Abstract
When simulations are very expensive and many are required, as for optimization or uncertainty quantification, a way to reduce cost is using surrogates. With multiple simulations to predict the quantity of interest, some being very expensive and accurate (high-fidelity simulations) and others cheaper but less accurate (low-fidelity simulations), it may be worthwhile to use multifidelity surrogates (MFSs). Moreover, if we can afford just a few high-fidelity simulations or experiments, MFS becomes necessary. Co-Kriging, which is probably the most popular MFS, replaces both low-fidelity and high-fidelity simulations by a single MFS. A recently proposed linear regression–based MFS (LR-MFS) offers the option to correct the LF simulations instead of correcting the LF surrogate in the MFS. When the low-fidelity simulation is cheap enough for use in an application, such as optimization, this may be an attractive option. In this paper, we explore the performance of LR-MFS using exact and surrogate-replaced low-fidelity simulations. The problem studied is a cylindrical dispersal of 100-μ m-diameter solid particles after detonation and the quantity of interest is a measure of the amplification of the departure from axisymmetry. We find very substantial accuracy improvements for this problem using the LR-MFS with exact low-fidelity simulations. Inspired by these results, we also compare the performance of co-Kriging to the use of Kriging to correct exact low-fidelity simulations and find a similar accuracy improvement when simulations are directly used. For this problem, further improvements in accuracy are achievable by taking advantage of inherent parametric symmetries. These results may alert users of MFSs to the possible advantages of using exact low-fidelity simulations when this is affordable.
Similar content being viewed by others
Notes
In this case, multiphase refers to the phases, gas, and particles.
The multiplicative correction approach is an MFS option that is not included in this paper; however, the reader can refer to Fernández-Godino et al. (2019c) if interested. This MFS is constructed using as training points the quotient between yHF(x) and yLF(x) functions at the nested training data points.
Abbreviations
- δ(x):
-
Discrepancy function
- \(\hat {\delta }(\mathbf {x})\) :
-
Discrepancy function surrogate, also known as additive correction
- ρ :
-
Constant scaling factor
- y HF(x):
-
High-fidelity simulation
- \(\hat {y}_{HF}(\mathbf {x})\) :
-
High-fidelity surrogate
- y LF(x):
-
Low-fidelity simulation
- \(\hat {y}_{LF}(\mathbf {x})\) :
-
Low-fidelity surrogate
- \(\hat {y}_{\hat {add}}(\mathbf {x})\) :
-
Multifidelity surrogate that uses additive correction and where the prediction is performed using a low-fidelity surrogate
- \(\hat {y}_{\hat {comp}}(\mathbf {x})\) :
-
Multifidelity surrogate that uses comprehensive correction and where the prediction is performed using a low-fidelity surrogate
- \(\hat {y}_{add}(\mathbf {x})\) :
-
Multifidelity surrogate that uses additive correction and where the prediction is performed using low-fidelity simulations
- \(\hat {y}_{comp}(\mathbf {x})\) :
-
Multifidelity surrogate that uses comprehensive correction and where the prediction is performed using low-fidelity simulations
References
Annamalai S, Rollin B, Ouellet F, Neal C, Jackson TL, Balachandar S (2016) Effects of initial perturbations in the early moments of an explosive dispersal of particles. J Fluids Eng 138(7):070903. https://doi.org/10.1115/1.4030954
Blatman G, Sudret B (2010) Efficient computation of global sensitivity indices using sparse polynomial chaos expansions. Reliab Eng Syst Saf 95(11):1216–1229. https://doi.org/10.1016/j.ress.2010.06.015. http://www.sciencedirect.com/science/article/pii/S0951832010001493
Blatman G, Sudret B (2011) Adaptive sparse polynomial chaos expansion based on least angle regression. J Comput Phys 230(6):2345–2367. https://doi.org/10.1016/j.jcp.2010.12.021. http://www.sciencedirect.com/science/article/pii/S0021999110006856 http://www.sciencedirect.com/science/article/pii/S0021999110006856
Bouhlel MA, Hwang JT, Bartoli N, Lafage R, Morlier J, Martins JRRA (2019) A Python surrogate modeling framework with derivatives. Adv Eng Softw. https://doi.org/10.1016/j.advengsoft.2019.03.005
Cameron RH, Martin WT (1947) The orthogonal development of non-linear functionals in series of Fourier-Hermite functionals. Ann Math 48(2):385–392. http://www.jstor.org/stable/1969178
Cressie N (1993) Statistics for spatial data: Wiley series in probability and statistics. Wiley, New York. https://doi.org/10.1002/9781119115151
Dobrat B, Crawford P (1981) Handbook, LLNL explosives. Lawrence Livermore National Laboratory. OSTI Identifier 6530310
