pith. sign in

arxiv: 2409.09137 · v2 · submitted 2024-09-13 · 🧮 math.NA · cs.NA

Robust optimal design of large-scale Bayesian nonlinear inverse problems

Abhijit Chowdhary , Ahmed Attia , Alen Alexanderian This is my paper

Pith reviewed 2026-05-23 20:30 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords robust optimal experimental designBayesian inverse problemsPDE-constrained optimizationexpected information gainsensor placementeigenvalue sensitivityworst-case optimizationinfinite-dimensional problems
0
0 comments X

The pith

A framework finds designs for nonlinear Bayesian PDE inverse problems that stay optimal despite uncertainties in the model or prior.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out a worst-case approach to robust optimal experimental design for infinite-dimensional nonlinear Bayesian inverse problems constrained by PDEs. It approximates the expected information gain utility, derives its gradients analytically using eigenvalue sensitivity, and solves the resulting combinatorial max-min problem with a probabilistic optimization method. These steps produce sensor placements or similar designs that remain effective when elements of the inverse problem vary. A reader would care because practical inverse problems always carry uncertainty in the forward model, prior, or noise model, so non-robust designs can degrade sharply under small changes.

Core claim

The central claim is that a worst-case scenario formulation, paired with efficient expected-information-gain approximations and eigenvalue-sensitivity gradient evaluations, yields a scalable method that solves the robust optimal design problem for large-scale PDE-constrained Bayesian inversions and returns designs stable to variations in the inverse-problem elements, as shown for elliptic-PDE sensor placement.

What carries the argument

Eigenvalue sensitivity techniques that supply analytical forms and fast evaluation of the gradient of the expected information gain utility with respect to the uncertain inverse-problem elements inside a worst-case max-min formulation.

If this is right

  • The method scales to infinite-dimensional problems governed by PDEs without requiring full discretization of the design space at every step.
  • Analytical gradients obtained from eigenvalue sensitivities allow gradient-based optimization of the otherwise combinatorial max-min problem.
  • The probabilistic optimization paradigm converts the robust design task into a tractable stochastic program whose solution yields a design stable to the considered uncertainties.
  • The framework is demonstrated to produce effective robust sensor placements for an elliptic PDE inverse problem.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same sensitivity machinery could be reused to quantify how much each uncertain element contributes to design degradation, guiding which uncertainties deserve the most modeling effort.
  • If the EIG approximations extend without retuning, the framework could be applied directly to time-dependent or highly nonlinear PDEs common in fluid or structural inverse problems.
  • Embedding the robust optimizer inside existing Bayesian inversion libraries would let practitioners request a design that is already protected against their own prior and model uncertainties.

Load-bearing premise

The EIG approximations and eigenvalue sensitivity calculations remain accurate enough that the worst-case designs stay near-optimal when the actual nonlinear PDE model, prior, or noise model differs from the ones used in the design computation.

What would settle it

A numerical test that perturbs the forward model, prior covariance, or noise level within realistic ranges, recomputes the true expected information gain for both the robust design and a non-robust design, and checks whether the robust design retains higher utility in the worst-case perturbations.

Figures

Figures reproduced from arXiv: 2409.09137 by Abhijit Chowdhary, Ahmed Attia, Alen Alexanderian.

Figure 1
Figure 1. Figure 1: Results of the two sensor experiment. Left: Scatter plot of the utility [PITH_FULL_IMAGE:figures/full_fig_p020_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparison of U and a double-loop Monte Carlo estimator of the EIG. This is done across different designs and realizations of θ. Each marker represents a different realization of the noise parameter, while the color/shape indicates a different design. additional insight into the correlation between U and DKL for inverse problems of this type. 4.2. 64 Sensor, Budget 8, Experiment. Here, we consider an ROED … view at source ↗
Figure 3
Figure 3. Figure 3: Optimization trajectory of the 64 sensor, budget 8, experiment. The iterations [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Results of the 64 sensor, budget 8, experiment. Left: Optimal design discov [PITH_FULL_IMAGE:figures/full_fig_p022_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: A visualization of the quality of (ξ opt , θ opt) for the 64 sensor, budget 8, ROED experiment. Here we compare the utility of (ξ opt , θ opt) against (ξ opt , θ) for random θ and (ξ, θ opt) for random ξ. These are evaluated using the utility U (left) and the low-rank EIG D (r) KL (right). is significantly higher than that of U(ξ, θ opt) for random designs. This indicates that the discovered design is near… view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of objective values across different designs and realizations of [PITH_FULL_IMAGE:figures/full_fig_p024_6.png] view at source ↗
read the original abstract

