pith. sign in

arxiv: 2603.00033 · v2 · submitted 2026-02-05 · 🧮 math.OC · math.PR

Iterative Data-Consistent Inversion with Multiple Push-forward Constraints

Tianyi Jiang , Troy Butler , Timothy Wildey , Tim Kutta , Haonan Wang This is my paper

Pith reviewed 2026-05-16 06:33 UTC · model grok-4.3

classification 🧮 math.OC math.PR
keywords data-consistent inversionpush-forward constraintsmultiple constraintsf-divergenceuncertainty quantificationiterative methodsinverse problemsprobability measures
0
0 comments X

The pith

Iterative data-consistent inversion converges to the parameter measure that satisfies all multiple push-forward constraints while minimizing cumulative f-divergence.

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

The paper addresses the challenge of finding a probability measure on uncertain parameters such that pushing it forward through a model matches several distinct observed probability measures on output quantities of interest. It introduces an iterative scheme that applies the single-constraint data-consistent inversion repeatedly, updating the measure each time to incorporate one new constraint while preserving proximity to prior iterates. The authors establish that this procedure converges to a measure satisfying every constraint at once. The converged measure minimizes the sum of f-divergences from the initial measure across all constraints. When the process begins from a uniform distribution, the result is the maximum-entropy distribution consistent with the full set of constraints.

Core claim

The iterative DCI scheme is shown to converge to a solution of the multiple push-forward constraint problem. This iterative solution minimizes the cumulative f-divergence across all constraints and, under uniform initializations, represents the maximal entropy solution (the I-projection) onto the intersection of the solution sets.

What carries the argument

The iterative Data-Consistent Inversion scheme, which applies the single-constraint optimal update repeatedly to accumulate satisfaction of multiple push-forward constraints.

If this is right

  • The final measure satisfies every individual push-forward constraint simultaneously.
  • The method handles high-dimensional parameter spaces without constructing or approximating a joint observed measure.
  • Convergence holds for the stated class of problems under the given initialization.
  • The solution coincides with the I-projection onto the intersection when initialized uniformly.

Where Pith is reading between the lines

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

  • The scheme could be adapted to streaming data by treating each new observation batch as an additional constraint in sequence.
  • Extensions to other divergence families might preserve the same convergence structure if the single-step optimality carries over.
  • In coupled multi-physics models the iteration could reduce the need for expensive joint sampling of all outputs together.

Load-bearing premise

The single-constraint DCI solution is optimal in the f-divergence sense and the iterative updates converge to a non-empty intersection of the individual solution sets.

What would settle it

A numerical test on a low-dimensional example in which the final iterate's push-forward deviates from one of the observed measures by more than the discretization error after the claimed convergence.

Figures

Figures reproduced from arXiv: 2603.00033 by Haonan Wang, Tianyi Jiang, Tim Kutta, Timothy Wildey, Troy Butler.