Dubreuil S, Berveiller M, Petitjean F, Salaün M (2014) Construction of bootstrap confidence intervals on sensitivity indices computed by polynomial chaos expansion. Reliab Eng Syst Saf 121(Supplement C):263–275. https://doi.org/10.1016/j.ress.2013.09.011. http://www.sciencedirect.com/science/article/pii/S0951832013002688
Fernández-Godino MG, Balachandar S, Haftka RT (2019a) On the use of symmetries in building surrogate models. J Mech Des 141(6):061402. https://doi.org/10.1115/1.4042047
Fernández-Godino MG, Ouellet F, Haftka R, Balachandar S (2019b) Early time evolution of circumferential perturbation of initial particle volume fraction in explosive cylindrical multiphase dispersion. J Fluids Eng 141:0913021–09130220. https://doi.org/10.1115/1.4043055
Fernández-Godino MG, Park C, Kim NH, Haftka RT (2019c) Issues in deciding whether to use multifidelity surrogates. AIAA J 57(5):2039–2054. https://doi.org/10.2514/1.J057750
Kennedy MC, O’Hagan A (2000) Predicting the output from a complex computer code when fast approximations are available. Biometrika 87(1):1–13. https://doi.org/10.1093/biomet/87.1.1
Le Gratiet L, Garnier J (2014) Recursive co-kriging model for design of computer experiments with multiple levels of fidelity. Int J Uncertain Quantif 365–386
Myers DE (1982) Matrix formulation of co-kriging. Math Geol 14(3):249–257. https://doi.org/10.1007/BF01032887
Ouellet F, Annamalai S, Rollin B (2017) Effect of a bimodal initial particle volume fraction perturbation in an explosive dispersal of particles. In: AIP Conference proceedings, vol 1793. AIP Publishing, p 150011. https://doi.org/10.1063/1.4971740
Queipo NV, Haftka RT, Shyy W, Goel T, Vaidyanathan R, Tucker PK (2005) Surrogate-based analysis and optimization. Prog Aerosp Sci 41(1):1–28. https://doi.org/10.1016/j.paerosci.2005.02.001
Seber GA, Lee AJ (2012) Linear regression analysis, vol 329. Wiley, New York. ISBN 978-0-471-41540-4
Sobol IM (2001) Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates. Math Comput Simul 55(1):271–280. http://www.sciencedirect.com/science/article/pii/S0378475400002706
Sudret B (2008) Global sensitivity analysis using polynomial chaos expansions. Reliab Eng Syst Saf 93(7):964–979. https://doi.org/10.1016/j.ress.2007.04.002. http://www.sciencedirect.com/science/article/pii/S0951832007001329. Bayesian Networks in Dependability
Vauclin R (2014) Développement de modèles réduits multifidélité en vue de l’optimisation de structures aéronautiques. In: Rapport Institut Supérieur de l’Aéronautique et de l’Espace – École nationale Supérieure des mines de Saint-Étienne
Zhang Y, Kim NH, Park C, Haftka RT (2018) Multifidelity surrogate based on single linear regression. AIAA J 56(12):4944–4952. https://doi.org/10.2514/1.J057299
Acknowledgments
We gratefully acknowledge the contribution of a reviewer, Professor Andy Keane, who suggested trying the additive Kriging.
Funding
This work was partially supported by the Center for Compressible Multiphase Turbulence, the U.S. Department of Energy, National Nuclear Security Administration, Advanced Simulation and Computing Program, as a Cooperative Agreement under the Predictive Science Academic Alliance Program, under Contract No. DE-NA0002378.
This work was partially supported by the French National Research Agency (ANR) through the ReBReD project under grant ANR-16-CE10-0002 and by a ONERA internal project MUFIN dedicated about multi-fidelity.
This work was partially performed under U.S. Government contract 89233218CNA000001 for Los Alamos National Laboratory (LANL), which is operated by Triad National Security, LLC for the U.S. Department of Energy/National Nuclear Security Administration. Approved for public release LA-UR-19-22491.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Responsible Editor: Shapour Azarm
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendices
Appendix A: Identification of the significant variables
1.1 1.1 Methodology
The objective of this appendix is to present the sensitivity analysis of the quantity of interest studied in the paper (the normalized effective Fourier effective perturbation described in Section 3) with respect to the variables used to parametrized the PVF perturbation and described in Section 2 by (1). We recall that this perturbation is modeled by:
with the constraint:
Moreover, as the variables Φ1, Φ2, and Φ3 are used to model the phase shift, one can arbitrary set Φ1 = 0 and consider the phase shift with respect to the first mode. As a consequence, the problem counts seven independent variables concatenated into the vector \(X = \left \lbrace A_{1},\! A_{2},\!k_{1},k_{2},\!k_{3},{\varPhi }_{2},{\varPhi }_{3}\right \rbrace \). The quantity of interest can thus be expressed as:
in which the mapping \({\mathscr{M}}\) involves the numerical resolution of the multiphase explosion and the post-processing of the solution as explained in Sections 2 and 3.
