Pith. sign in

REVIEW 5 major objections 5 minor 54 references

An Ensemble Information Filter: Retrieving Markov-information from the SPDE discretisation

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The Ensemble Information Filter encodes Markov structure from SPDE dynamics into the precision matrix, making ad-hoc localisation unnecessary.

desk verdict A well-argued new combination of graph-sparse precision estimation and information-filter updates; the strong 'no localisation' claim needs tempering and reproducibility, but it deserves a proper referee. read the letter →

arxiv 2501.09016 v1 pith:5B45KSBT submitted 2025-01-15 stat.ME math.DS

Berent {AA}nund Str{o}mnes Lunde This is my paper · ORCID
classification stat.MEmath.DS MSC 62M2060H1565C30
keywords dataassimilationensembleKalmanfilterinformationMarkovrandomfieldprecisionmatrixSPDElocalisationGaussian
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that spurious correlations and ensemble collapse in ensemble data assimilation are finite-sample overfitting problems: estimating a dense Gaussian covariance from a small ensemble adds degrees of freedom that grow quadratically with resolution, so the fitted model Q diverges from the true P. The proposed Ensemble Information Filter (EnIF) instead estimates the precision matrix within the sparsity pattern of a graph G supplied by the SPDE discretisation, so conditional-independence structure is enforced before any statistical fitting. The central claim is that, if G encodes the local dependence of the dynamics, the posterior ensemble suffers far less from estimation error, spurious correlations, and collapse than standard EnKF or ES, and no ad-hoc distance-based localisation is needed. Numerical experiments on resolution convergence, AR-1 dependence strength, Lorenz-96 filtering, stochastic heat-equation smoothing, and a 2D anisotropic exponential GRF parameter problem support the claim.

What carries the argument

The central mechanism is the precision matrix $\Lambda = \Sigma^{-1}$ with the zero pattern of a conditional-independence graph $G=(V,E)$: $\Lambda_{ij} \neq 0$ if and only if $i=j$ or $(i,j)\in E$. The graph comes from the SPDE's local operators: for a linear local operator $L$, the precision operator is $Q = L^*L$, and a finite-element or finite-difference discretisation with piecewise-linear basis functions and a diagonal mass matrix gives banded sparse $\Lambda$. Estimation is performed through the affine Knothe-Rosenblatt factor $C$ of an inverse triangular map with $\Lambda = C^\top C$; the update uses the information-filter equations $\eta_{t|t} = \eta_{t|t-1} + H^\top \Lambda_{\epsilon r}\tilde d$ and $\Lambda_{t|t} = \Lambda_{t|t-1} + H^\top \Lambda_{\epsilon r} H$, avoiding dense matrix inversion. Fill-reducing permutations for the triangular factor, plus sparse regression for $H$ via a monotone-LASSO boosting algorithm, keep the procedure statistically and computationally scalable.

What would settle it

Take a linear-Gaussian test system with known precision matrix that has a dense off-diagonal block, such as two distant spatial regions coupled through an unobserved mediator, run EnIF with a graph G that omits that block, and increase the ensemble size: if the KLD between the EnIF posterior and the exact posterior does not converge to zero as n grows, the localisation-free and statistically consistent claims fail for misspecified G. A concrete version would use a non-Markovian 2D GRF with a deliberately over-sparsified graph and compare ensemble variance after conditioning against the exact conditional variance.

Watch

Extended reading notes

Core claim

The load-bearing statement is that the precision matrix, not the covariance, is the right object for ensemble updates: conditional independence appears as exact zeros in the precision, the information-filter update adds observation information to a sparse matrix without dense inversion, and the number of parameters to estimate equals vertices plus edges of G, growing linearly with resolution instead of quadratically. For smoothing under linear local SPDE dynamics, the Markov property is exact, via the Rozanov link between local linear operators and Gaussian Markov fields and the finite-element derivation with a diagonal lumped mass matrix. For filtering, the marginal distribution may lose strict spatial Markov properties, and the paper explicitly treats G there as a parsimonious approximation justified by a locality-preference argument and by the bias-reduction logic of the Kullback-Leibler divergence. The experimentally supported conclusion is that the posterior ensemble then tracks the exact Gaussian update far better than EnKF or ES, including at high resolution, without tuning any localisation radius.