Figure 1
Figure 1. Figure 1: Schematic that conceptualizes the iterative DCI process when [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (Left) Uniform initial samples (blue circles) and data-generating samples (orange pluses) on [PITH_FULL_IMAGE:figures/full_fig_p021_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The KDE estimates of densities associated with Figure 2 after the first epoch of iterations. (Top [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The KDE estimates of densities associated with Figure 2 after the fifth epoch of iterations. (Top [PITH_FULL_IMAGE:figures/full_fig_p023_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The KDE estimates of densities associated with Figure 2 after the final epoch (46th) of iterations. [PITH_FULL_IMAGE:figures/full_fig_p024_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The KL divergences between the QoI marginals associated with the push-forward of the updated [PITH_FULL_IMAGE:figures/full_fig_p024_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: (Top Row:) The updated density obtained by using both the joint QoI data shown in Figure 2 [PITH_FULL_IMAGE:figures/full_fig_p025_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: (Top Row:) The updated density obtained by incorrectly defining the observed QoI distribution [PITH_FULL_IMAGE:figures/full_fig_p026_8.png] view at source ↗
Figure 8
Figure 8. Figure 8: This “wrong” example illustrates an important, but perhaps subtle, point about Step 2 in the [PITH_FULL_IMAGE:figures/full_fig_p027_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: The joint data space for three linear QoI maps (eqs. (18)-(20)) on a 2-dimensional parameter space [PITH_FULL_IMAGE:figures/full_fig_p027_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: The KDE estimates of densities associated with parameter samples from Figure 2 and QoI samples [PITH_FULL_IMAGE:figures/full_fig_p028_10.png] view at source ↗
Figure 10
Figure 10. Figure 10: Since each epoch now includes three iterations, each iteration of the last epoch is shown in the [PITH_FULL_IMAGE:figures/full_fig_p029_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Two random realizations of the permeability field generated using the truncated KL expansion. [PITH_FULL_IMAGE:figures/full_fig_p030_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: The approximated pressure fields corresponding to the two random realizations of the permeability [PITH_FULL_IMAGE:figures/full_fig_p030_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: The magnitude of the approximated velocity fields corresponding to the two random realizations [PITH_FULL_IMAGE:figures/full_fig_p031_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: The first eight randomly generated locations for sensors are used to generate a 12-dimensional [PITH_FULL_IMAGE:figures/full_fig_p032_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: KDE estimates of the marginal QoI densities for [PITH_FULL_IMAGE:figures/full_fig_p033_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: KDE estimates of the marginal QoI densities for [PITH_FULL_IMAGE:figures/full_fig_p034_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Sample means of the first 16 KL modes for the data-generating distribution (blue bars) and [PITH_FULL_IMAGE:figures/full_fig_p035_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: KDE estimates of the marginal QoI densities for [PITH_FULL_IMAGE:figures/full_fig_p036_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Sample means of the first 16 KL modes for the data-generating distribution (blue bars) and [PITH_FULL_IMAGE:figures/full_fig_p036_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: The trends of the updated means of the first five KL modes as a function of iteration when [PITH_FULL_IMAGE:figures/full_fig_p037_20.png] view at source ↗
read the original abstract

A foundational challenge in uncertainty quantification involves estimating a probability measure on the space of uncertain parameters such that its push-forward through a computational model matches an observed probability measure on the output data associated with quantities of interest (QoI). When multiple, distinct sets of observational data are available, the desired parameter measure should simultaneously satisfy multiple push-forward constraints associated with various subsets of the QoI. In this work, we present a convergent measure-theoretic framework for solving this problem based on an iterative application of Data-Consistent Inversion (DCI). We first rigorously establish the theoretical optimality of the DCI solution to the standard problem, proving that it minimizes the $f$-divergence over the space of all possible pullback measures that satisfy the push-forward constraint. This optimality property provides the foundation for our iterative DCI scheme, which is shown to converge to a solution of the multiple push-forward constraint problem. This iterative solution minimizes the cumulative $f$-divergence across all constraints and, under uniform initializations, represents the maximal entropy solution (the I-projection) onto the intersection of the solution sets. We provide a rigorous convergence analysis for the proposed method and demonstrate its practical utility through numerical examples, including a high-dimensional parameter space governed by partial differential equations, where the iterative approach robustly avoids the complexities associated with approximating high-dimensional joint observed measures.

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 develops a convergent iterative Data-Consistent Inversion (DCI) framework for recovering a parameter probability measure whose push-forwards simultaneously match multiple observed output measures associated with distinct quantities of interest. It first proves that the single-constraint DCI solution is the unique minimizer of an f-divergence over all pullback measures satisfying the push-forward constraint, then shows that the iteration converges to a measure minimizing the cumulative f-divergence across all constraints; under uniform initialization this limit coincides with the I-projection (maximum-entropy solution) onto the intersection of the individual solution sets. Convergence is established in the weak topology on probability measures, relying on continuity of the forward maps and absolute continuity of the observed measures. The approach is illustrated on numerical examples, including a high-dimensional PDE-governed system.

Significance. If the optimality and convergence claims hold, the work supplies a rigorous, measure-theoretic alternative to joint-distribution approaches for multi-data inverse problems in uncertainty quantification. The avoidance of high-dimensional joint observed measures, combined with explicit f-divergence minimization and I-projection properties, offers both theoretical grounding and practical scalability for complex forward models. The provision of convergence guarantees under stated topological hypotheses strengthens the foundation for iterative DCI methods.