We consider robust optimal experimental design (ROED) for nonlinear Bayesian inverse problems governed by partial differential equations (PDEs). An optimal design is one that maximizes some utility quantifying the quality of the solution of an inverse problem. However, the optimal design is dependent on elements of the inverse problem such as the simulation model, the prior, or the measurement error model. ROED aims to produce an optimal design that is aware of the additional uncertainties encoded in the inverse problem and remains optimal even after variations in them. We follow a worst-case scenario approach to develop a new framework for robust optimal design of nonlinear Bayesian inverse problems. The proposed framework a) is scalable and designed for infinite-dimensional Bayesian nonlinear inverse problems constrained by PDEs; b) develops efficient approximations of the utility, namely, the expected information gain; c) employs eigenvalue sensitivity techniques to develop analytical forms and efficient evaluation methods of the gradient of the utility with respect to the uncertainties we wish to be robust against; and d) employs a probabilistic optimization paradigm that properly defines and efficiently solves the resulting combinatorial max-min optimization problem. The effectiveness of the proposed approach is illustrated for optimal sensor placement problem in an inverse problem governed by an elliptic PDE.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a framework for robust optimal experimental design (ROED) of nonlinear Bayesian inverse problems governed by PDEs. It adopts a worst-case scenario formulation to produce designs robust to uncertainties in the model, prior, or noise model. The framework develops efficient approximations to the expected information gain (EIG) utility, employs eigenvalue sensitivity analysis to obtain analytical gradients of the utility with respect to the uncertain elements, and uses a probabilistic optimization approach to solve the resulting combinatorial max-min problem. Scalability to infinite-dimensional settings is claimed, with effectiveness illustrated via an optimal sensor placement example governed by an elliptic PDE.

Significance. If the EIG approximations and eigenvalue sensitivity techniques remain accurate under worst-case perturbations, the work would provide a computationally tractable route to robust designs in large-scale PDE-constrained inverse problems. The probabilistic treatment of the max-min problem and the use of sensitivity methods for gradient evaluation are strengths that could improve efficiency over naive sampling approaches. The elliptic PDE demonstration shows practical applicability in a common setting, though broader impact depends on validation for strongly nonlinear forward maps.

major comments (2)
  1. [Abstract and illustration] Abstract and elliptic PDE illustration: the central claim of effectiveness for nonlinear Bayesian inverse problems rests on the EIG approximations and eigenvalue sensitivity producing gradients accurate enough for the max-min optimizer. However, the provided illustration uses an elliptic PDE, for which the parameter-to-observation map is often effectively linear; this does not test whether first-order eigenvalue perturbation analysis misses higher-order effects when worst-case uncertainties push the problem far from the nominal point.
  2. [Method and numerical results] The worst-case formulation combined with the proposed approximations is load-bearing for the robustness claim, yet no quantitative assessment (e.g., comparison of approximate vs. exact EIG gradients or bias in the resulting designs) is supplied for cases where the forward map exhibits clear nonlinearity.
minor comments (2)
  1. Notation for the uncertain elements (model, prior, noise) and their parameterization in the max-min problem could be introduced earlier and used consistently to improve readability.
  2. [Abstract] The abstract states the framework is 'scalable' for infinite-dimensional problems; a brief statement on how the low-rank or Laplace-type EIG approximations scale with mesh size or parameter dimension would strengthen the claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and the opportunity to clarify aspects of our work. Below we respond point-by-point to the major comments.

read point-by-point responses

  1. Referee: [Abstract and illustration] Abstract and elliptic PDE illustration: the central claim of effectiveness for nonlinear Bayesian inverse problems rests on the EIG approximations and eigenvalue sensitivity producing gradients accurate enough for the max-min optimizer. However, the provided illustration uses an elliptic PDE, for which the parameter-to-observation map is often effectively linear; this does not test whether first-order eigenvalue perturbation analysis misses higher-order effects when worst-case uncertainties push the problem far from the nominal point.