Load-bearing premise

The load-bearing premise is that the user-supplied conditional-independence graph G correctly represents, or parsimoniously approximates, the true dependence structure of the prior; if G is misspecified, EnIF forces the posterior update to ignore dependence that actually exists.

Editorial extensions

If this is right

  • At fixed ensemble size, EnIF's average KLD to the true posterior stabilises as resolution increases, whereas sample-covariance EnKF and ES diverge; the paper demonstrates this for a 1D Matérn/Ornstein-Uhlenbeck process.
  • EnIF adapts automatically to dependence strength: in AR-1 experiments, weak dependence produces almost no off-diagonal update while strong dependence produces longer-range updates, with no radius parameter to tune.
  • The framework covers filtering, smoothing, and parameter estimation: smoothing appends past states to $u_t$ and encodes temporal Markov structure in G, while static parameters use graphs from SPDE or hierarchical priors.
  • The EnIF-MDA variant repeats the ES-MDA proof in the canonical parametrisation, so iterative multiple-data-assimilation updates remain available.
  • If H is unknown, sparse regression with early stopping yields a sparse estimate, preserving sparsity in the posterior precision and enabling high-dimensional applications such as 40,000-parameter GRFs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper's experiments, the degrees-of-freedom argument suggests EnIF should also reduce filter divergence in weakly observed high-dimensional nonlinear systems where the localisation radius is hard to tune; this is a testable prediction the paper does not run.
  • The optimal Markov order in the Lorenz-96 experiment grows with ensemble size, which points to a model-selection rule for G itself: choose the smallest graph whose cross-validated loss is statistically indistinguishable from the best, an implicit but not fully developed extension.
  • For non-Markovian priors such as the 2D exponential GRF, G is necessarily approximate; the paper's success there hints at robustness to mild misspecification, but it also predicts that a seriously wrong G would trade spurious correlations for structured bias in the posterior.
  • The sparse-regression component could be tested separately: replacing monotone-LASSO with another sparse estimator inside EnIF would isolate how much of the method's performance comes from precision sparsity versus from sparse H estimation.
Share X Bluesky LinkedIn Reddit HN

Formalized claims in Lean

  1. Claim #1: The load-bearing statement is that the precision matrix, not the covariance, is the right object for ensemble updates: conditional independence appears as exact zeros in the precision, the information-filter update adds observation information to a sparse matrix without dense inversion, and the number of parameters to estimate equals vertices plus edges of G, growing linearly with resolution inste

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes the Ensemble Information Filter (EnIF), a precision-based ensemble update that encodes a conditional independence graph G in the precision matrix, motivated by Markov properties of SPDE discretisations. The author proves in Appendix A that, when the prior precision and observation operator are known, the three-step EnIF update is algebraically equivalent to the EnKF update. The central claim is that estimating the precision under G eliminates the need for ad-hoc localisation while adapting to the strength of dependence. Numerical experiments on a 1D Matérn/OU process, an AR-1 process, the Lorenz-96 system, a 2D stochastic heat equation, and a non-Markovian exponential GRF are used to support filtering, smoothing, and parameter-estimation performance.

Significance. The proposed direction is valuable: the SPDE-to-Markov link for smoothing is theoretically grounded (Section 3.1), the algebraic equivalence in Appendix A is a useful pedagogical and methodological contribution, and the KLD-based framing of ensemble estimation is appealing. If the localisation-free claim were established, EnIF would be a significant alternative to EnKF/ES in high-dimensional DA. However, the key advantage depends on the user-supplied graph G and the Markov order, which are structural hyperparameters selected by validation in Section 6.1, and the paper does not provide a misspecification analysis. The experimental support is limited by the absence of error bars and repeated trials. The central idea is promising but the main claim is not yet fully supported.

