IEnSF: Iterative Ensemble Score Filter for Reducing Error in Posterior Score Estimation in Nonlinear Data Assimilation

Feng Bao; Guannan Zhang; Zezhong Zhang

REVIEW 3 major objections 4 minor 1 cited by

IEnSF: Iterative Ensemble Score Filter for Reducing Error in Posterior Score Estimation in Nonlinear Data Assimilation

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read An iterative outer loop progressively reduces posterior score error in nonlinear score-based data assimilation by replacing a heuristic likelihood weighting with a Gaussian-mixture-derived correction.

desk verdict The exact GM-prior posterior score (Theorem 3.5) is a real step beyond EnSF's heuristic, but the headline claim of reduced score error is not actually measured and the main approximation is a Jensen gap with no bound. read the letter →

arxiv 2510.20159 v2 pith:KFLUQDKY submitted 2025-10-23 math.NA cs.NA

Zezhong Zhang , Feng Bao , Guannan Zhang This is my paper · ORCID

classification math.NAcs.NA MSC 62M2062F1565C3060H10

keywords score-basedgenerativemodeldataassimilationensemblescorefilterGaussianmixturepriorposteriorestimationLorenz-96reverse-timeSDEiterativeBayesianinference

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 proposes IEnSF, an iterative version of the Ensemble Score Filter, to fix a structural error in how posterior scores are approximated in nonlinear data assimilation. It shows that when the prior is a Gaussian mixture, the exact posterior score splits into an observation-weighted mixture of prior component scores plus a time-scaled conditional expectation of the likelihood score. The central claim is that replacing that expectation with a single evaluation at a Kalman-updated point, then repeatedly solving the reverse diffusion SDE to refine a reference Gaussian, progressively reduces the posterior score estimation error. The authors demonstrate reduced KL divergence to exact posteriors in linear and nonlinear low-dimensional problems and lower RMSE than LETKF on a 1000-dimensional Lorenz-96 system with sparse, nonlinear arctan observations. A sympathetic reader would care because the method offers a principled replacement for heuristic likelihood weighting, with the iterative loop as the mechanism that corrects bias.

What carries the argument

The central object is Theorem 3.5's posterior-score decomposition and its computable surrogate in Eq. (57). The decomposition isolates the only intractable object, the conditional expectation of the likelihood score; the surrogate replaces it with a single evaluation at μ̄*_0(z_t) = μ* + α_t Σ* (α_t² Σ* + β_t² I)^{-1}(z_t − α_t μ*), computed from an iteratively refitted Gaussian N(μ*,Σ*). The analytic factor J(t) replaces the heuristic damping schedule h(t)=1−t, and the observation-dependent weight w_obs(k,z_t,y) lets the observation reshape the prior mixture weights. Together these pieces carry the argument that reverse-time SDE sampling is driven by a posterior score corrected toward the B

What would settle it

Take a low-dimensional nonlinear observation map, such as arctan or a norm-based operator, for which the conditional expectation E[S_{Y|X}(y|Z_0)|Z_t=z_t,Y=y] can be computed to high accuracy by dense Monte Carlo; run IEnSF with and without the one-point replacement. If the KL divergence between IEnSF posterior samples and an MCMC ground truth does not decrease monotonically with iteration, or is worse than EnSF in that setting, the claim that iteration gradually reduces posterior score error is falsified.

Watch

Extended reading notes

Core claim

Under a Gaussian-mixture prior, the exact posterior score of the diffusion process is S_{Z_t|Y}(z_t|y) = sum_k p_{ξ|Z_t,Y}(k|z_t,y) S_{Z_t|k}(z_t) + J(t) E[S_{Y|X}(y|Z_0) | Z_t=z_t, Y=y]. This identity is the paper's theoretical anchor: it shows that the observation should enter both through the mixture weights and through a precisely time-scaled likelihood-score term, not through the heuristic additive damping used by EnSF. The practical task is to approximate the intractable conditional expectation. The one-point replacement evaluates S_{Y|X} at a point μ̄*_0(z_t) derived from a Gaussian reference posterior N(μ*,Σ*) via a Kalman update, and an outer iterative loop re-fits that reference fr

Load-bearing premise

The load-bearing premise is that a single evaluation at the Kalman-updated reference mean captures the conditional expectation of the likelihood score: exact only when the observation operator is linear, and uncontrolled for strongly nonlinear operators, with the iterative loop additionally assuming that re-fitting the reference Gaussian from its own samples converges to the true posterior.

