FrontUQ 2024
​
Workshop on Frontiers of Uncertainty Quantification 2024
September 24-27 2024, Braunschweig, Germany
Tuesday
Session: UQ Software and Tutorials
(Chair: L. Seelinger)
09.00 - 10.00
Introduction to UM-Bridge and Software-Tools
10.00 - 10.30
UQ Problem used for Tutorials
Software Tutorial 1
11.00 - 12.00
N. Lüthen: Uncertainty Quantification with UQLab and UM-Bridge
Software Tutorial 2
13.00 - 14.00
C. Piazzola: The Sparse Grids Matlab Kit
Software Tutorial 3
14.30 - 15.30
S. Dolgov: TT Toolbox
Software Tutorial 4
16.00 - 17.00
C. Krill: UQpy 4.2: Scientific Machine Learning
Wednesday
08.15 - 08.30
Opening
Keynote I
08.30 - 09.30
J. Schaefer: Industry Perspective on UQ to Enable High-Fidelity Predictive Modeling for Aerospace Design and Analysis
Session I: UQ for Certification by Analysis & Digital Twins
(Chair: P. Bekemeyer)
10.00 - 10.30
D. Di Francesco: Towards risk-optimal certification by analysis
10.30 - 11.00
L. Werthen-Brabants: Towards Trustworthy Neural Networks for Certification by Analysis - Fuel Tank Flammability Reduction System
11.00 - 11.30
J. Unger: Uncertainty Quantification and Model Extension for Digital Twins through Model Bias Identification
11.30 - 12.00
D. Valente: Provenance-Driven Framework for Robust Aerospace System Performance
Keynote II
13.00 - 14.00
R. Tempone: Stochastic Optimization: Adaptive Variance Reduction and Bayesian Quasi-Newton Methods
Session II: Mathematical Methodologies for UQ I
(Chair: L. Seelinger)
14.00 - 14.30
T. Zhou: Information bottleneck based uncertainty quantification
14.30 - 15.00
E. Lovbak: Markov Chain Monte Carlo for Particle Solvers
15.00 - 15.30
B. Kent: Adaptive-in-time stochastic collocation approximation for parametric parabolic PDEs
Session III: Design under Uncertainties (Chair: U. Römer)
16.00 - 16.30
J. Parekh: Identification and Handling of Uncertainties in Computational Aerodynamics
16.30 - 17.00
S. Baars: Thompson sampling and partitioned surrogates for multidisciplinary design optimization
17.00 - 17.30
M. Alder: Probabilistic Technology Assessment of Complex Transportation Systems
Thursday
Keynote III
08.30 - 09.30
R. Dwight: Statistical methods for generalizable data-driven turbulence modelling
Session IV: Forward Propagation of Uncertainties
(Chair: P. Bekemeyer)
10.00 - 10.30
F. Lößle: Uncertainty Quantification in Aircraft Noise Calculation: Current Status and Challenges at DLR
10.30 - 11.00
H. Geisler: A new paradigm for engineering simulations under uncertainties: Time-separated Stochastic Mechanics
11.00 - 11.30
J. Bachner: Uncertainty Propagation for Multi-Hole Pneumatic Probes in Turbomachinery Flows
11.30 - 12.00
M. Pollak: Surrogate Modeling for Analysis and Design of Hollow Fiber Membrane Humidifiers for PEM Fuel Cells
Keynote IV
13.00 - 14.00
E. Ullmann: Rare event estimation with PDE-based models
Session V: Mathematical Methodologies for UQ II
(Chair: L. Seelinger)
14.00 - 14.30
L. Kluge: Efficient Bayesian Inference in Cosmological Simulations with Multilevel Delayed Acceptance
14.30 - 15.00
P. Hristov: Backcalculation for design under general uncertainty: An introduction and a tutorial
15.00 - 15.30
K. Tüting: A modeling perspective on tracing uncertainties in dynamic systems
Session VI: State Estimation and Monitoring under Uncertainties (Chair: U. Römer)
16.00 - 16.30
D. Pölzleitner: Feature and Extrapolation Aware Uncertainty Quantification for AI-based State Estimation
16.30 - 17.00
N. Dridi: Uncertainty Quantification Using Bayesian Neural Networks
17.00 - 17.30
D. Tyagi: Damage Localisation and Quantification from Modal Data using Sparsity Promoting Priors
Friday
Keynote V
09.00 - 10.00
R. Butler: Certification for Design: Re-shaping the Testing Pyramid for Composite Aerostructures
Session VII: Surrogate Modeling for UQ (Chair: P. Bekemeyer)
10.30 - 11.00
V. Narouie: Polynomial Chaos-based Statistical Finite Element Analysis with Non-Conjugate Prior
11.00 - 11.30
F. Zacchei: Multi-Fidelity Delayed Acceptance for PDE Inverse Problems with Progressive Neural Network Surrogates
11.30 - 12.00
D. Anton: Statistical calibration of constitutive models from full-field data using physics-informed neural networks