    Authors: The elliptic PDE example employs an uncertain conductivity field as the inversion parameter, rendering the parameter-to-observation map nonlinear. Nevertheless, we accept that the chosen regime may not exhibit strong nonlinearity and that first-order eigenvalue perturbation may miss higher-order effects under large worst-case shifts. We will revise the numerical results section to include an explicit discussion of the degree of nonlinearity present in the example, the range of perturbations considered, and the conditions under which the first-order sensitivity remains reliable. revision: partial

  2. Referee: [Method and numerical results] The worst-case formulation combined with the proposed approximations is load-bearing for the robustness claim, yet no quantitative assessment (e.g., comparison of approximate vs. exact EIG gradients or bias in the resulting designs) is supplied for cases where the forward map exhibits clear nonlinearity.

    Authors: We agree that the manuscript does not supply a direct quantitative comparison of approximate versus exact EIG gradients or resulting design bias for forward maps with clear nonlinearity. The current numerical study is limited to the elliptic PDE setting as a demonstration of scalability. We will add a statement in the conclusions section acknowledging this limitation and identifying the provision of such assessments for strongly nonlinear maps as an important direction for future work. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation chain self-contained against external benchmarks

full rationale

The abstract and description outline a framework using EIG approximations, eigenvalue sensitivity, and probabilistic optimization for ROED, but contain no equations, fitted parameters renamed as predictions, or self-citations that bear the central load. No self-definitional steps, uniqueness theorems imported from authors, or ansatzes smuggled via citation are present. The claims rest on proposed methods whose validity is to be assessed externally (e.g., via the elliptic PDE illustration), not by reduction to the inputs themselves. This is the normal case of an independent derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; ledger left empty pending full text.

pith-pipeline@v0.9.0 · 5743 in / 1142 out tokens · 35769 ms · 2026-05-23T20:30:45.613566+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages

  1. [1]

    Alexanderian , Optimal experimental design for infinite-dimensional Bayesian inverse problems governed by PDEs: a review , Inverse Problems, 37 (2021), p

    A. Alexanderian , Optimal experimental design for infinite-dimensional Bayesian inverse problems governed by PDEs: a review , Inverse Problems, 37 (2021), p. 043001

    work page 2021

  2. [2]

    Alexanderian, P

    A. Alexanderian, P. J. Gloor, and O. Ghattas , On Bayesian A- and D-optimal experi- ROED OF INFINITE-DIMENSIONAL BAYESIAN NONLINEAR INVERSE PROBLEMS 25 mental designs in infinite dimensions , Bayesian Analysis, 11 (2016), pp. 671 – 695

    work page 2016

  3. [3]

    Alexanderian, R

    A. Alexanderian, R. Nicholson, and N. Petra , Optimal design of large-scale nonlinear Bayesian inverse problems under model uncertainty , Inverse Problems, (2024)

    work page 2024

  4. [4]

    Alexanderian, N

    A. Alexanderian, N. Petra, G. Stadler, and I. Sunseri , Optimal design of large-scale Bayesian linear inverse problems under reducible model uncertainty: Good to know what you don’t know, SIAM/ASA Journal on Uncertainty Quantification, 9 (2021), pp. 163–184

    work page 2021

  5. [5]

    Alexanderian and A

    A. Alexanderian and A. K. Saibaba , Efficient D-optimal design of experiments for infinite- dimensional Bayesian linear inverse problems , SIAM Journal on Scientific Computing, 40 (2018), pp. A2956–A2985

    work page 2018

  6. [6]

    A. C. Atkinson and A. N. Donev , Optimum Experimental Designs , Oxford, 1992

    work page 1992

  7. [7]

    Attia , Probabilistic approach to black-box binary optimization with budget constraints: Application to sensor placement , arXiv preprint arXiv:2406.05830, (2024)

    A. Attia , Probabilistic approach to black-box binary optimization with budget constraints: Application to sensor placement , arXiv preprint arXiv:2406.05830, (2024)

    work page arXiv 2024

  8. [8]

