Precision and Cholesky Factor Estimation for Gaussian Processes

Daniel Sanz-Alonso; Jiaheng Chen

REVIEW 1 major objections 5 minor 1 cited by

Precision and Cholesky Factor Estimation for Gaussian Processes

T0 review · 1 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Estimating a Gaussian process's precision matrix at many observation locations is shown to require only a polylogarithmic number of samples in the matrix size, despite the matrix being ill-conditioned.

desk verdict Polylogarithmic sample complexity for ill-conditioned GP precision estimation is a genuinely new result, and the proof is mostly solid, but the reduction to the lattice theorem in Theorem 2.3 has a repairable gap in the decay constant. read the letter →

arxiv 2412.08820 v2 pith:D56R6TLI submitted 2024-12-11 math.ST cs.NAmath.NAstat.TH

Jiaheng Chen , Daniel Sanz-Alonso This is my paper · ORCID

classification math.STcs.NAmath.NAstat.TH MSC 62G0562M4060G0560G15

keywords precisionestimationCholeskyfactorGaussianprocessesscreeningeffectsamplecomplexitylocalregressionlatticegraphconditionnumber

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 studies the statistical problem of estimating the precision matrix (inverse covariance) of a Gaussian process from independent replicates observed at $M$ scattered locations, along with the Cholesky factor of that matrix. Its central claim is that the sample size needed for a relative-error guarantee grows only poly-logarithmically in $M$ — roughly $(\log M)^d$ instead of the order-$M$ samples that general precision-matrix estimation theory would require — despite the fact that the target matrices are ill-conditioned, with condition numbers growing polynomially in $M$. The mechanism is the screening effect: conditioning on nearby values makes distant values nearly independent, so the dense precision matrix is approximately sparse in a physically local sense, though in dimension $d \ge 2$ it cannot be made banded by any reordering. The main theorems establish high-probability spectral-norm error bounds for precision estimation (Theorems 2.3 and 2.8) and for Cholesky factor estimation under maximin ordering (Theorem 2.7), with an extra logarithmic factor in the Cholesky case. The results matter because they identify a regime in which ill-conditioning is not an obstacle to sample-efficient estimation, with implications for sampling, transport, and spatial statistics.

What carries the argument

The central object is the screening effect — the exponential decay of conditional correlation between two process values given the values at intermediate locations — which converts the dense, ill-conditioned precision matrix into an approximately sparse one. The carrying machinery is a local regression estimator on a $d$-dimensional lattice: the lattice is divided into blocks of side length $b = \lceil \log(N\kappa(\Omega)) \rceil$, and each block of the precision is estimated by inverting the sample covariance restricted to a small neighborhood of that block; the neighborhood size is fixed (a 2-block radius), while the block size grows logarithmically to control the bias term $\kappa(\Omega)e^{-3b}$. Two auxiliary devices lift this lattice estimator to the general setting: Hall's marriage theorem matches homogeneously scattered observation points to nearby lattice points with distortion at most $h$, and a block-Cholesky decomposition of the covariance matrix under the maximin ordering, whose diagonal blocks are uniformly well-conditioned, organizes the Cholesky-factor estimation across scales. The error analysis splits each local error into a statistical term from the sample covariance and a bias term from the exponential decay of precision entries.

What would settle it

Simulate a Matérn process on the unit square using the whole-space covariance (no boundary conditioning), then for points approaching the boundary measure the conditional correlation given intermediate points as the mesh size h decreases. If the maximal such correlation decays like $e^{{-c dist/h}}$ with a constant c that is uniform up to the boundary, the screening assumption behind the polylogarithmic sample-complexity result holds in that setting; if the decay rate worsens or saturates inside a boundary layer, the relative-error bound in Theorem 2.3 would acquire a boundary term and the stated sample complexity would fail for that model.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that for a Gaussian process whose precision operator satisfies Assumption 2.1 (symmetry, positive definiteness, boundedness, and locality in a Sobolev-space setting) and whose observation locations are homogeneously scattered (Assumption 2.2), there exists an estimator of the $M$-by-$M$ precision matrix that, whenever $N \ge C_1 \log^d(N/h)$, achieves with high probability a relative spectral-norm error bounded by $C_3(\log^d(N/h)/N)^{1/2}$. Because the mesh size $h$ satisfies $M \asymp h^{-d}$, this makes the sample complexity polylogarithmic in $M$. The same conclusion extends, under the maximin ordering of observation locations, to the upper-triangular Cholesky factor of the precision, with an additional factor $\log(1/h)$ that the paper shows can be dropped if a modified block Cholesky factor is estimated instead. The proofs establish that the precision matrix belongs to a class of exponentially decaying lattice operators, estimate it blockwise by inverting local sample covariance matrices on blocks of size $\log(N\kappa(\Omega))$, and reduce scattered locations to a regular lattice by a matching argument.