major comments (2)
  1. [Section 4] Convergence Analysis (Section 4): the claim that the iteration converges to a non-empty intersection of solution sets rests on hypotheses of forward-map continuity and absolute continuity of observed measures; these must be stated explicitly as numbered assumptions (e.g., A1–A3) with a precise reference to the weak-topology compactness argument that guarantees existence of the limit.
  2. [Section 3] Optimality of single-constraint DCI (Section 3): the variational argument establishing f-divergence minimality over pullback measures should include an explicit verification that the minimizer is attained within the admissible set, particularly when the observed measure is only absolutely continuous rather than equivalent.
minor comments (2)
  1. [Introduction] Notation: the definition of the cumulative f-divergence functional should appear in the introduction or preliminaries rather than only in the convergence theorem statement.
  2. [Numerical Examples] Numerical examples: the caption of the high-dimensional PDE figure should specify the exact dimension of the parameter space and the number of push-forward constraints used.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading, positive assessment of the work, and constructive comments on the presentation of the theoretical results. We address each major comment below and will incorporate the suggested clarifications into the revised manuscript.

read point-by-point responses

  1. Referee: [Section 4] Convergence Analysis (Section 4): the claim that the iteration converges to a non-empty intersection of solution sets rests on hypotheses of forward-map continuity and absolute continuity of observed measures; these must be stated explicitly as numbered assumptions (e.g., A1–A3) with a precise reference to the weak-topology compactness argument that guarantees existence of the limit.

    Authors: We agree that explicitly numbering the assumptions will improve clarity and rigor. In the revised manuscript, we will introduce three numbered assumptions at the start of Section 4: Assumption A1 (continuity of each forward map with respect to the weak topology on probability measures), Assumption A2 (absolute continuity of each observed measure with respect to Lebesgue measure on the output space), and Assumption A3 (tightness of the sequence of iterates, which follows from the Prokhorov theorem in the Polish space setting). We will also add an explicit reference to the compactness argument (invoking Prokhorov's theorem to guarantee relative compactness of the sequence in the weak topology, followed by identification of the limit via the continuity of the push-forward operators) to ensure the existence of the limit point in the intersection of the solution sets. revision: yes

  2. Referee: [Section 3] Optimality of single-constraint DCI (Section 3): the variational argument establishing f-divergence minimality over pullback measures should include an explicit verification that the minimizer is attained within the admissible set, particularly when the observed measure is only absolutely continuous rather than equivalent.

    Authors: We appreciate this observation. The current proof establishes lower semi-continuity of the f-divergence and closedness of the constraint set, but we will strengthen it by adding an explicit verification step that the minimizer is attained. Specifically, we will invoke the direct method: show that any minimizing sequence is tight (using the absolute continuity of the observed measure to control the tails via the Radon-Nikodym derivative), extract a weakly convergent subsequence by Prokhorov's theorem, and pass to the limit using the continuity of the push-forward map to confirm the limit lies in the admissible set. This addition will be placed immediately after the variational inequality in the revised Section 3. revision: yes

Circularity Check

0 steps flagged

Derivation is self-contained; no circular reductions identified

full rationale

The paper derives optimality of the single-constraint DCI solution through a direct variational argument over pullback measures that minimize f-divergence, then extends this to an iterative scheme whose fixed point satisfies all constraints simultaneously. Convergence is analyzed in the space of probability measures under explicit hypotheses on continuity of forward maps and absolute continuity of observed measures. No quoted step reduces a claimed prediction or optimality result to a fitted input by construction, nor relies on a load-bearing self-citation chain; the I-projection property under uniform initialization follows from standard information-projection facts applied to the intersection of solution sets. The derivation chain is therefore independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The framework rests on standard existence and absolute-continuity assumptions from measure theory together with the optimality property of the single-step DCI solution; no new free parameters or invented entities are introduced in the abstract.

axioms (2)
  • domain assumption Existence of probability measures whose push-forwards match each observed output measure
    Required for the individual solution sets and their intersection to be non-empty.
  • standard math The single-constraint DCI solution minimizes the chosen f-divergence
    This optimality property is invoked to justify the iterative construction.

pith-pipeline@v0.9.0 · 5542 in / 1375 out tokens · 34026 ms · 2026-05-16T06:33:41.364388+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel echoes
    ?
    echoes

    ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

    We prove that the DCI solution ... minimizes the f-divergence over the space of all possible pullback measures ... This iterative solution minimizes the cumulative f-divergence ... represents the maximal entropy solution (the I-projection) onto the intersection of the solution sets.