major comments (5)
  1. [§4.2, item 3; abstract] The claim that EnIF 'eliminates the need for ad-hoc localisation' because locality is encoded in the precision before estimation is not established. The graph G and the Markov order are user-supplied structural inputs: Section 6.1 selects the Markov order by comparing training and test negative log-likelihood (orders 2, 5, and 7 for n=100, 200, and 500), which is a tuning procedure over a hyperparameter that plays the same role as a localisation radius. The paper should either provide a principled, parameter-free method for selecting G or compare EnIF against localised EnKF/ES where the localisation radius is selected by the same cross-validation criterion; without this, the 'no tuning' claim is not supported.
  2. [§3.2, §6.3] The paper concedes in Section 3.2 that filtering marginals 'may lose strict spatial Markov properties' and only 'likely still exhibit a preference for locality', yet the central claim relies on G being a parsimonious representation of the true dependence structure. Section 6.3 applies EnIF to a 2D anisotropic exponential GRF that is explicitly non-Markovian, imposing a graph without assessing the resulting misspecification. There is no bound, diagnostic, or empirical check on the error from an incorrect G. This is load-bearing because the claimed automatic adaptivity collapses if G is misspecified; the manuscript needs an analysis or at least a principled diagnostic for G misspecification.
  3. [§5, §6.3] The numerical evidence for the central advantages lacks error bars and repeated experiments. Figure 3, Figure 4, Figure 5, and Figure 8 are single-realisation comparisons, and the text makes comparative claims such as 'EnIF achieves a lower KLD' (Section 5.3) and 'EnIF updates are significantly more localised' (Section 6.3) without quantifying uncertainty. The code is referenced only for the Section 6.3 implementation; the main experiments are not reproducible from the manuscript. Error bars or ensemble-of-ensembles statistics are needed to support the claimed superiority.
  4. [§4.2, item 2; §5] The precision estimation step is not specified in enough detail to be reproducible. The paper says to estimate the non-zero elements of a permutation-optimised affine KR-map CπG, but does not give the objective function, the regularisation, or the algorithm for doing so; the experiments refer to an external implementation (Lunde et al., 2022) and a GitHub repository. Without a precise estimator definition, the statistical properties of the EnIF cannot be assessed, and the claimed consistency and scalability are unverifiable.
  5. [§4.2, item 3; §5.2] The statement that EnIF 'will adapt fully to the strength of dependence in the data, requiring no tuning of the radius of convergence' is not demonstrated. Section 5.2 shows that EnIF tracks the exact update for different AR-1 dependence strengths φ, but the graph in that experiment is the known AR-1 graph, so the test does not exercise the regime where G must be inferred or approximated. The adaptive claim is about the estimated precision adapting to the data, but the precision estimate is constrained by G, whose selection is itself a tuning problem (see also the Section 6.1 Markov-order selection). A convincing demonstration would need to compare EnIF against localisation with the radius selected by the same validation criterion.
minor comments (5)
  1. [§5.2, Figure 4 caption] The caption contains a typo: 'correation' should be 'correlation'.
  2. [§6.1, Figure 6 caption] The caption describes the Euler graph as connecting xj−2, xj−1, xj, and xj+1, but the surrounding text says the RK4 graph connects j−6 through j+3; a precise statement of the RK4 stencil and its derivation would improve reproducibility.
  3. [§2.1, Appendix B] The notation for the state vector u and its realisations u^{(i)} is sometimes inconsistent (bold versus non-bold); please unify the notation throughout.
  4. [§5.3] The narrative states 'At very small values of c, the localisation effect is negligible' and 'further increases in c weaken the udpates too much'; apart from the typo 'udpates', the direction of increasing radius in Figure 5 should be clarified so the interpretation of the x-axis is unambiguous.
  5. [§3.1, Eq. (31)] The block-precision formula in Equation (31) is stated without derivation; providing the derivation or a reference for the non-stationary case would help readers verify the Markov structure.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: EnIF's update is algebraically the EnKF and the Markov graph is an external structural input, though the 'no localisation tuning' claim is weakened by validation-based Markov-order selection and a single non-load-bearing self-citation.

full rationale

The paper's derivation chain is self-contained and does not reduce to its own inputs by construction. The EnIF update is shown in Appendix A to be algebraically equivalent to the standard EnKF/Kalman update when the precision and observation map are known, so the method has an independent benchmark rather than defining its target in terms of fitted values. The Markov graph G is an external structural input derived from the SPDE/FEM discretisation or supplied by the user, and the precision entries are subsequently estimated from the ensemble under that fixed graph; no predicted quantity is a renamed fitted parameter. The only self-citation is Lunde et al. (2022) for the direct precision-estimation variant in Section 6.3, which is not load-bearing for the central claim. Two caveats are worth noting but are not circularity under the defined patterns: Section 3.2 concedes that filtering marginals 'may lose strict spatial Markov properties but will likely still exhibit a preference for locality,' and Section 6.1 selects the Markov order by comparing training and test loss, which is a tuning procedure analogous to choosing a localisation radius. These are correctness and generality concerns about the claim that EnIF 'eliminates the need for ad-hoc localisation,' not instances of the paper deriving a result from an equivalent input. Overall, the core derivation is externally grounded and not circular.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The method's adjustment knobs are modeling choices: the graph G and Markov order, regularisation hyperparameters for sparse regression, and the tuning parameters of the baseline methods used in comparisons.