Editorial extensions

If this is right

If the central claim holds, score-based ensemble filters can drop the heuristic likelihood-weighting schedule in favor of an analytically derived time scale that adapts to ensemble covariance.
In high-dimensional sparse-observation settings, the observation-dependent mixture weights provide a correlation-driven update path for unobserved components, mitigating the stalling that plagues the original EnSF.
The iterative loop yields an operational convergence criterion: continue refining the reference Gaussian until successive fits stop changing, so the number of refinement iterations is self-tuned rather than fixed.
Because the final posterior is still sampled by the reverse SDE, the method inherits non-Gaussian posterior flexibility while using a Gaussian only as an auxiliary evaluation tool.
When the observation operator is linear, the one-point replacement is exact, so the IEnSF posterior score recovers the true Bayesian update and reduces to a direct generalization of EnSF without structural bias.

Reading between the lines

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

Editorial inference: the same one-point-replacement step is exactly where the method would fail under strongly saturated observation maps; computing or bounding the Jensen gap for arctan- and norm-type operators, rather than relying on RMSE comparison, would turn the method into a certified filter.
Editorial inference: the iteration structure suggests a natural adaptive schedule for the variance-splitting parameter γ — start with a near-particle prior representation and increase Gaussianity as the reference posterior sharpens, potentially improving convergence in strongly non-Gaussian posteriors.
Editorial inference: a testable extension is to apply the reverse-SDE posterior-score correction outside filtering entirely, for example to Bayesian inverse problems where the Gaussian-mixture prior is replaced by a learned or kernel-density prior, with J(t) as the only problem-specific ingredient.
Editorial inference: the claimed reduction in posterior-score error should be measurable directly — comparing S_{Y|X}(y|μ̄*_0(z_t)) against a dense Monte Carlo estimate of the conditional expectation along reverse paths would provide a diagnostic that RMSE comparisons cannot.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Bottom line: the theoretical core is sound and genuinely improves on EnSF, but the paper's headline claim—that the iterative algorithm reduces posterior score estimation error—is not supported by the experiments as reported. The derivation of the exact posterior score under a Gaussian mixture prior, with the J(t) scaling and observation-aware mixture weights, is a legitimate contribution. It is not in the cited EnSF papers, and the appendices carry the derivation carefully. The low-dimensional experiments show IEnSF clearly beating EnSF on KL divergence and matching a particle filter in the harmonic oscillator; the 1000-D Lorenz-96 result against LETKF under the arctan observation operator is encouraging if it holds up.

Soft spots, in order of importance. First, the computable score in Eq. (57) replaces an intractable conditional expectation with a single evaluation of the likelihood score at a Kalman-style point from a reference Gaussian. That is a Jensen gap, and the paper gives neither a bound nor a diagnostic. The low-dimensional tests happen to live in regimes where prior and posterior are close, so they do not stress this gap. Second, Algorithm 1 is a fixed-point iteration: it refits the reference Gaussian to samples produced using that same reference, and convergence to the true posterior is not proved. The empirical convergence after a few iterations is nice, but it is not the claim in the abstract. Third, the abstract says the method reduces posterior score estimation error, yet no experiment measures score error directly; everything is sample-level KL or RMSE. Fourth, the Lorenz-96 results are single-trajectory RMSE without error bars, the ensemble size for IEnSF is not stated, and hyperparameters are grid-searched on the evaluation trajectory. Fifth, Eqs. (48)–(49) as printed look wrong: they show a bare quadratic form where the Gaussian density should have exp(-Q/2) and a determinant. If the implementation follows the text, the mixture weights are not the ones derived; if it does not, the text should be fixed.

