# Inverse problems for PDEs and other spatial models

## a.k.a. Bayesian calibration, model uncertainty

Inverse problems where the model is full of spatioemtporal correlations. I am particularly thinking about this in the context of PDE solvers, particularly approximate ones.

Suppose I have a PDE, possibly with some unknown parameters in the driving equation. All being equal I can do not too badly at approximating that with tools already mentioned. What if I wish simultaneously infer some unknown inputs? Then we consider it as an inverse problem. This is not quite the same as the predictive problem that many of the methods consider. However, we are free to use simulation-based inference to solve, or MCMC methods to do so for any of the forward-operator-learning approaches. To train the model to solve target the inverse problem directly, we might consider GANs or variational inference. At this point we are more or less required to start thinking about this in a probabilistic network or we will miss essential uncertainty quantification.

We are also encouraged to think about this as an approximation problem. We are surprised to find anything like a clean closed-form solution for the posterior distribution of some parameter in a PDE. Worse, the hypothetical solution is probably not even particularly computationally convenient. Why would it be? This will be especially trick in Bayesian inversion.

As far as how we might proceed, Liu, Yeo, and Lu (2020) is one approach, generalizing the approach of F. Sigrist, KΓΌnsch, and Stahel (2015b), but for advection/diffusion equations specifically. Generic methods include Bao et al. (2020); Jo et al. (2019); Lu, Mao, and Meng (2019); Raissi, Perdikaris, and Karniadakis (2019); Tait and Damoulas (2020); Xu and Darve (2020); Yang, Zhang, and Karniadakis (2020); Zhang, Guo, and Karniadakis (2020); Zhang et al. (2019).

## Gaussian process case

Alexanderian (2021) states a βwell-knownβ result, that the solution of a Bayesian linear inverse problem with Gaussian prior and noise models is a Gaussian posterior measure $$\mu_{\text {post }}^{y}=\mathcal{N}\left(m_{\text {MAP }}, \mathcal{C}_{\text {post }}\right)$$, where $\mathcal{C}_{\text {post }}=\left(\mathcal{F}^{*} \boldsymbol{\Gamma}_{\text {noise }}^{-1} \mathcal{F}+\mathcal{C}_{\text {pr }}^{-1}\right)^{-1} \quad \text { and } \quad m_{\text {MAP }}=\mathcal{C}_{\text {post }}\left(\mathcal{F}^{*} \boldsymbol{\Gamma}_{\text {noise }}^{-1} \boldsymbol{y}+\mathcal{C}_{\text {pr }}^{-1} m_{\text {MAP }}\right)$