  • IndisputableMonolith/Foundation/BranchSelection.lean branch_selection unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    the sequence of Radon-Nikodym derivatives ... satisfies the conditions necessary to apply a generalization of the k=2 convergence results found in [37]

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages

  1. [1]

    Calvetti, D., Kaipio, J., and Somersalo, E., Inverse problems in the Bayesian framework,Inverse Prob- lems, 30(11):110301, 2014

    work page 2014

  2. [2]

    Gelman, A., Carlin, J.B., Stern, H.S., Dunson, D.B., Vehtari, A., and Rubin, D.B.,Bayesian Data Analysis, Third Edition, Chapman and Hall/CRC, 2013

    work page 2013

  3. [3]

    Cotter, S., Dashti, M., and Stuart, A., Approximation of Bayesian Inverse Problems,SIAM Journal of Numerical Analysis, 48:322–345, 2010

    work page 2010

  4. [4]

    and O’Hagan, A., Bayesian calibration of computer models,Journal of the Royal Sta- tistical Society: Series B (Statistical Methodology), 63(3):425–464, 2001

    Kennedy, M.C. and O’Hagan, A., Bayesian calibration of computer models,Journal of the Royal Sta- tistical Society: Series B (Statistical Methodology), 63(3):425–464, 2001. 39

    work page 2001

  5. [5]

    Berger, J., Moreno, E., Pericchi, L., Bayarri, M., Bernardo, J., Cano, J., De la Horra, J., Mart´ ın, J., R´ ıos-Ins´ ua, D., Betr` o, B., Dasgupta, A., Gustafson, P., Wasserman, L., Kadane, J., Srinivasan, C., Lavine, M., O’Hagan, A., Polasek, W., Robert, C., Goutis, C., Ruggeri, F., Salinetti, G., and Sivaganesan, S., An overview of robust Bayesian anal...

    work page 1994

  6. [6]

    Fitzpatrick, B., Bayesian analysis in inverse problems,Inverse Problems, 7(5):675, 1991

    work page 1991

  7. [7]

    and Lucka, F., Maximum a posteriori estimates in linear inverse problems with log-concave priors are proper Bayes estimators,Inverse Problems, 30(11):114004, Oct 2014

    Burger, M. and Lucka, F., Maximum a posteriori estimates in linear inverse problems with log-concave priors are proper Bayes estimators,Inverse Problems, 30(11):114004, Oct 2014

    work page 2014

  8. [8]

    Petra, N., Martin, J., Stadler, G., and Ghattas, O., A Computational Framework for Infinite- Dimensional Bayesian Inverse Problems, Part II: Stochastic Newton MCMC with Application to Ice Sheet Flow Inverse Problems,SIAM Journal on Scientific Computing, 36(4):A1525–A1555, 2014

    work page 2014

  9. [9]

    Alexandrian, A., Petra, N., Stadler, G., and Ghattas, O., A Fast and Scalable Method for A-Optimal Design of Experiments for Infinite-dimensional Bayesian Nonlinear Inverse Problems,SIAM Journal on Scientific Computing, 38:A243–A272, 2016

    work page 2016

  10. [10]

    Wikle, C., Berliner, L., and Cressie, N., Hierarchical Bayesian space-time models,Environmental and Ecological Statistics, 5(2):117–154, Jun 1998

    work page 1998

  11. [11]

    Salvatier, J., Wiecki, T., and Fonnesbeck, C., Probabilistic programming in Python using PyMC3, PeerJ Computer Science, 2:e55, 2016

    work page 2016

  12. [12]

    1–27, 2014

    Butler, T., Estep, D., Tavener, S., Dawson, C., and Westerink, J., A Measure-Theoretic Computa- tional Method For Inverse Sensitivity Problems III: Multiple Quantities of Interest,SIAM Journal on Uncertainty Quantification, pp. 1–27, 2014

    work page 2014

  13. [13]

    Butler, T., Jakeman, J., and Wildey, T., Combining push-forward measures and Bayes’ rule to construct consistent solutions to stochastic inverse problems,SIAM Journal on Scientific Computing, 40(2):A984– A1011, 2018

    work page 2018

  14. [14]

    Bingham, D., Butler, T., and Estep, D., Inverse problems for physics-based process models,Annual Review of Statistics and Its Application, 11:461–482, 2024

    work page 2024

  15. [15]

    Butler, T., Wildey, T., and Yen, T., Data-consistent inversion for stochastic input-to-output maps, Inverse Problems, 36(8):085015, 2020

    work page 2020

  16. [16]

    Butler, T., Spence, R., Wildey, T., and Yen, T., Stability and convergence of solutions to stochastic inverse problems using approximate probability densities,International Journal for Uncertainty Quan- tification, 15(4):21–51, 2025

    work page 2025

  17. [17]

    and Pollard, D., Conditioning as disintegration,Statistica Neerlandica, 51(3):287–317, 1997