None of this is fatal. The theorem is a real result, and the method is plausible. But the paper currently overstates what is established. It deserves a serious referee, not a desk reject, because the theoretical derivation is a legitimate finding and the approach may well be important. The referee should ask for code, corrected equations, a direct or at least indirect check of the score error, and basic robustness reporting. This is a paper to engage with, but not one whose central claim should be taken on faith.

Referee Report

3 major / 4 minor

Summary. The paper proposes the Iterative Ensemble Score Filter (IEnSF), a modification of the Ensemble Score Filter for nonlinear data assimilation. Under a Gaussian mixture prior (Assumption 3.1), the authors derive an exact expression for the posterior score (Theorem 3.5) as the sum of observation-weighted component scores plus a time-scaled conditional expectation of the likelihood score. Since the conditional expectation and mixture weights are intractable, they approximate the weights by linearizing the observation operator (Section 3.2.1) and replace the conditional expectation by a single evaluation of the likelihood score at a Kalman-type mean computed from a reference Gaussian (Section 3.2.2). This reference is iteratively refitted from posterior samples generated by the same approximate score (Algorithm 1). Numerical experiments include low-dimensional Bayesian inference with Gaussian and non-Gaussian posteriors, a harmonic oscillator, and a 1000-dimensional Lorenz-96 model with direct and arctan observation operators, reporting KL divergences and RMSEs relative to Kalman, particle, MCMC, EnKF, EnSF, and LETKF baselines.

Significance. The exact posterior score decomposition in Theorem 3.5 is a genuinely useful theoretical contribution: it replaces the heuristic damping function h(t) in EnSF with the analytically derived J(t), and it introduces observation-dependent prior weights that are absent from the original EnSF. The appendices provide a careful derivation of the theorem, and the low-dimensional experiments show that the iterative scheme can reduce KL divergence to a known posterior in the tested cases. However, the paper's headline claim—that the algorithm 'can gradually reduce the posterior score estimation error'—is not established by the theory or by the experiments as reported. The practical scheme rests on two unproven premises: a single-point Jensen-type approximation of an intractable conditional expectation, and convergence of the self-referential fixed-point iteration in Algorithm 1. Neither is accompanied by an error bound or a direct measurement of posterior score error. The numerical evidence is therefore suggestive but does not support the abstract's central assertion.

major comments (3)

[Section 3.2.2, Eq. (55)] The approximation μ_{S_Y|X}(z_t,y) ≈ S_Y|X(y|μ̄*_0(z_t)) replaces an intractable conditional expectation by a single likelihood-score evaluation. For nonlinear observation operators (arctan, quadratic-norm), S_Y|X is nonlinear in z_0, and the resulting Jensen gap is uncontrolled. No bound, Lipschitz condition, or numerical diagnostic is provided. Since Eq. (57) is the score used in reverse-SDE sampling, this gap directly affects the paper's central claim of reducing posterior score error. The authors acknowledge the approximation error after Eq. (50), but do not quantify it or show that the iterative refinement reduces it.
[Section 3.2.3 / Algorithm 1] The iterative refinement fits N(μ*,Σ*) to samples generated using the same approximate score (Eq. (57)). This is a self-referential fixed point: the reference controls the likelihood-score evaluation, and the samples determine the next reference. The paper proves neither contraction to the true posterior nor monotone reduction of score error or KL divergence. Figures 2 and 7 show empirical KL decrease in specific low-dimensional settings, but the general claim 'gradually reduce the posterior score estimation error' in the abstract is not supported by a theorem. A convergence or error-reduction result, even under restrictive conditions, is needed for the theoretical claim as stated.
[Section 4] The abstract and introduction claim that IEnSF 'substantially reduces the error in posterior score estimation.' However, Section 4 does not report a single direct measurement of score error. The low-dimensional experiments report KL divergence between posterior samples and a Gaussian ground truth (Figures 2, 5, 7), which is a distribution-level discrepancy, not a score error. The 1000D Lorenz-96 experiments (Figures 9 and 10) report only RMSE, on a single trajectory, with no error bars and no stated ensemble size for IEnSF. For a paper whose central claim is about score error, the absence of any estimate of ||S_hat - S_true|| (e.g., in a low-dimensional nonlinear case with MCMC-based ground truth) is a significant gap. Either add such a measurement or weaken the claim.