The methodology used to study the sensitivity of the quantity of interest F to the input parameter X is to perform a global sensitivity analysis by computing variance-based sensitivity indices. Consequently, it is assumed that X is a vector of seven independent random variables (the probability distribution will be discussed in the following) and F is a random variable of unknown probability distribution. Then, the approach proposed in Sudret (2008) and improved in Blatman and Sudret (2010) for the computation of sensitivity indices (Sobol’ indices 2001) by sparse polynomial chaos expansion (PCE) (Blatman and Sudret 2011) is applied. Assuming that F is a second-order random variable, it can be shown (Cameron and Martin 1947) that:
where \(\left \lbrace \phi _{i} \right \rbrace _{i\in \mathbb {N}}\) is a polynomial basis orthogonal with respect to the probability distribution of X and Ci are unknown coefficients.
Sparse PCE consists in the construction of a sparse polynomial basis \(\left \lbrace \phi _{i} \right \rbrace _{\alpha \in {\mathscr{A}}}\), where α = (α1,⋯ , αn) is a multi-index used to identify the polynomial acting with the power αi on the variable Xi and \({\mathscr{A}}\) is a set of indices α. In practice, \({\mathscr{A}}\) is a subset of the set \({\mathscr{B}}\) which contains all the indices α up to a dimension d, i.e., \(card({\mathscr{B}})=\frac {(d+n)!}{d!n!}\). The objective of sparse approach is to find an accurate polynomial basis \(\left \lbrace \phi _{i} \right \rbrace _{\alpha \in {\mathscr{A}}}\) such that \(card({\mathscr{A}})<<card({\mathscr{B}})\). In the present case, this is achieved by Least Angle Regression, i.e., unknown coefficients Ci are computed by iteratively solving a mean square problem and selecting, at each iteration, the polynomial which is the most correlated with the residual (see Blatman and Sudret (2011) for details).
Finally, one gets the following approximation:
from which the sensitivity index can be derived. Indeed, the orthogonality of the polynomial basis \(\left \lbrace \phi _{i} \right \rbrace _{\alpha \in {\mathscr{A}}}\) allows to write the expectation and the variance in the following form:
In addition, the idea pointed out in Sudret (2008) is to identify the PCE with the ANOVA decomposition, from which one can show that, the first-order sensitivity index of the variable Xi reads:
where \(L_{i}=\left \lbrace \alpha \in {\mathscr{A}},~ \forall ~j\neq i~\alpha _{j}=0 \right \rbrace \), i.e., only the polynomials acting exclusively on the variable Xi are considered.
The total sensitivity index is also available by:
where \(L^{+}_{i}=\left \lbrace \alpha \in {\mathscr{A}},~ \alpha _{i}\neq 0 \right \rbrace \), i.e., all the polynomials acting on the variable Xi are considered (allows considering interactions between Xi and the other variables).
One can note that the approximation of the sensitivity index obtained by sparse PCE relies on an accurate approximation of the surrogate response by the sparse PCE; however, the link between the accuracy of the PCE approximation and the accuracy of the approximated sensitivity index is not straightforward. In order to access the quality of the sensitivity index computed by PCE, a bootstrap approach proposed in Dubreuil et al. (2014) is set up and detailed in the next section.
1.1.1 1.2 Application to the high-fidelity computation of normalized Fourier effective perturbation
The probabilistic surrogate of the seven independent input parameters is detailed in Table 4. Uniform distributions are assumed for each component of the random vector X and variability ranges are defined, which basically define the domain of interest for the sensitivity analysis.
Then, a design of experiments of 711 points is drawn by LHS in order to estimate the sensitivity index by sparse PCE. The maximum order of the polynomials is set to d = 4. In order to assess the accuracy of the obtained sensitivity index, the following bootstrap procedure is proposed. Among the 711 points, 611 are used as a training set to compute the PCE approximation (least angle regression approach) and 100 are used as a validation set. The training and validation sets are randomly chosen among the 711 points. The bootstrap approach consists in repeating B times this procedure changing each time the training and the validation set. This leads to B different PCE approximations and thus to a sample of B sensitivity indices. This sample is further used to estimate the coefficient of variation of the sensitivity index estimators obtained by sparse PCE. Moreover, for each bootstrap sample, the relative L2 norm of the relative error (𝜖2) is computed on the validation set as well as the coefficient of determination (R2) computed on the training set.