free parameters (4)
  • Conditional independence graph G / Markov order = user-specified; optimal Markov order approx 2, 5, 7 for Lorenz-96 with n = 100, 200, 500
    The graph determines which entries of the precision matrix are forced to zero. It is supplied by the practitioner (Section 7) or selected by test loss (Section 6.1), so it is a hand-chosen or model-selected input, not derived automatically.
  • Localisation radius c (baseline ES) = swept across values in Figure 5
    Used only for the comparison method, not for EnIF. The comparison result depends on how this baseline parameter is tuned.
  • Adaptive localisation threshold 3/sqrt(n) = 3/sqrt(n)
    Fixed threshold for the baseline adaptive localisation in Figure 2 and Section 6.3. This is a tuning choice for the comparison method.
  • Monotone LASSO step size and early-stopping criterion = not specified numerically
    Hyperparameters of Algorithm 1, used to estimate H in Section 6.3. They require tuning and are not stated in the paper.
assumptions (5)
  • standard math For linear local SPDEs of the form L u = W, the solution is Gaussian and Markov, with precision operator L*L (Rozanov 1977; Lindgren et al. 2011).
    Section 3 uses this external mathematical result to justify encoding sparsity from the SPDE discretisation. The paper does not prove it.
  • domain assumption The KR map S is restricted to affine maps, so the approximating distribution Q is Gaussian.
    Section 4 states 'We will further assume that S is of the affine type'. Without this, the precision-matrix information-filter update is not justified. This limits the method's consistency for nonlinear or non-Gaussian priors.
  • domain assumption Mass lumping M~=diag(row sums) preserves the Markov property and does not increase the solution error (Lindgren et al. 2011, Appendix C.5).
    Section 3.1 uses the diagonal approximation to keep the spatial blocks band-sparse. Without it, the spatial Markov structure is lost.
  • ad hoc to paper For filtering, the marginal p(ut) retains enough locality that a parsimonious Markov approximation is valid.
    Section 3.2 concedes that marginalising over time destroys exact spatial Markov properties. The method still imposes a Markov graph, which is an approximation rather than an exact consequence of the SPDE.
  • domain assumption H is known or can be estimated with minimal statistical error and with a sparse structure.
    Section 4.2 states this contingency. In high dimensions H estimation requires regularisation (Appendix E), which introduces additional tuning and potential misspecification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Ensemble Information Filter: Retrieving Markov-information from the SPDE discretisation." pith.science (2026). https://pith.science/paper/5B45KSBT

@misc{pith2026250109016,
  author       = {Pith},
  title        = {Pith review of: An Ensemble Information Filter: Retrieving Markov-information from the SPDE discretisation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5B45KSBT}},
  note         = {Machine review of arXiv:2501.09016}
}
read the original abstract

Ensemble-based Data Assimilation faces significant challenges in high-dimensional systems due to spurious correlations and ensemble collapse. These issues arise from estimating dense dependencies with limited ensemble sizes. This paper introduces the Ensemble Information Filter, which encodes Markov properties directly into the statistical model's precision matrix, leveraging structure from SPDE dynamics to constrain information to propagate locally. EnIF eliminates the need for ad-hoc localisation, improving statistical consistency and scalability. Numerical experiments demonstrate its advantages in filtering, smoothing, and parameter estimation, making EnIF a robust and efficient solution for large-scale data assimilation problems.

Figures

Figures reproduced from arXiv: 2501.09016 by the authors.