    Chang, J. and Pollard, D., Conditioning as disintegration,Statistica Neerlandica, 51(3):287–317, 1997

    work page 1997

  18. [18]

    and Mikelsons, L., Solving Stochastic Inverse Problems with Stochastic BayesFlow, In2023 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM), pp

    Zhang, Y. and Mikelsons, L., Solving Stochastic Inverse Problems with Stochastic BayesFlow, In2023 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM), pp. 966–972, 2023. 40

    work page 2023

  19. [19]

    Rumbell, T., Parikh, J., Kozloski, J., and Gurev, V., Novel and flexible parameter estimation methods for data-consistent inversion in mechanistic modelling,Royal Society Open Science, 10(11):230668, 2023

    work page 2023

  20. [20]

    Mattis, S., Steffen, K., Butler, T., Dawson, C., and Estep, D., Learning quantities of interest from dynamical systems for observation-consistent inversion,Computer Methods in Applied Mechanics and Engineering, 388:114230, 2022

    work page 2022

  21. [21]

    and Wildey, T., Solving stochastic inverse problems for property–structure linkages using data-consistent inversion and machine learning,JOM, 73(1):72–89, 2021

    Tran, A. and Wildey, T., Solving stochastic inverse problems for property–structure linkages using data-consistent inversion and machine learning,JOM, 73(1):72–89, 2021

    work page 2021

  22. [22]

    Bruder, L., Gee, M., and Wildey, T., Data-Consistent Solutions to Stochastic Inverse Problems Us- ing a Probabilistic Multi-Fidelity Method Based on Conditional Densities,International Journal for Uncertainty Quantification, 10(5), 2020

    work page 2020

  23. [23]

    Fleury, A., Bouttier, F., and Bergot, T., Calibration of parameter perturbations for ensemble prediction using data-consistent inversion,Monthly Weather Review, 153(4):655–672, April 2025

    work page 2025

  24. [24]

    Butler, T. and Hakula, H., What do we hear from a drum? A data-consistent approach to quantifying irreducible uncertainty on model inputs by extracting information from correlated model output data, Computer Methods in Applied Mechanics and Engineering, 370:113228, October 2020

    work page 2020

  25. [25]

    and Raftery, A., Inference for Deterministic Simulation Models: The Bayesian Melding Ap- proach,Journal of the American Statistical Association, 95(452):1244–1255, 2000

    Poole, D. and Raftery, A., Inference for Deterministic Simulation Models: The Bayesian Melding Ap- proach,Journal of the American Statistical Association, 95(452):1244–1255, 2000

    work page 2000

  26. [26]

    Butler, T., Jakeman, J., and Wildey, T., Convergence of probability densities using approximate models for forward and inverse problems in uncertainty quantification,SIAM Journal on Scientific Computing, 40(5):A3523–A3548, 2018

    work page 2018

  27. [27]

    Pilosov, M., del Castillo-Negrete, C., Yen, T., Butler, T., and Dawson, C., Parameter estimation with maximal updated densities,Computer Methods in Applied Mechanics and Engineering, 407:115906, 2023

    work page 2023

  28. [28]

    Roper, T., Hakula, H., and Butler, T., A machine-learning enabled framework for quantifying uncertain- ties in parameters of computational models,Computers & Mathematics with Applications, 182:184–212, 2025

    work page 2025

  29. [29]

    and Wu, Y.,Information Theory: From Coding to Learning, Cambridge University Press,

    Polyanskiy, Y. and Wu, Y.,Information Theory: From Coding to Learning, Cambridge University Press,

    work page

  30. [30]

    Draft of October 20, 2022

    work page 2022

  31. [31]

    Guntuboyina, A., Lower bounds for the minimax risk usingf-divergences, and applications,IEEE Transactions on Information Theory, 57(4):2386–2399, 2011

    work page 2011

  32. [32]

    Blei, D.M., Kucukelbir, A., and McAuliffe, J., Variational inference: A review for statisticians,Journal of the American Statistical Association, 112(518):859–877, April 2017

    work page 2017

  33. [33]

    and Leibler, R., On information and sufficiency,The Annals of Mathematical Statistics, 22:79–86, 1951