Over the B bootstrap repetitions, the estimated mean values are E[𝜖2] ≈ 1.54 × 10− 3 and E[R2] ≈ 9.74 × 10− 1 and the coefficients of variation are \(cv_{\epsilon ^{2}}\approx 1.57\times 10^{-1}\) and \(cv_{R^{2}}\approx 1.86\times 10^{-3}\). These first results allowed confidence in the accuracy of the PCE approximation. Note that the relatively large coefficient of variation \(cv_{\epsilon ^{2}}\) should be considered with respect to its very low mean value. Concerning the sensitivity index, Table 5 presents the first order and total index with the mean values and coefficient of variation estimated by bootstrap.
First of all, the results presented in Table 5 show that when a sensitivity index has a significant value (values highlighted in bold) its estimation is quite accurate as the coefficient of variation is relatively low (less than 3%). One can also note that for low values of sensitivity indices the coefficients of variation are quite large; however, as these sensitivity indices are, at least, two orders of magnitude lower than the significant ones, their poor estimation is not detrimental for the purpose of sensitivity analysis. With respect to the results of the sensitivity analysis, one can conclude that the variance of the quantity of interest is mainly driven by the first five variables namely A1, A2, k1, k2, and k3. It is also interesting to note the strong interaction between the amplitude variables (that have almost no first-order effect) with the wavenumbers which is consistent with the shape of the perturbation (in (1) the interactions between A1, k1 and A2, k2 clearly appear) and the constraint on the amplitude (2), which explains the interaction between A1, A2, and k3, and justifies that \(\hat {S}_{T_{k_{3}}}\) has the highest value.
Based on this result, it has been decided to consider only the five variables A1, A2, k1, k2, and k3 for the surrogate construction of the quantity of interest F.
Appendix B: Percentage contribution of low-fidelity and high-fidelity data in LR-MFS using up to second-order polynomial basis functions
The contribution of the LF data points to the MFSs is studied in order to understand why the performance of the surrogates that use yLF instead of the LF surrogate, \(\hat {y}_{LF}\), for prediction, worked overwhelmingly better for the metric F. First, notice that in our case n = 21 in (8), which is the number of coefficients of a quadratic polynomial in five variables. Also, note that for prediction, we can choose between using yLF or \(\hat {y}_{LF}\); however, for training purposes, yLF is used. Therefore, ρ and \(\hat {\delta }(x)\) coefficients in (7) and in (14) are identical. Equation (8) shows explicitly the contribution of each surrogate, HF and LF, to the comprehensive MF approximation using the \(\hat {y}_{LF}\); however, we can choose to use yLF, i.e., the LF simulations directly, instead. The first term represents the contribution of the LF simulation, \(\rho \hat {y}_{LF}(x)\) (using LF surrogate) or ρyLF(x) (using LF simulations), to the MFS. The second term, \({\sum }_{1=1}^{p} X_{i}(x) b_{i}\), represents the contribution of the HF simulation to the MFS.
Figure 8 shows the mean contribution in percentage of the LF and HF information to the comprehensive MFS predicted using yLF, i.e., \(\hat {y}_{comp}\). Using yLF or \(\hat {y}_{LF}\) for predicting the MF comprehensive correction gives the same LF and HF mean contributions; therefore, only one plot was included. The mean contribution was calculated averaging the contribution of each of the 92 validation points. The contribution of the LF model is defined as \(\rho \hat {y}_{LF}\), while the contribution of the HF model is defined as \({\sum }_{1=1}^{p} X_{i}(x) b_{i}\). A negative percentage indicates, in this case, that the contribution has a negative sign. Values higher than 100% indicate that the mean of the contribution is higher than the metric value on average, i.e., calculating the term \(\rho \hat {y}_{LF}\) for each of the validation points and then taking the average. Therefore, ρ does not need to be higher than 1; instead, the term \(\rho \hat {y}_{LF}\) needs to be higher than 1 on average. Naturally, the sum of the LF and HF contributions adds to 100%. For more than 50 HF data points, the contribution is dominated by the LF information (≈ 85%). This helps explain why when the LF simulation output is used directly (without constructing a surrogate), the improvement is substantial in Fig. 4. That is, for metric F, the results of the MF comprehensive prediction using yLF, \(\hat {y}_{comp}\), perform substantially better than the ones that use the LF surrogate, \(\hat {y}_{\hat {comp}}\). When the source of LF is changed for prediction instead of a surrogate, the results change drastically, which is expected due to the high correlation between HF and LF simulations (0.92).
Rights and permissions
About this article
Cite this article
Fernández-Godino, M.G., Dubreuil, S., Bartoli, N. et al. Linear regression-based multifidelity surrogate for disturbance amplification in multiphase explosion. Struct Multidisc Optim 60, 2205–2220 (2019). https://doi.org/10.1007/s00158-019-02387-4
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00158-019-02387-4