## References

Alexanderian, Alen. 2021. arXiv:2005.12998 [Math], January.
Bao, Gang, Xiaojing Ye, Yaohua Zang, and Haomin Zhou. 2020. Inverse Problems 36 (11): 115003.
Brehmer, Johann, Gilles Louppe, Juan Pavez, and Kyle Cranmer. 2020. Proceedings of the National Academy of Sciences 117 (10): 5242β49.
Cotter, S. L., M. Dashti, and A. M. Stuart. 2010. SIAM Journal on Numerical Analysis 48 (1): 322β45.
Cranmer, Kyle, Johann Brehmer, and Gilles Louppe. 2020. Proceedings of the National Academy of Sciences, May.
Dashti, Masoumeh, and Andrew M. Stuart. 2015. arXiv:1302.6989 [Math], July.
Franklin, Joel N. 1970. Journal of Mathematical Analysis and Applications 31 (3): 682β716.
Grigorievskiy, Alexander, Neil Lawrence, and Simo SΓ€rkkΓ€. 2017. In arXiv:1610.08035 [Stat].
Jo, Hyeontae, Hwijae Son, Hyung Ju Hwang, and Eunheui Kim. 2019. arXiv:1907.12925 [Cs, Math], July.
Kaipio, Jari, and E. Somersalo. 2005. Statistical and Computational Inverse Problems. Applied Mathematical Sciences. New York: Springer-Verlag.
Kennedy, Marc C., and Anthony OβHagan. 2001. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 63 (3): 425β64.
Lasanen, Sari. 2012. Inverse Problems & Imaging 6 (2): 267.
Lassas, Matti, Eero Saksman, and Samuli Siltanen. 2009. Inverse Problems and Imaging 3 (1): 87β122.
Liu, Xiao, Kyongmin Yeo, and Siyuan Lu. 2020. Journal of the American Statistical Association 0 (0): 1β18.
Lu, Lu, Pengzhan Jin, and George Em Karniadakis. 2020. arXiv:1910.03193 [Cs, Stat], April.
Lu, Lu, Zhiping Mao, and Xuhui Meng. 2019. In, 6.
Lu, Lu, Xuhui Meng, Zhiping Mao, and George Em Karniadakis. 2021. SIAM Review 63 (1): 208β28.
Mosegaard, Klaus, and Albert Tarantola. 2002. In International Geophysics, 81:237β65. Elsevier.
OβHagan, A. 2006. Reliability Engineering & System Safety, The Fourth International Conference on Sensitivity Analysis of Model Output (SAMO 2004), 91 (10): 1290β300.
Perdikaris, Paris, and George Em Karniadakis. 2016. Journal of the Royal Society, Interface 13 (118): 20151107.
Plumlee, Matthew. 2017. Journal of the American Statistical Association 112 (519): 1274β85.
Raissi, Maziar, Paris Perdikaris, and George Em Karniadakis. 2017a. November.
βββ. 2017b. November.
Raissi, Maziar, P. Perdikaris, and George Em Karniadakis. 2019. Journal of Computational Physics 378 (February): 686β707.
Roininen, Lassi, Janne M. J. Huttunen, and Sari Lasanen. 2014. Inverse Problems & Imaging 8 (2): 561.
SΓ€rkkΓ€, Simo, A. Solin, and J. Hartikainen. 2013. IEEE Signal Processing Magazine 30 (4): 51β61.
Sigrist, Fabio Roman Albert. 2013. Application/pdf. ETH Zurich.
Sigrist, Fabio, Hans R. KΓΌnsch, and Werner A. Stahel. 2015a. Application/pdf. Journal of Statistical Software 63 (14).
βββ. 2015b. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 77 (1): 3β33.
Stuart, A. M. 2010. Acta Numerica 19: 451β559.
Stuart, Andrew M., and Aretha L. Teckentrup. 2016. arXiv:1603.02004 [Math], December.
Tait, Daniel J., and Theodoros Damoulas. 2020. arXiv:2006.15641 [Cs, Stat], June.
Tarantola, Albert. 2005. Inverse Problem Theory and Methods for Model Parameter Estimation. SIAM.
βββ. n.d. Mapping Of Probabilities.
Teckentrup, Aretha L. 2020. arXiv:1909.00232 [Cs, Math, Stat], July.
Welter, David E., Jeremy T. White, Randall J. Hunt, and John E. Doherty. 2015. USGS Numbered Series 7-C12. Techniques and Methods. Reston, VA: U.S. Geological Survey.
White, Jeremy T., Michael N. Fienen, and John E. Doherty. 2016a. pyEMU: A Python Framework for Environmental Model Uncertainty Analysis Version .01. U.S. Geological Survey.
βββ. 2016b. Environmental Modelling & Software 85 (November): 217β28.
Xu, Kailai, and Eric Darve. 2020. In arXiv:2011.11955 [Cs, Math].
Yang, Liu, Dongkun Zhang, and George Em Karniadakis. 2020. SIAM Journal on Scientific Computing 42 (1): A292β317.
Zammit-Mangion, Andrew, Michael Bertolacci, Jenny Fisher, Ann Stavert, Matthew L. Rigby, Yi Cao, and Noel Cressie. 2021. Geoscientific Model Development Discussions, July, 1β51.
Zhang, Dongkun, Ling Guo, and George Em Karniadakis. 2020. SIAM Journal on Scientific Computing 42 (2): A639β65.
Zhang, Dongkun, Lu Lu, Ling Guo, and George Em Karniadakis. 2019. Journal of Computational Physics 397 (November): 108850.

### No comments yet. Why not leave one?

GitHub-flavored Markdown & a sane subset of HTML is supported.