Abstract
This work considers the challenging problem of identifying the statistical properties of random fields from indirect observations. To this end, a Bayesian approach is introduced, whose key step is the nonparametric approximation of the likelihood function from limited information. When the likelihood function is based on the evaluation of an expensive computer code, this work also proposes a method to select iteratively new design points to reduce the uncertainties on the results that are due to the approximation of the likelihood. Two applications are finally presented to illustrate the efficiency of the proposed procedure: a first one based on analytic data, and a second one dealing with the identification of the random elasticity field of an heterogeneous microstructure.
Similar content being viewed by others
References
Akaike H (1974) A new look at the statistical model identification. IEEE Trans Autom Control 19(6):716–723
Arnst M, Ghanem R, Soize C (2010) Identification of bayesian posteriors for coefficients of chaos expansions. J Comput Phys 229(9):3134–3154
Atwell J, King B (2001) Proper orthogonal decomposition for reduced basis feedback controllers for parabolic equations. Math Comput Modell 33(1–3):1–19
Auffray Y, Barbillon P, Marin JM (2012) Maximin design on non hypercube domains and kernel interpolation. Stat Comput 22(3):703–712
Bilionis I, Zabaras N (2015) Bayesian uncertainty propagation using gaussian processes. In: Ghanem R, Higdon D, Owhadi H (eds) Handbook of uncertainty quantification. Springer, New York
Box G, Jenkins G (1970) Time series analysis: forecasting and control. Holden-Day, San Francisco
Chen P, Schwab C (2015) Sparse-grid, reduced basis bayesian inversion. Comput Methods Appl Mech Eng 297:84–115
Clouteau D, Cottereau R, Lombaert G (2013) Dynamics of structures coupled with elastic media—a review of numerical models and methods. J Sound Vib 332:2415–2436
Conrad PR, Marzouk YM, Pillai NS, Smith A (2016) Accelerating asymptotically exact MCMC for computationally intensive models via local approximations. J Am Stat Assoc 111:1591–1607
Conrad PR, Davis A, Marzouk YM, Pillai NS, Smith A (2018) Parallel local approximation MCMC for expensive models. SIAM/ASA J Uncertain Quantif 6(1):39–373
Damblin G, Barbillonz P, Keller M, Pasanisi A, Parent E (2013) Adaptive numerical designs for the calibration of computer codes. SIAM/ASA J Uncertain Quantif 6(1):151–179
Draguljić D, Santner TJ, Dean AM (2012) Noncollapsing space-filling designs for bounded nonrectangular regions. Technometrics 54(2):169–178
Emery J, Grigoriu M, Field RF Jr (2016) Bayesian methods for characterizing unknown parameters of material models. Appl Math Model 13–14:6395–6411
Fang K, Li R, Sudjianto A (2006) Design and modeling for computer experiments. Computer science and data analysis series. Chapman & Hall, London
Fang K, Lin D (2003) Uniform experimental designs and their applications in industry. Handb Stat 22:131–178
Fielding M, Nott DJ, Liong SY (2011) Efficient MCMC schemes for computationally expensive posterior distributions. Technometrics 53(1):16–28
Ghanem R, Spanos PD (2003) Stochastic finite elements: a spectral approach, rev edn. Dover Publications, New York
Guilleminot J, Soize C (2013) On the statistical dependence for the components of random elasticity tensors exhibiting material symmetry properties. J Elast 111:109–130
Higdon D, Lee H, Holloman C (2003) Markov chain monte carlo based approaches for inference in computationally intensive inverse problems. Bayesian Stat 7:181–197
Higdon D, Gattiker J, Williams B, Rightley M (2008) Computer model calibration using high-dimensional output. J Am Stat Assoc 103(482):570–583
Joseph VR, Gul E, Ba S (2015) Maximum projection designs for computer experiments. Biometrika 102(2):371–380
Kennedy M, O’Hagan A (2001) Bayesian calibration of computer models. J R Stat Soc 63:425–464
Lai WM, Rubin D, Krempl E (2010) Introduction to continuum mechanics. Elsevier, Inc, Amsterdam
Le Maître O, Knio O (2010) Spectral methods for uncertainty quantification. Springer, New York
Lekivetz R, Jones B (2015) Fast flexible space-filling designs for nonrectangular regions. Qual Reliab Eng Int 31(5):829–837
Li J, Marzouk YM (2014) Adaptive construction of surrogates for the bayesian solution of inverse problems. SIAM J Sci Comput 36:A1163–A1186
Mak S, Joseph VR (2018) Minimax and minimax projection designs using clustering. J Comput Graph Stat 27(1):166–178. https://doi.org/10.1080/10618600.2017.1302881
Marin JM, Robert CP (2007) Bayesian core. Springer, New York
Marzouk YM, Najm HN (2009) Dimensionality reduction and polynomial chaos acceleration of bayesian inference in inverse problems. J Comput Phys 228(6):1862–1902
Marzouk YM, Xiu D (2009) A stochastic collocation approach to bayesian inference in inverse problems. Commun Comput Phys 6:826–847
Matthies H, Zander E, Rosi B, Litvinenko A (2016) Parameter estimation via conditional expectation: a Bayesian inversion. Adv Model Simul Eng Sci 3(1):24
McKay M, Beckman R, Conover W (1979) A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics 21:239–245
Nguyen MT, Desceliers C, Soize C, Allain J, Gharbi H (2015) Multiscale identification of random elasticity field at mesoscale of a heterogeneous microstructure using multiscale experimental observations. J Multiscale Comput Eng 13(4):281–295