minor comments (4)

[Section 2.2.2] Typo: 'satisfies satisfies the normalization condition' should read 'satisfies the normalization condition'.
[Section 3.1, after Eq. (23)] Typo: 'postieor' should be 'posterior'.
[Section 4.3] The reproducibility of the Lorenz-96 experiments is limited: report the ensemble size used for IEnSF, the value of γ in Eq. (22), the smoothing parameters η1, η2, the iteration count M, and the convergence tolerance tol. Also report results over multiple independent truth trajectories or initial ensembles, with error bars or percentiles, rather than a single trajectory as in Figures 9 and 10.
[Appendix C] The variance-splitting identity Var(GMM) = (1−γ²)σ̄² + γ²σ̄² assumes independence between the mixture means and the shared covariance; for finite samples this is exact only in expectation. This is not a blocking issue, but a short clarification would help.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the posterior score derivation is self-contained and the iterative refinement is a fixed-point scheme, not a fit renamed as a prediction.

full rationale

The paper's derivation chain is not circular. The exact posterior score in Theorem 3.5 is derived from the stated Gaussian-mixture prior (Assumption 3.1) using the forward/reverse diffusion kernels in Propositions 3.3–3.4 and standard Bayes conditioning; it does not assume the target posterior or the eventual error-reduction claim. The computable approximation in Eq. (57) replaces the intractable conditional expectation E[S_Y|X(y|Z_0)|Z_t=z_t,Y=y] with a single evaluation at the Kalman-type point μ̄*_0(z_t) (Eqs. (50)–(55)); the paper explicitly identifies this as a Jensen-gap approximation and notes it is exact only for linear observation operators, so it is presented as an acknowledged approximation rather than as derived from the target. Algorithm 1 iteratively refits the auxiliary Gaussian N(μ*,Σ*) from samples generated using that same reference; this is a self-consistent fixed-point iteration analogous to EM or iterated filtering, not a case where the output equals the input by construction. The convergence of this iteration is asserted without proof, and the Jensen gap is uncontrolled, but those are rigor/accuracy limitations, not definitional circularity. Crucially, the numerical validation is anchored to external ground truths: the analytic Kalman posterior (Figs. 1–2), long-run MCMC (Figs. 3–4), and an independently simulated Lorenz-96 truth trajectory (Figs. 8–10), with KL divergence and RMSE measured against those references rather than against the algorithm's own fitted parameters. The citations to the authors' prior EnSF work are background and baseline comparisons, not the authority for the central theorem or the error-reduction claim. Therefore, no load-bearing step reduces to its own inputs by construction.

Assumptions & free parameters 5 free parameters · 7 assumptions · 1 invented entities

The method rests on seven explicit or implicit premises. The GM-prior assumption (Assumption 3.1) and Gaussian observation noise are domain assumptions that make the diffusion-process marginals and reverse kernels Gaussian mixtures (Prop. 3.3); the linear-SDE solution Z_t = α_t Z_0 + β_t ε (Lemma A.1) and the conditional-independence of Y and Z_t given Z_0 are standard math/domain facts. The local linearization of M (Eq. 47) and the Gaussian reference posterior N(μ*,Σ*) (Eq. 54) are ad hoc approximations introduced to make the intractable terms tractable, and the fixed-point convergence of the reference-refinement loop is an unproven implicit assumption. Free parameters: γ controls the variance split of the GM prior, η1/η2 smooth the reference updates, M/tol control the outer loop, and localization/inflation were grid-searched per method on the L96 problem; none of these values is reported.