    Kullback, S. and Leibler, R., On information and sufficiency,The Annals of Mathematical Statistics, 22:79–86, 1951. 41

    work page 1951

  34. [34]

    and Harremoes, P., Renyi Divergence and Kullback-Leibler Divergence,IEEE Transac- tions on Information Theory, 60:3797–3820, 2014

    van Erven, T. and Harremoes, P., Renyi Divergence and Kullback-Leibler Divergence,IEEE Transac- tions on Information Theory, 60:3797–3820, 2014

    work page 2014

  35. [35]

    and Horel, T., Optimal bounds betweenf-divergences and integral probability metrics,J

    Agrawal, R. and Horel, T., Optimal bounds betweenf-divergences and integral probability metrics,J. Mach. Learn. Res., 22(1):1–59, 2021

    work page 2021

  36. [36]

    and Tusn´ ady, G., Information geometry and alternating minimization procedures,Statistics and Decisions, Supplemental Issue Number 1:205–237, 1984

    Csisz´ ar, I. and Tusn´ ady, G., Information geometry and alternating minimization procedures,Statistics and Decisions, Supplemental Issue Number 1:205–237, 1984

    work page 1984

  37. [37]

    and Byrne, W., Convergence theorems for generalized alternating minimization pro- cedures,J

    Gunawardana, A. and Byrne, W., Convergence theorems for generalized alternating minimization pro- cedures,J. Mach. Learn. Res., 6:2049–2073, December 2005

    work page 2049

  38. [38]

    Ruschendorf, L., Convergence of the Iterative Proportional Fitting Procedure,The Annals of Statistics, 23(4):1160 – 1174, 1995

    work page 1995

  39. [39]

    and Gerolin, A., An Optimal Transport Approach for the Schr¨ odinger Bridge Problem and Convergence of Sinkhorn Algorithm,Journal of Scientific Computing, 85(2), October 2020

    Marino, S. and Gerolin, A., An Optimal Transport Approach for the Schr¨ odinger Bridge Problem and Convergence of Sinkhorn Algorithm,Journal of Scientific Computing, 85(2), October 2020

    work page 2020

  40. [40]

    Csiszar, I.,I-Divergence Geometry of Probability Distributions and Minimization Problems,The Annals of Probability, 3(1):146 – 158, 1975

    work page 1975

  41. [41]

    and Shenitzer, A., The principle of maximum entropy,The Mathematical Intelligencer, 7(1):42–48, 1985

    Guiasu, S. and Shenitzer, A., The principle of maximum entropy,The Mathematical Intelligencer, 7(1):42–48, 1985

    work page 1985

  42. [42]

    and Scott, D., Variable kernel density estimation,The Annals of Statistics, 20(3):1236–1265, 1992

    Terrell, G. and Scott, D., Variable kernel density estimation,The Annals of Statistics, 20(3):1236–1265, 1992

    work page 1992

  43. [43]

    Ganis, B., Klie, H., Wheeler, M., Wildey, T., Yotov, I., and Zhang, D., Stochastic collocation and mixed finite elements for flow in porous media,Computer methods in applied mechanics and engineering, 197(43):3547–3559, 2008

    work page 2008

  44. [44]

    Wheeler, M., Wildey, T., and Yotov, I., A multiscale preconditioner for stochastic mortar mixed finite elements,Computer Methods in Applied Mechanics and Engineering, 200(9):1251–1262, 2011

    work page 2011

  45. [45]

    Zhang, D. and Lu, Z., An efficient, high-order perturbation approach for flow in random porous media via Karhunen-Lo´ eve and polynomial expansions,Journal of Computational Physics, 194(2):773–794, 2004

    work page 2004

  46. [46]

    User/Reference Guide for MrHyDE - A framework for solving Multi-resolution Hybridized Differential Equations – Version 1.0, May 2024

    Wildey, T. User/Reference Guide for MrHyDE - A framework for solving Multi-resolution Hybridized Differential Equations – Version 1.0, May 2024. SAND2024-06292

    work page 2024

  47. [47]

    MrHyDE library.https://github.com/sandialabs/MrHyDE

    Wildey, T. MrHyDE library.https://github.com/sandialabs/MrHyDE. Accessed: 09/25/2024

    work page 2024

  48. [48]

    and Czado, C., Evading the curse of dimensionality in nonparametric density estimation with simplified vine copulas,Journal of Multivariate Analysis, 151:69–89, 2016

    Nagler, T. and Czado, C., Evading the curse of dimensionality in nonparametric density estimation with simplified vine copulas,Journal of Multivariate Analysis, 151:69–89, 2016. 42

    work page 2016