Nouy A (2010) Proper generalized decomposition and separated representations for the numerical solution of high dimensional stochastic problems. Arch Comput Methods Eng 17:403–434
Nouy A, Soize C (2014) Random fields representations for stochastic elliptic boundary value problems and statistical inverse problems. Eur J Appl Math 25:339–373
Perrin G (2019) Adaptive calibration of a computer code with time-series output. Reliab Eng Syst Saf. https://doi.org/10.1016/j.ress.2019.106728
Perrin G, Cannamela C (2017) A repulsion-based method for the definition and the enrichment of opotimized space filling designs in constrained input spaces. J Soc Fr Stat 158(1):37–67
Perrin G, Soize C, Duhamel D, Funfschilling C (2012) Identification of polynomial chaos representations in high dimension from a set of realizations. SIAM J Sci Comput 34(6):2917–2945
Perrin G, Soize C, Duhamel D, Funfschilling C (2013) Karhunen–loéve expansion revisited for vector-valued random fields: scaling, errors and optimal basis. J Comput Phys 242:607–622
Perrin G, Soize C, Duhamel D, Funfschilling C (2014) A posteriori error and optimal reduced basis for stochastic processes defined by a finite set of realizations. SIAM/ASA J Uncertain Quantif 2:745–762
Perrin G, Soize C, Marque-Pucheu S, Garnier J (2017) Nested polynomial trends for the improvement of gaussian process-based predictors. J Comput Phys 346:389–402
Perrin G, Soize C, Ouhbi N (2018) Data-driven kernel representations for sampling with an unknown block dependence structure under correlation constraints. J Comput Stat Data Anal 119:139–154
Rasmussen CE (2003) Gaussian processes to speed up hybrid monte carlo for expensive bayesian integrals. Bayesian Stat 7:651–659
Rubinstein RT, Kroese D (2008) Simulation and the Monte Carlo method. Wiley, Hoboken
Santner TJ, Williams B, Notz W (2003) The design and analysis of computer experiments. Springer, New York
Scott DW, Sain SR (2004) Multidimensional density estimation. In: Handbook of statistics, vol 24, pp 229–261
Sinsbeck M, Nowak W (2017) Sequential design of computer experiments for the solution of Bayesian inverse. SIAM/ASA J Uncertain Quantif 5:640–664
Soize C (2006) Non-gaussian positive-definite matrix-valued random fields for elliptic stochastic partial differential operators. Comput Methods Appl Mech Eng 195:26–64
Soize C (2008) Tensor-valued random fields for meso-scale stochastic model of anisotropic elastic microstructure and probabilistic analysis of representative volume element size. Probab Eng Mech 23:307–323
Soize C (2010) Identification of high-dimension polynomial chaos expansions with random coefficients for non-Gaussian tensor-valued random fields using partial and limited experimental data. Comput Methods Appl Mech Eng 199(33–36):2150–2164
Soize C (2011) A computational inverse method for identification of non-Gaussian random fields using the Bayesian approach in very high dimension. Comput Methods Appl Mech Eng 200(45–46):3083–3099
Soize C, Ghanem R (2016) Data-driven probability concentration and sampling on manifold. J Comput Phys 321(September 2015):242–258
Soize C, Ghanem R (2017) Probabilistic learning on manifold for optimization under uncertainties. Proc Uncecomp 2017:1–15
Stinstra E, den Hertog D, Stehouwer P, Vestjens A (2003) Constrained maximin designs for computer experiments. Technometrics 45(4):340–346
Stinstra E, den Hertog D, Stehouwer P, Vestjens A (2010) Uniform designs over general input domains with applications to target region estimation in computer experiments. Comput Stat Data Anal 51(1):219–232
Stuart AM (2010) Inverse problems: a bayesian perspective. Acta Numerica 19:451–559
Tian M, Li D, Cao Z, Phoon K, Wang Y (2016) Bayesian identification of random field model using indirect test data. Eng Geol 210:197–211
Tsilifis P, Ghanem RG, Hajali P (2017) Efficient Bayesian experimentation using an expected information gain lower bound. SIAM/ASA J Uncertain Quantif 5:30–62
Wan J, Zabaras N (2011) A Bayesian approach to multiscale inverse problems using the sequential Monte Carlo method. Inverse Probl 27:105004
Wand MP, Jones MC (1995) Kernel smoothing. Encycl Stat Behav Sci 60(60):212
Whittle P (1951) Hypothesis testing in time series, Ph.D. thesis. University of Uppsala
Whittle P (1983) Prediction and regulation by linear least-square methods. University of Minnesota Press, Minneapolis
Williams M (2011) The eigenfunctions of the Karhunen–Loeve integral equation for a spherical system. Propab Eng Mech 26:202–207
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
1.1 A.1. Proof of the equality of Eq. 9
Let \(\varvec{A},\varvec{B},\varvec{D}\) be the block decomposition matrices of \(\widehat{\varvec{C}}^{-1}\):
Using the Schur complement, if follows that:
It comes
This leads to the searched result.
Rights and permissions
About this article
Cite this article
Perrin, G., Soize, C. Adaptive method for indirect identification of the statistical properties of random fields in a Bayesian framework. Comput Stat 35, 111–133 (2020). https://doi.org/10.1007/s00180-019-00936-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00180-019-00936-5