surrogate capable of full-field, multi-physics prediction—including velocity, vorticity, and pressure—for moving-boundary swimmers, demonstrating strong generalization across geometries and parameters. The two-dimensional model achieves a mean relative L² error of 3.51% on five extrapolated high-Reynolds-number trajectories, while the three-dimensional model yields errors of 3.44% (velocity), 5.58% (vorticity), and 19.2% (pressure) on interpolated trajectories.

0 citationsRead paper

The role of physical models in the validation and calibration of numerical models -- The example of the Lillebælt Bridge

Apr 28, 2026

This study addresses the growing marginalization of physical models by reasserting their value in engineering research and education, using a scaled model of the Lillebet Bridge to tackle challenges in validating and calibrating complex numerical models. Through operational modal analysis, the experimentally determined natural frequencies and damping ratios yielded high-fidelity dynamic characteristics. These data effectively supported the calibration and validation of the corresponding numerical model, underscoring the critical role of physical models as a bridge between theory and experimentation. The findings provide a reliable benchmark for hybrid simulation methodologies, reaffirming the indispensable contribution of physical modeling to structural dynamics research and pedagogy.

0 citationsRead paper

Data Driven Calibration of Analytical Concrete Creep Models Considering Preloading Effects Using Gaussian Processes

Apr 28, 2026

Traditional concrete creep models struggle to accurately capture the influence of preloading on both the magnitude and variability of creep, leading to insufficient predictive accuracy. This study proposes a data-driven calibration approach based on Gaussian process regression that, for the first time, systematically incorporates preloading intensity, application timing, and concrete age into an analytical creep model. The proposed method not only significantly enhances prediction accuracy but also enables quantification of predictive uncertainty and supports optimal experimental design. By doing so, it provides a robust theoretical foundation for the sustainable design of concrete structures.

0 citationsRead paper

Assessing the dynamic response of long-span bridges under simultaneous wind and traffic loads

Apr 27, 2026

This study addresses the limitations of conventional approaches that treat wind and traffic loads in isolation, thereby neglecting their nonlinear coupling effects on the dynamic response of long-span bridges. A comprehensive vehicle–wind–bridge coupled dynamic model is developed, integrating Kármán wind spectra with Davenport coherence for turbulent wind fields, ISO road roughness profiles, three-dimensional vehicle dynamics, and quasi-steady aerodynamic theory to perform time-history analyses. Moving beyond the assumption of linear superposition, the research uncovers the nonlinear interaction mechanisms under multi-source loading and accurately quantifies bridge response characteristics. The findings establish a new paradigm and provide a scientific foundation for evaluating service performance, formulating serviceability criteria, and optimizing structural design of long-span bridges.

0 citationsRead paper

K$α$LOS finds Consensus: A Meta-Algorithm for Evaluating Inter-Annotator Agreement in Complex Vision Tasks

Mar 28, 2026

This work addresses the limitations of current visual task benchmarks, which are often hindered by annotation noise and unable to disentangle genuine model improvements from inconsistencies in labeling. Conventional evaluation metrics further fail to account for spatial correspondences among instances. To overcome these challenges, the authors propose KαLOS, a meta-algorithm grounded in a “localize-then-evaluate” principle that reformulates complex vision tasks into nominal reliability matrices. KαLOS employs a data-driven approach to calibrate localization parameters tailored to diverse tasks, enabling the first standardized and interpretable assessment of annotation consistency in complex visual settings. The method supports fine-grained diagnostics—such as annotator vigor and collaborative clustering—without relying on heuristic assumptions or circular validation. Experiments demonstrate that KαLOS robustly distinguishes signal from noise across multiple vision tasks, establishing a reliable standard for data quality evaluation in modern computer vision benchmarks.

0 citationsRead paper