free parameters (5)

γ (variance-splitting parameter) = not reported (grid-searched in L96 experiments)
Controls construction of the GM prior: μ_k = √(1−γ²)(x_k−x̄)+x̄, Σ=γΣ̄ (Eq. 22, Appendix C). Determines how Gaussian vs. particle-like the prior mixture is; the method's accuracy depends on it.
η1, η2 (reference-posterior smoothing) = not reported
Introduced in Alg. 1, step 7, 'for numerical stability' when updating μ*,Σ*; values never specified.
M (max refinement iterations) and tol (convergence threshold) = not reported
Alg. 1, lines 3 and 6; figures suggest ~3-5 iterations suffice but the stopping metric d (unnamed) and tolerance are unspecified.
Inflation/localization settings for Σ̄ (IEnSF) and LETKF = best-performing per grid search, values not given
Section 4.3: 'we fine-tune hyperparameters such as inflation and covariance localization through grid search, and the best-performing configuration is reported for each method.'
Reverse-SDE discretization (integrator, step count) = not reported
The reverse SDE/ODE in Alg. 1 line 4 is never specified beyond 'simulate the reverse SDE/ODE'; affects sampling accuracy and cost.

assumptions (7)

domain assumption The prior distribution is a Gaussian mixture with uniform weights and shared covariance (Assumption 3.1, Eq. 21)
Used to make the diffusion-process marginals and reverse kernels Gaussian mixtures (Prop. 3.3); in practice enforced by the variance-splitting construction of Eq. (22).
domain assumption Observation noise is zero-mean Gaussian with known covariance Σ_obs (Eq. 18)
Gives the likelihood score S_{Y|X} = −∇M^T Σ_obs^{-1}(M(x)−y) used throughout; the derivation of I2 in Appendix B relies on the identity ∇p = p·S.
standard math The forward diffusion is a linear SDE with the α_t=1−t, β_t²=t schedule so that Z_t = α_t Z_0 + β_t ε (Lemma A.1)
Standard linear-SDE solution; Gaussian marginals and reverse kernels follow from Gaussian conditioning.
domain assumption Y is conditionally independent of Z_t given Z_0 (Appendix B, Eq. B.11)
Justified by the observation model (Y depends on state only); needed to keep p_{Y|Z_t,ξ} an integral against the reverse kernel.
ad hoc to paper The observation operator M is locally linearized about the reverse-kernel means for the mixture weights (Eq. 47, Eqs. 48-49)
Makes w_obs analytically tractable; the paper argues the linearization is only inside the weight integral, not on the final posterior.
ad hoc to paper The reference posterior for the unknown true posterior X|Y is Gaussian N(μ*,Σ*) (Section 3.2.2, Eq. 54)
Required to evaluate the single-point likelihood score; iterated in Alg. 1, but no proof that the fixed point is the true posterior.
ad hoc to paper The fixed-point iteration in Alg. 1 converges to the true posterior
Assumed implicitly; the paper claims 'converges toward the true posterior' based on empirical KL/RMSE plots only.

invented entities (1)

Reference posterior Gaussian N(μ*,Σ*) independent evidence
purpose: Auxiliary Gaussian used to define the single-point evaluation point μ̄*_0(z_t) and as the iteration's fixed-point anchor in Alg. 1
Not a physical entity; it is an algorithmic approximation. It has a falsifiable handle: the final posterior samples are compared to external ground truth (analytic KF posterior, MCMC, known L96 truth), so a wrong fixed point would show up as high KL/RMSE.

how reviews work

0 comments

Cite this review

Pith. "Pith review of IEnSF: Iterative Ensemble Score Filter for Reducing Error in Posterior Score Estimation in Nonlinear Data Assimilation." pith.science (2026). https://pith.science/paper/KFLUQDKY