    Attia and E

    A. Attia and E. Constantinescu , Optimal experimental design for inverse problems in the presence of observation correlations , SIAM Journal on Scientific Computing, 44 (2022), pp. A2808–A2842

    work page 2022

  9. [9]

    Attia, S

    A. Attia, S. Leyffer, and T. Munson , Robust A-optimal experimental design for Bayesian inverse problems, (2023), https://arxiv.org/abs/arXiv:2305.03855

    work page arXiv 2023

  10. [10]

    Attia, S

    A. Attia, S. Leyffer, and T. S. Munson , Stochastic learning approach for binary optimiza- tion: Application to Bayesian optimal design of experiments , SIAM Journal on Scientific Computing, 44 (2022), pp. B395–B427

    work page 2022

  11. [11]

    Bangerth, A framework for the adaptive finite element solution of large-scale inverse problems, SIAM Journal on Scientific Computing, 30 (2008), pp

    W. Bangerth, A framework for the adaptive finite element solution of large-scale inverse problems, SIAM Journal on Scientific Computing, 30 (2008), pp. 2965–2989

    work page 2008

  12. [12]

    Bartuska, L

    A. Bartuska, L. Espath, and R. Tempone, Small-noise approximation for Bayesian optimal experimental design with nuisance uncertainty, Comput. Methods Appl. Mech. Engrg., 399 (2022), p. 115320

    work page 2022

  13. [13]

    Biedermann and H

    S. Biedermann and H. Dette , A note on maximin and Bayesian D-optimal designs in weighted polynomial regression, Mathematical Methods of Statistics, 12 (2003), p. 358–370

    work page 2003

  14. [14]

    Bui-Thanh, O

    T. Bui-Thanh, O. Ghattas, J. Martin, and G. Stadler , A computational framework for infinite-dimensional Bayesian inverse problems part i: The linearized case, with appli- cation to global seismic inversion , SIAM Journal on Scientific Computing, 35 (2013), pp. A2494–A2523

    work page 2013

  15. [15]

    Chaloner and I

    K. Chaloner and I. Verdinelli, Bayesian experimental design: A review , Statistical Science, 10 (1995)

    work page 1995

  16. [16]

    P. Chen, U. Villa, and O. Ghattas , Taylor approximation and variance reduction for pde- constrained optimal control under uncertainty , Journal of Computational Physics, 385 (2019), p. 163–186

    work page 2019

  17. [17]

    Chowdhary, Scalable Uncertainty Quantification for Infinite-Dimensional Bayesian In- verse Problems., PhD thesis, Feb

    A. Chowdhary, Scalable Uncertainty Quantification for Infinite-Dimensional Bayesian In- verse Problems., PhD thesis, Feb. 2025, https://www.lib.ncsu.edu/resolver/1840.20/45106

    work page 2025

  18. [18]

    Chowdhary, S

    A. Chowdhary, S. E. Ahmed, and A. Attia , PyOED: An extensible suite for data assimila- tion and model-constrained optimal design of experiments , ACM Trans. Math. Softw., 50 (2024)

    work page 2024

  19. [19]

    Chowdhary, S

    A. Chowdhary, S. Tong, G. Stadler, and A. Alexanderian , Sensitivity analysis of the information gain in infinite-dimensional bayesian linear inverse problems , International Journal for Uncertainty Quantification, 14 (2024)

    work page 2024

  20. [20]

    D. R. Cox , Planning of Experiments , Wiley Classics Library, John Wiley & Sons, Nashville, TN, Apr. 1992

    work page 1992

  21. [21]

    Daon and G

    Y. Daon and G. Stadler, Mitigating the influence of the boundary on PDE-based covariance operators, 2018

    work page 2018

  22. [22]

    Darges, A

    J. Darges, A. Alexanderian, and P. Gremaud , Variance-based sensitivity of bayesian in- verse problems to the prior distribution , International Journal for Uncertainty Quantifica- tion

    work page

  23. [23]

    Dette, V

    H. Dette, V. B. Melas, and A. Pepelyshev , Standardized maximin e-optimal designs for the michaelis-menten model , Statistica Sinica, 13 (2003), p. 1147–1163

    work page 2003

  24. [24]