Figure 1
Figure 1. Left: 50 realisations of 1-d Matérn (κ = 0.1) Gaussian process as in Example 1 sampled on x ∈ [0, 1], unconditioned on an observation of the endpoint (green dot) at x = 1.0. Right: Conditioned realisations overlayd the unconditional ones from the left plot. The update is a special case of the EnKF and is expressed in Example 1. Red line showcases the first Matérn realisation (left) and how this realisation is update… view at source ↗
Figure 2
Figure 2. Left: Conditioned realisations as in Figure 1 right, but exchanging true covariance with sample covariance [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Average KLD values (49) for the EnIF (orange), EnKF/ES (green), and the Euler-Maruyama scheme (blue) relative to the true analytical posterior of the 1D Matérn process (κ = 1.0) corresponding to the Ornstein-Uhlenbeck process (50). The x-axis shows the level of resolution, indicating the number of discretisation points and thus the number of vertices. 5. EnKF equivalence under dense estimates: Estimating Λ using an … view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Exact, ES and EnIF updates of the first realisation in an [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: KLD values (49) vs. localisation radius of ES/EnKF. Both the [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Conditional independence graphs induced by integrating the Lorenz-96 dynamics (with 40 states) in Equation [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Negative log-likelihood loss versus Markov order for the Lorenz-96 filtering problem. Loss is computed for [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: The average (n = 100) ensemble update across the GRFs at varying discretisation, using different update algorithms. The GRFs are sampled using an anisotropic exponential covariance function. This is a special case of the Matérn covariance, which in the case of two dime…
Figure 9
Figure 9. Figure 9: The graph corresponding to an AR-1 process [PITH_FULL_IMAGE:figures/full_fig_p020_9.png]
Figure 10
Figure 10. Figure 10: Timing LASSO from the python package sklearn and the Monotone LASSO in Algorithm 1, versus an [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 51 canonical work pages

  1. [1]

    Aanonsen, S. I., G. N vdal, D. S. Oliver, A. C. Reynolds, and B. Vall \`e s (2009). The ensemble K alman filter in reservoir engineering--a review. Spe Journal\/ 14\/ (03), 393--412

    work page 2009

  2. [2]

    Akaike, H. (1974). A new look at the statistical model identification. IEEE transactions on automatic control\/ 19\/ (6), 716--723

    work page 1974

  3. [3]

    Amestoy, P. R., T. A. Davis, and I. S. Duff (2004). Algorithm 837: Amd, an approximate minimum degree ordering algorithm. ACM Transactions on Mathematical Software (TOMS)\/ 30\/ (3), 381--388

    work page 2004

  4. [4]

    Anderson, J. L. (2003). A local least squares framework for ensemble filtering. Monthly Weather Review\/ 131\/ (4), 634--642

    work page 2003

  5. [5]

    Learning non-Gaussian graphical models via Hessian scores and triangular transport

    Baptista, R., Y. Marzouk, R. E. Morrison, and O. Zahm (2021). Learning non- G aussian graphical models via H essian scores and triangular transport. arXiv preprint arXiv:2101.03093\/

    work page Pith review arXiv 2021

  6. [6]

    On the representation and learning of monotone triangular transport maps

    Baptista, R., Y. Marzouk, and O. Zahm (2020). On the representation and learning of monotone triangular transport maps. arXiv preprint arXiv:2009.10303\/

    work page Pith review arXiv 2020

  7. [7]

    Burgers, G., P. J. Van Leeuwen, and G. Evensen (1998). Analysis scheme in the ensemble K alman filter. Monthly weather review\/ 126\/ (6), 1719--1724

    work page 1998

  8. [8]

    Claeskens, G. and N. L. Hjort (2008). Model selection and model averaging. Cambridge B ooks\/

    work page 2008

Show all 54 references
  1. [9]

    Cuthill, E. and J. McKee (1969). Reducing the bandwidth of sparse symmetric matrices. In Proceedings of the 1969 24th national conference , pp.\ 157--172

    1969

  2. [10]

    Emerick, A. A. and A. C. Reynolds (2013). Ensemble smoother with multiple data assimilation. Computers & Geosciences\/ 55 , 3--15

    2013

  3. [11]

    Evensen, G. (1994). Sequential data assimilation with a nonlinear quasi-geostrophic model using M onte C arlo methods to forecast error statistics. Journal of Geophysical Research: Oceans\/ 99\/ (C5), 10143--10162

    1994

  4. [12]

    Evensen, G. (2003). The ensemble K alman filter: T heoretical formulation and practical implementation. Ocean dynamics\/ 53\/ (4), 343--367

    2003

  5. [13]

    Evensen, G. et al. (2009). Data assimilation: the ensemble Kalman filter , Volume 2. Springer

    2009

  6. [14]

    Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of S tatistics\/ , 1189--1232

    2001

  7. [15]

    Hamill, T. M., J. S. Whitaker, and C. Snyder (2001). Distance-dependent filtering of background error covariance estimates in an ensemble K alman filter. Monthly Weather Review\/ 129\/ (11), 2776--2790

    2001

  8. [16]

    Taylor, R

    Hastie, T., J. Taylor, R. Tibshirani, and G. Walther (2007). Forward stagewise regression and the monotone lasso. Electronic Journal of Statistics\/ 1 , 1--29

    2007

  9. [17]

    Tibshirani, and J

    Hastie, T., R. Tibshirani, and J. H. Friedman (2009). The elements of statistical learning: data mining, inference, and prediction , Volume 2. Springer

    2009

  10. [18]

    Hoerl, A. E. and R. W. Kennard (1970). Ridge regression: Biased estimation for nonorthogonal problems. Technometrics\/ 12\/ (1), 55--67

    1970

  11. [19]

    Houtekamer, P. L. and H. L. Mitchell (2001). A sequential ensemble K alman filter for atmospheric data assimilation. Monthly Weather Review\/ 129\/ (1), 123--137

    2001

  12. [20]

    Houtekamer, P. L. and H. L. Mitchell (2005). Ensemble K alman filtering. Quarterly Journal of the Royal Meteorological Society\/ 131\/ (613), 3269--3289

    2005

  13. [21]

    Huber, P. J. et al. (1967). The behavior of maximum likelihood estimates under nonstandard conditions. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability , Volume 1, pp.\ 221--233. Berkeley, CA: University of California Press

    1967

  14. [22]

    Hunt, B. R., E. J. Kostelich, and I. Szunyogh (2007). Efficient data assimilation for spatiotemporal chaos: A local ensemble transform K alman filter. Physica D: Nonlinear Phenomena\/ 230\/ (1-2), 112--126

    2007

  15. [23]

    Kalman, R. (1960). A new approach to linear filtering and prediction problems. Trans. ASME, D\/ 82 , 35--44

    1960

  16. [24]

    Kalman, R. E. and R. S. Bucy (1961). New results in linear filtering and prediction theory. Journal of Basic Engineering\/ 83\/ (1), 95--108

    1961

  17. [25]

    Karypis, G. and V. Kumar (1998). A fast and high quality multilevel scheme for partitioning irregular graphs. SIAM Journal on scientific Computing\/ 20\/ (1), 359--392

    1998

  18. [26]

    Kleppe, T. S. (2024). Log-density gradient covariance and automatic metric tensors for R iemann manifold M onte C arlo methods. Scandinavian Journal of Statistics\/

    2024

  19. [27]

    Kloeden, P. E. and E. Platen (1992). Numerical solution of stochastic differential equations . Springer

    1992

  20. [28]

    Konishi, S. and G. Kitagawa (1996). Generalised information criteria in model selection. Biometrika\/ 83\/ (4), 875--890

    1996

  21. [29]

    Nielsen, C

    Kristensen, K., A. Nielsen, C. W. Berg, H. Skaug, and B. M. Bell (2016). T M B : A utomatic differentiation and L aplace approximation. Journal of Statistical Software\/ 70\/ (i05)

    2016

  22. [30]

    Bolin, and H

    Lindgren, F., D. Bolin, and H. Rue (2022). The SPDE approach for G aussian and non- G aussian fields: 10 years and still running. Spatial Statistics\/ , 100599

    2022

  23. [31]

    Rue, and J

    Lindgren, F., H. Rue, and J. Lindstr \"o m (2011). An explicit link between G aussian fields and G aussian M arkov random fields: the stochastic partial differential equation approach. Journal of the Royal Statistical Society: Series B (Statistical Methodology)\/ 73\/ (4), 423--498

    2011

  24. [32]

    Lorenz, E. N. (1995). Predictability: A problem partly solved. In Seminar on Predictability , Shinfield Park, Reading. ECMWF

    1995

  25. [33]

    Lorenz, E. N. and K. A. Emanuel (1998). Optimal sites for supplementary weather observations: Simulation with a small model. Journal of the Atmospheric Sciences\/ 55\/ (3), 399--414

    1998

  26. [34]

    Lunde, B. . S., F. Curic, and S. Sortland (2022). Graph SPME : M arkov precision matrix estimation and asymptotic S tein-type shrinkage. arXiv preprint arXiv:2205.07584\/