@misc{pith2026251020159,
  author       = {Pith},
  title        = {Pith review of: IEnSF: Iterative Ensemble Score Filter for Reducing Error in Posterior Score Estimation in Nonlinear Data Assimilation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KFLUQDKY}},
  note         = {Machine review of arXiv:2510.20159}
}

read the original abstract

The Ensemble Score Filter (EnSF) is a score-based diffusion model approach for solving high-dimensional and nonlinear data assimilation problems. While initial applications of EnSF to the Lorenz-96 model and the quasi-geostrophic system showed potential, the current method employs a heuristic weighted sum to combine the prior and the likelihood score functions. This introduces a structural error into the estimation of the posterior score function in the nonlinear setting. This work addresses this challenge by developing an iterative ensemble score filter (IEnSF) that applies an iterative algorithm as an outer loop around the reverse-time stochastic differential equation solver. When the state dynamics or the observation operator is nonlinear, the iterative algorithm can gradually reduce the posterior score estimation error by improving the accuracy of approximating the conditional expectation of the likelihood score function. The number of iterations required depends on the distance between the prior and posterior distributions. Numerical experiments demonstrate that the IEnSF algorithm substantially reduces the error in posterior score estimation in the nonlinear setting and thus improves the accuracy of tracking high-dimensional dynamical systems.

Figures

Figures reproduced from arXiv: 2510.20159 by the authors.

**Figure 2.** KL divergence and posterior samples for the linear observation model. [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗

**Figure 3.** Non-Gaussian posterior sampling with a nonlinear observation operator. Results are shown for IEnSF [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗

Figures from the paper (7 more)

**Figure 4.** Figure 4: Bias correction in posterior sampling under a nonlinear observation operator. Compared methods include [PITH_FULL_IMAGE:figures/full_fig_p019_4.png]

**Figure 5.** Figure 5: KL divergence of posterior ensembles against the KF posterior for the harmonic oscillator problem. Each [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]

**Figure 6.** Figure 6: Posterior means and spreads for the two state dimensions of the harmonic oscillator. PF, EnKF, and IEnSF [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]

**Figure 7.** Figure 7: Average KL divergence between the IEnSF and KF posteriors as a function of iteration number. The KL [PITH_FULL_IMAGE:figures/full_fig_p021_7.png]

**Figure 8.** Figure 8: RMSE divergence of the Lorenz-96 trajectory over time. Even with the true initial condition, small numerical [PITH_FULL_IMAGE:figures/full_fig_p022_8.png]

**Figure 9.** Figure 9: RMSE of observed and unobserved components for the direct observation model. Both LETKF and IEnSF [PITH_FULL_IMAGE:figures/full_fig_p022_9.png]

**Figure 10.** Figure 10: RMSE of observed and unobserved components for the nonlinear [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Shopping Companion: Benchmarking and Training LLM Agents for Long-Horizon Preference-Grounded E-Commerce Tasks
cs.CL 2026-03 unverdicted novelty 5.0 of 10

A new long-horizon preference-grounded shopping benchmark shows SOTA LLMs below 70% success, while a 4B model fine-tuned with tool-wise process rewards beats stronger baselines.

Reference graph

Works this paper leans on

33 extracted references · 3 linked inside Pith · cited by 1 Pith paper

[1]

G. Evensen, Sequential data assimilation with a nonlinear quasi-geostrophic model using monte carlo methods to forecast error statistics, Journal of Geophysical Research: Oceans 99 (C5) (1994) 10143–10162

1994
[2]

P. L. Houtekamer, H. L. Mitchell, Data assimilation using an ensemble kalman filter technique, Monthly weather review 126 (3) (1998) 796–811

1998
[3]

Shaman, A

J. Shaman, A. Karspeck, Forecasting seasonal outbreaks of influenza, Proceedings of the Na- tional Academy of Sciences 109 (50) (2012) 20425–20430