    Fedorov , Optimal experimental design , WIREs Computational Statistics, 2 (2010), pp

    V. Fedorov , Optimal experimental design , WIREs Computational Statistics, 2 (2010), pp. 581–589

    work page 2010

  25. [25]

    Go and T

    J. Go and T. Isaac , Robust expected information gain for optimal Bayesian experimental design using ambiguity sets, in Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence, J. Cussens and K. Zhang, eds., vol. 180 of Proceedings of Machine Learning Research, PMLR, 01–05 Aug 2022, pp. 728–737

    work page 2022

  26. [26]

    Halko, P

    N. Halko, P. G. Martinsson, and J. A. Tropp , Finding structure with randomness: Proba- bilistic algorithms for constructing approximate matrix decompositions , SIAM Review, 53 (2011), pp. 217–288. 26 A. CHOWDHARY, A. ATTIA, AND A. ALEXANDERIAN

    work page 2011

  27. [27]

    C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gommers, P. Virtanen, D. Courna- peau, E. Wieser, J. Taylor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, M. H. van Kerkwijk, M. Brett, A. Haldane, J. F. del R ´ıo, M. Wiebe, P. Pe- terson, P. G ´erard-Marchant, K. Sheppard, T. Reddy, W. Weckesser, H. Ab- basi, C. Gohlke, and T. E. Oliphant , Array...

    work page doi:10.1038/s41586-020-2649-2 2020

  28. [28]

    Huan and Y

    X. Huan and Y. M. Marzouk , Simulation-based optimal bayesian experimental design for nonlinear systems, Journal of Computational Physics, 232 (2013), p. 288–317, https://doi. org/10.1016/j.jcp.2012.08.013

    work page doi:10.1016/j.jcp.2012.08.013 2013

  29. [29]

    Kaipio and V

    J. Kaipio and V. Kolehmainen , Approximate marginalization over modeling errors and un- certainties in inverse problems , Bayesian Theory and Applications, (2013), pp. 644–672

    work page 2013

  30. [30]

    Kolehmainen, T

    V. Kolehmainen, T. Tarvainen, S. R. Arridge, and J. P. Kaipio, Marginalization of unin- teresting distributed parameters in inverse problems-application to diffuse optical tomog- raphy, International Journal for Uncertainty Quantification, 1 (2011)

    work page 2011

  31. [31]

    Koval, A

    K. Koval, A. Alexanderian, and G. Stadler, Optimal experimental design under irreducible uncertainty for linear inverse problems governed by PDEs , Inverse Problems, 36 (2020)

    work page 2020

  32. [32]

    Kullback and R

    S. Kullback and R. A. Leibler, On information and sufficiency, The Annals of Mathematical Statistics, 22 (1951), pp. 79–86

    work page 1951

  33. [33]

    L. C. Lau and H. Zhou , A local search framework for experimental design , (2020)

    work page 2020

  34. [34]

    P. D. Lax, Linear algebra and its applications, Pure and Applied Mathematics: A Wiley Series of Texts, Monographs and Tracts, Wiley-Blackwell, Chichester, England, 2 ed., Aug. 2007

    work page 2007

  35. [35]

    Levitin and B

    E. Levitin and B. Polyak, Constrained minimization methods, USSR Computational Math- ematics and Mathematical Physics, 6 (1966), pp. 1–50

    work page 1966

  36. [36]

    D. C. Liu and J. Nocedal, On the limited memory BFGS method for large scale optimization , Mathematical Programming, 45 (1989), p. 503–528

    work page 1989

  37. [37]

    Logg, K.-A

    A. Logg, K.-A. Mardal, and G. Wells , Automated Solution of Differential Equations by the Finite Element Method: The FEniCS Book , vol. 84 of Lecture Notes in Computational Science and Engineering, Springer Berlin Heidelberg, Berlin, Heidelberg, 2012

    work page 2012

  38. [38]

    Mozumder, T

    M. Mozumder, T. Tarvainen, S. Arridge, J. P. Kaipio, C. D’Andrea, and V. Kolehmainen, Approximate marginalization of absorption and scattering in fluores- cence diffuse optical tomography, Inverse Problems & Imaging, 10 (2016), p. 227

    work page 2016

  39. [39]

