Nonlinear state-space models are ubiquitous in modeling real-world dynamical systems. Sequential Monte Carlo (SMC) techniques, also known as particle methods, are a well-known class of parameter estimation methods for this general class of state-space models. Existing SMC-based techniques rely on excessive sampling of the parameter space, which makes their computation intractable for large systems or tall data sets. Bayesian optimization techniques have been used for fast inference in state-space models with intractable likelihoods. These techniques aim to find the maximum of the likelihood function by sequential sampling of the parameter space through a single SMC approximator. Various SMC approximators with different fidelities and computational costs are often available for sample-based likelihood approximation. In this paper, we propose a multi-fidelity Bayesian optimization algorithm for the inference of general nonlinear state-space models (MFBO-SSM), which enables simultaneous sequential selection of parameters and approximators. The accuracy and speed of the algorithm are demonstrated by numerical experiments using synthetic gene expression data from a gene regulatory network model and real data from the VIX stock price index.

Calculation of phase diagrams is one of the fundamental tools in alloy design-more specifically under the framework of Integrated Computational Materials Engineering. Uncertainty quantification of phase diagrams is the first step required to provide confidence for decision making in property-or performance-based design. As a manner of illustration, a thorough probabilistic assessment of the CALPHAD model parameters is performed against the available data for a Hf-Si binary case study using a Markov Chain Monte Carlo sampling approach. The plausible optimum values and uncertainties of the parameters are thus obtained, which can be propagated to the resulting phase diagram. Using the parameter values obtained from deterministic optimization in a computational thermodynamic assessment tool (in this case Thermo-Calc) as the prior information for the parameter values and ranges in the sampling process is often necessary to achieve a reasonable cost for uncertainty quantification. This brings up the problem of finding an appropriate CALPHAD model with high-level of confidence which is a very hard and costly task that requires considerable expert skill. A Bayesian hypothesis testing based on Bayes' factors is proposed to fulfill the need of model selection in this case, which is applied to compare four recommended models for the Hf-Si system. However, it is demonstrated that information fusion approaches, i.e., Bayesian model averaging and an error correlation-based model fusion, can be used to combine the useful information existing in all the given models rather than just using the best selected model, which may lack some information about the system being modelled.

No abstract

No abstract

No abstract

No abstract

Available computational models for many engineering design applications are both expensive and and of a black-box nature. This renders traditional optimization techniques difficult to apply, including gradient-based optimization and expensive heuristic approaches. For such situations, Bayesian global optimization approaches, that both explore and exploit a true function while building a metamodel of it, are applied. These methods often rely on a set of alternative candidate designs over which a querying policy is designed to search. For even modestly high-dimensional problems, such an alternative set approach can be computationally intractable, due to the reliance on excessive exploration of the design space. To overcome this, we have developed a framework for the optimization of expensive black-box models, which is based on active subspace exploitation and a two-step knowledge gradient policy. We demonstrate our approach on three benchmark problems and a practical aerostructural wing design problem, where our method performs well against traditional direct application of Bayesian global optimization techniques.

This paper presents a novel compositional multidisciplinary uncertainty analysis methodology for systems with feedback couplings, and model discrepancy. Our approach incorporates aspects of importance resampling, density estimation, and Gibbs sampling to ensure that, under mild assumptions, our method is provably convergent in distribution. A key feature of our approach is that disciplinary models can all be executed offline and independently. Offline data is synthesized in an online phase that does not require any further model evaluations or any full coupled system level evaluations. We demonstrate our approach on a simple aerodynamics-structures system.