2012
[4]

R. E. Kalman, A new approach to linear filtering and prediction problems (1960)

1960
[5]

Evensen, Data assimilation: the ensemble Kalman filter, Springer, 2009

G. Evensen, Data assimilation: the ensemble Kalman filter, Springer, 2009

2009
[6]

A. H. Jazwinski, Stochastic processes and filtering theory, Courier Corporation, 2013

2013
[7]

N. J. Gordon, D. J. Salmond, A. F. Smith, Novel approach to nonlinear/non-gaussian bayesian state estimation, in: IEE proceedings F (radar and signal processing), Vol. 140, IET, 1993, pp. 107–113

1993
[8]

Gelb, et al., Applied optimal estimation, MIT press, 1974

A. Gelb, et al., Applied optimal estimation, MIT press, 1974

1974

Show all 33 references

[9]

S. J. Julier, J. K. Uhlmann, New extension of the kalman filter to nonlinear systems, in: Signal processing, sensor fusion, and target recognition VI, Vol. 3068, Spie, 1997, pp. 182–193

1997
[10]

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

2007
[11]

Kitagawa, Monte carlo filter and smoother for non-gaussian nonlinear state space models, Journal of computational and graphical statistics 5 (1) (1996) 1–25

G. Kitagawa, Monte carlo filter and smoother for non-gaussian nonlinear state space models, Journal of computational and graphical statistics 5 (1) (1996) 1–25

1996
[12]

Doucet, N

A. Doucet, N. De Freitas, N. J. Gordon, et al., Sequential Monte Carlo methods in practice, Vol. 1, Springer, 2001

2001
[13]

Bengtsson, P

T. Bengtsson, P. Bickel, B. Li, Curse-of-dimensionality revisited: Collapse of the particle fil- ter in very large scale systems, in: Probability and statistics: Essays in honor of David A. Freedman, Vol. 2, Institute of Mathematical Statistics, 2008, pp. 316–335

2008
[14]

Snyder, T

C. Snyder, T. Bengtsson, P. Bickel, J. Anderson, Obstacles to high-dimensional particle filter- ing, Monthly Weather Review 136 (12) (2008) 4629–4640

2008
[15]

M. K. Pitt, N. Shephard, Filtering via simulation: Auxiliary particle filters, Journal of the American statistical association 94 (446) (1999) 590–599

1999
[16]

W. R. Gilks, C. Berzuini, Following a moving target—monte carlo inference for dynamic bayesian models, Journal of the Royal Statistical Society: Series B (Statistical Methodology) 63 (1) (2001) 127–146

2001
[17]

Rebeschini, R

P. Rebeschini, R. van Handel, Can local particle filters beat the curse of dimensionality, Annals of Applied Probability 25 (2013) 2809–2866. 24

2013
[18]

Kingma, T

D. Kingma, T. Salimans, B. Poole, J. Ho, Variational diffusion models, Advances in neural information processing systems 34 (2021) 21696–21707

2021
[19]

Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, B. Poole, Score-based gener- ative modeling through stochastic differential equations, International Conference on Learning Representations (2021)

2021
[20]

J. Ho, A. Jain, P. Abbeel, Denoising diffusion probabilistic models, Advances in neural infor- mation processing systems 33 (2020) 6840–6851

2020
[21]

J. Ho, T. Salimans, A. Gritsenko, W. Chan, M. Norouzi, D. J. Fleet, Video diffusion models, Advances in neural information processing systems 35 (2022) 8633–8646

2022
[22]

Karras, M

T. Karras, M. Aittala, T. Aila, S. Laine, Elucidating the design space of diffusion-based gen- erative models, Advances in neural information processing systems 35 (2022) 26565–26577

2022
[23]

Kawar, M

B. Kawar, M. Elad, S. Ermon, J. Song, Denoising diffusion restoration models, Advances in neural information processing systems 35 (2022) 23593–23606

2022
[24]