    O’Leary-Roseberry, U

    T. O’Leary-Roseberry, U. Villa, P. Chen, and O. Ghattas , Derivative-informed pro- jected neural networks for high-dimensional parametric maps governed by PDEs , Com- puter Methods in Applied Mechanics and Engineering, 388 (2022), p. 114199

    work page 2022

  40. [40]

    Plessix, A review of the adjoint-state method for computing the gradient of a functional with geophysical applications, Geophysical Journal International, 167 (2006), pp

    R.-E. Plessix, A review of the adjoint-state method for computing the gradient of a functional with geophysical applications, Geophysical Journal International, 167 (2006), pp. 495–503

    work page 2006

  41. [41]

    Pronzato and E

    L. Pronzato and E. Walter , Robust experiment design via maximin optimization , Mathe- matical Biosciences, 89 (1988), pp. 161–176

    work page 1988

  42. [42]

    Rainforth, A

    T. Rainforth, A. Foster, D. R. Ivanova, and F. B. Smith, Modern Bayesian Experimental Design, Statistical Science, 39 (2024), pp. 100 – 114

    work page 2024

  43. [43]

    C. R. Rojas, J. S. Welsh, G. C. Goodwin, and A. Feuer, Robust optimal experiment design for system identification, Automatica, 43 (2007), pp. 993–1008

    work page 2007

  44. [44]

    A. M. Stuart , Inverse problems: A Bayesian perspective , Acta Numerica, 19 (2010), p. 451–559

    work page 2010

  45. [45]

    Sunseri, A

    I. Sunseri, A. Alexanderian, J. Hart, and B. v. B. Waanders, Hyper-differential sensitivity analysis for nonlinear Bayesian inverse problems , International Journal for Uncertainty Quantification, 14 (2024), p. 1–20

    work page 2024

  46. [46]

    Telen, F

    D. Telen, F. Logist, E. Van Derlinden, and J. F. Van Impe , Robust optimal experiment design: A multi-objective approach , IFAC Proceedings Volumes, 45 (2012), pp. 689–694. 7th Vienna International Conference on Mathematical Modelling

    work page 2012

  47. [47]

    Telen, D

    D. Telen, D. Vercammen, F. Logist, and J. Van Impe , Robustifying optimal experiment design for nonlinear, dynamic (bio)chemical systems, Computers & Chemical Engineering, 71 (2014), pp. 415–425

    work page 2014

  48. [48]

    Uci ´nski, Optimal measurement methods for distributed parameter system identification , Systems and Control Series, CRC Press, Boca Raton, FL, 2005

    D. Uci ´nski, Optimal measurement methods for distributed parameter system identification , Systems and Control Series, CRC Press, Boca Raton, FL, 2005

    work page 2005

  49. [49]

    Villa, N

    U. Villa, N. Petra, and O. Ghattas , HIPPYlib: An Extensible Software Framework for Large-Scale Inverse Problems Governed by PDEs: Part I: Deterministic Inversion and Linearized Bayesian Inference, ACM Trans. Math. Softw., 47 (2021)

    work page 2021

  50. [50]

    Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, St´ efan J

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Lar- ROED OF INFINITE-DIMENSIONAL BAYESIAN NONLINEAR INVERSE PROBLEMS 27 son, C. J. Carey, ˙I. Polat, Y. Feng, E. W. Mo...

    work page doi:10.1038/s41592-019-0686-2 2020

  51. [51]

    Wald, Statistical decision functions which minimize the maximum risk , Annals of Mathe- matics, 46 (1945), p

    A. Wald, Statistical decision functions which minimize the maximum risk , Annals of Mathe- matics, 46 (1945), p. 265–280

    work page 1945

  52. [52]

    K. Wu, P. Chen, and O. Ghattas , A fast and scalable computational framework for large- scale high-dimensional bayesian optimal experimental design , SIAM/ASA Journal on Un- certainty Quantification, 11 (2023), pp. 235–261. Appendix A. Variational Tools. The gradient and the Hessian of the cost functional Φ (3.10a) are essential components of our proposed...

    work page 2023