    2022 arXiv

  27. [35]

    Moselhy, M

    Marzouk, Y., T. Moselhy, M. Parno, and A. Spantini (2016). An introduction to sampling via measure transport. arXiv preprint arXiv:1602.05023\/

    2016 arXiv

  28. [36]

    Baxter, P

    Mason, L., J. Baxter, P. Bartlett, and M. Frean (1999). Boosting algorithms as gradient descent. Advances in neural information processing systems\/ 12

    1999

  29. [37]

    Moore, J. B. and B. Anderson (1979). Optimal filtering . Prentice-Hall New York

    1979

  30. [38]

    ksendal, B. (2003). Stochastic differential equations. In Stochastic differential equations , pp.\ 65--84. Springer

    2003

  31. [39]

    Ott, E., B. R. Hunt, I. Szunyogh, A. V. Zimin, E. J. Kostelich, M. Corazza, E. Kalnay, D. Patil, and J. A. Yorke (2004). A local ensemble K alman filter for atmospheric data assimilation. Tellus A: Dynamic Meteorology and Oceanography\/ 56\/ (5), 415--428

    2004

  32. [40]

    Raanes, P. N., A. S. Stordal, and R. J. Lorentzen (2023). Review of ensemble gradients for robust optimisation. arXiv preprint arXiv:2304.12136\/

    2023 arXiv

  33. [41]

    Baptista, D

    Ramgraber, M., R. Baptista, D. McLaughlin, and Y. Marzouk (2023a). Ensemble transport smoothing. P art i: Unified framework. Journal of Computational Physics: X\/ 17 , 100134

    2023

  34. [42]

    Baptista, D

    Ramgraber, M., R. Baptista, D. McLaughlin, and Y. Marzouk (2023b). Ensemble transport smoothing. P art ii: Nonlinear updates. Journal of Computational Physics: X\/ 17 , 100133

    2023

  35. [43]

    Rosenblatt, M. (1952). Remarks on a multivariate transformation. The A nnals of M athematical S tatistics\/ 23\/ (3), 470--472

    1952

  36. [44]

    Rozanov, J. A. (1977). Markov random fields and stochastic partial differential equations. Mathematics of the USSR-Sbornik\/ 32\/ (4), 515

    1977

  37. [45]

    Rue, H. and L. Held (2005). Gaussian Markov random fields: theory and applications . Chapman and Hall/CRC

    2005

  38. [46]

    Martino, and N

    Rue, H., S. Martino, and N. Chopin (2009). Approximate B ayesian inference for latent G aussian models by using integrated nested L aplace approximations. Journal of the R oyal S tatistical S ociety: S eries B ( S tatistical M ethodology)\/ 71\/ (2), 319--392

    2009

  39. [47]

    Hanea, and A

    Sebacher, B., R. Hanea, and A. S. Stordal (2017). An adaptive pluri- G aussian simulation model for geological uncertainty quantification. Journal of Petroleum Science and Engineering\/ 158 , 494--508

    2017

  40. [48]

    Baptista, and Y

    Spantini, A., R. Baptista, and Y. Marzouk (2019). Coupling techniques for nonlinear ensemble filtering, arxiv. arXiv preprint arXiv:1907.00389\/ 30

    2019 arXiv

  41. [49]

    Bigoni, and Y

    Spantini, A., D. Bigoni, and Y. Marzouk (2018). Inference via low-dimensional couplings. The Journal of Machine Learning Research\/ 19\/ (1), 2639--2709

    2018

  42. [50]

    Stone, M. (1974). Cross-validatory choice and assessment of statistical predictions. Journal of the Royal Statistical Society. Series B (Methodological)\/ , 111--147

    1974

  43. [51]

    Takeuchi, K. (1976). Distribution of information statistics and validity criteria of models. Mathematical Science\/ 153 , 12--18

    1976

  44. [52]

    Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology\/ 58\/ (1), 267--288

    1996

  45. [53]

    Uhlenbeck, G. E. and L. S. Ornstein (1930). On the theory of the B rownian motion. Physical R eview\/ 36\/ (5), 823

    1930

  46. [54]

    Van der Vaart, A. W. (2000). Asymptotic statistics , Volume 3. Cambridge U niversity P ress

    2000

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.