Chung, J

H. Chung, J. Kim, M. T. Mccann, M. L. Klasky, J. C. Ye, Diffusion posterior sampling for general noisy inverse problems, arXiv preprint arXiv:2209.14687 (2022)

2022 arXiv
[25]

Zhang, C

Z. Zhang, C. Tatsuoka, D. Xiu, G. Zhang, Exact conditional score-guided generative modeling for amortized inference in uncertainty quantification, arXiv preprint arXiv:2506.18227 (2025)

2025 arXiv
[26]

F. Bao, Z. Zhang, G. Zhang, A score-based nonlinear filter for data assimilation, Journal of Computational Physics 514 (2024) 113207

2024
[27]

F. Bao, Z. Zhang, G. Zhang, An ensemble score filter for tracking high-dimensional nonlin- ear dynamical system, Computer Methods in Applied Mechanics and Engineering 432 (2024) 117447

2024
[28]

D. Lu, Y. Liu, Z. Zhang, F. Bao, G. Zhang, A diffusion-based uncertainty quantification method to advance e3sm land model calibration, Journal of Geophysical Research: Machine Learning and Computation 1 (3) (2024) e2024JH000234

2024
[29]

F. Bao, H. G. Chipilski, S. Liang, G. Zhang, J. S. Whitaker, Nonlinear ensemble filtering with diffusion models: Application to the surface quasigeostrophic dynamics, Monthly Weather Review 153 (7) (2025) 1155–1169

2025
[30]

Liang, H

S. Liang, H. Tran, F. Bao, H. G. Chipilski, P. J. van Leeuwen, G. Zhang, Ensemble score filter with image inpainting for data assimilation in tracking surface quasi-geostrophic dynamics with partial observations, arXiv preprint arXiv:2501.12419 (2025)

2025 arXiv
[31]

B. D. Anderson, Reverse-time diffusion equation models, Stochastic Processes and their Appli- cations 12 (3) (1982) 313–326

1982
[32]

Gaspari, S

G. Gaspari, S. E. Cohn, Construction of correlation functions in two and three dimensions, Quarterly Journal of the Royal Meteorological Society 125 (554) (1999) 723–757

1999
[33]

Z0 Zt # =

B. Oksendal, Stochastic differential equations: an introduction with applications, Springer Science & Business Media, 2013. 25 Appendix A. Derivation of Propositions 3.3 and 3.4 In this section, we derive Propositions 3.3 and 3.4. We begin with the following lemma on the distr...

2013

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

G. Evensen, Sequential data assimilation with a nonlinear quasi-geostrophic model using monte carlo methods to forecast error statistics, Journal of Geophysical Research: Oceans 99 (C5) (1994) 10143–10162

1994

[2] [2]

P. L. Houtekamer, H. L. Mitchell, Data assimilation using an ensemble kalman filter technique, Monthly weather review 126 (3) (1998) 796–811

1998

[3] [3]

Shaman, A

J. Shaman, A. Karspeck, Forecasting seasonal outbreaks of influenza, Proceedings of the Na- tional Academy of Sciences 109 (50) (2012) 20425–20430

2012

[4] [4]

R. E. Kalman, A new approach to linear filtering and prediction problems (1960)

1960

[5] [5]

Evensen, Data assimilation: the ensemble Kalman filter, Springer, 2009

G. Evensen, Data assimilation: the ensemble Kalman filter, Springer, 2009

2009

[6] [6]

A. H. Jazwinski, Stochastic processes and filtering theory, Courier Corporation, 2013

2013

[7] [7]

N. J. Gordon, D. J. Salmond, A. F. Smith, Novel approach to nonlinear/non-gaussian bayesian state estimation, in: IEE proceedings F (radar and signal processing), Vol. 140, IET, 1993, pp. 107–113

1993

[8] [8]

Gelb, et al., Applied optimal estimation, MIT press, 1974

A. Gelb, et al., Applied optimal estimation, MIT press, 1974

1974