Load-bearing premise

The whole theory assumes that values at nearby observations nearly screen out the influence of distant observations, with the screening strength staying uniform as the grid is refined, and that this holds under a zero Dirichlet boundary condition — which the paper itself notes is unresolved for Matérn-type processes on the whole space observed inside a bounded domain.

Editorial extensions

If this is right

Precision estimation for Gaussian processes with pointwise observations is feasible with $N \gtrsim (\log M)^d$ samples, where $M$ is the number of observation locations, instead of the $N \asymp M$ samples that sample-covariance inversion would require.
The lattice-graph estimator (Theorem 2.8) comes with an explicit dependence on the condition number $\kappa(\Omega)$ and a computational cost of $O(M N \log^{2d}(M N))$, versus $O(M^2 N + M^3)$ for forming and inverting the full sample covariance.
Under the maximin ordering, the upper Cholesky factor $U$ of $\Omega$ can be estimated with $N \gtrsim (\log M)^{d+2}$ samples, with relative error $O(\log(1/h) \sqrt{\log^d(N/h)/N})$; the extra log factor disappears if a modified block Cholesky factor is the target (Remark 4.7).
The Hall's-marriage matching argument reduces scattered-location problems to lattice problems with minimal geometric distortion, a reduction the paper suggests may simplify other analyses involving homogeneously scattered points.
For precision operators with integer smoothness parameter $s > d/2$, including Matérn-type processes with $s = \nu + d/2$, the assumptions cover the standard setting; the authors note extensions to local-average measurements for $s \le d/2$.

Reading between the lines

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

The boundary-condition caveat suggests a concrete prediction: for whole-space Matérn kernels restricted to a bounded observation domain, the local-regression estimator's error should develop a boundary-layer term of order comparable to the number of points near the boundary, so that the polylogarithmic sample complexity holds only after one conditions on boundary values or adds observations outsid
If the exponential decay of conditional correlations also holds for fractional precision operators — evidence the paper cites from numerical experiments and the Caffarelli–Silvestre extension — then polylog sample complexity would carry over to fractional-Laplacian Gaussian fields, which would make efficient estimation and simulation of such fields statistically feasible.
The block-size choice $b = \lceil \log(N\kappa(\Omega)) \rceil$ turns the theory into a practical recipe: choose the conditioning neighborhood in a Vecchia-style approximation by the logarithm of the sample size and the estimated condition number; this could be tested empirically for large spatial datasets as a cross-validated tuning rule.
The reduction via Hall's marriage theorem is likely reusable: any inverse-covariance estimation problem on irregular point clouds whose precision satisfies a local decay property could first be matched to a lattice, estimated there, and then mapped back, so the mismatch between scattered and regular designs may not be the bottleneck.

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

The paper proves polylogarithmic sample complexity for estimating precision matrices and Cholesky factors of Gaussian processes whose precision operators are ill-conditioned but approximately sparse. That is the real result: previous theory would require N roughly linear in the matrix size M, and the screening effect lets them get away with N on the order of (log M)^d. The Hall's marriage reduction from scattered locations to a lattice is a genuinely new technique, and the Cholesky estimation under maximin ordering is the first of its kind. The proofs are long but mostly self-contained: Theorem 2.8 is proved from concentration plus block-decay bounds, and Theorems 2.3 and 2.7 reduce to it with explicit arguments. The regression/Vecchia interpretation in Remark 2.10 is helpful. The citation pattern looks appropriate; the heavy reliance on Owhadi-Scovel and Schafer et al. is legitimate because the paper builds on their structural lemmas rather than restating them as new. There is a genuine gap in the written proof of Theorem 2.3. In display (3.3) the lattice precision is only shown to have row-tail decay of order e^{-ck}, and c can be smaller than 1 because the constant c1 was chosen small to make the Hall matching work. Theorem 2.8 is stated for the class F_{d,p} with tail e^{-k}, so the application of Theorem 2.8 to this matrix is not justified as written. This is not a cosmetic mismatch: the bias term in Proposition 5.1 relies on the e^{-3b} tail, and with a slower tail you need to enlarge the block size by a factor of c^{-1}. The fix is straightforward: restate Theorem 2.8 for a decay constant alpha and take b = O(alpha^{-1} log(N kappa)), and the polylogarithmic rates survive. But the paper should spell this out. The boundary-conditioning limitation is real but explicitly flagged in Remark 2.5; it lies outside Assumption 2.1 and does not affect the internal results. The extension to non-integer s is conjectural and correctly labeled as such. No code or data, but this is a theory paper and the proofs carry the weight. Overall the central argument holds up, modulo the decay-constant fix. This paper is for statisticians working on high-dimensional precision estimation and spatial statistics, and for numerical analysts interested in sparse Cholesky factorizations. It deserves a serious referee rather than a desk reject. I would send it out, with the expectation of a revision that patches the gap in the proof of Theorem 2.3.

Referee Report

1 major / 5 minor

Summary. The paper studies estimation of large precision matrices and upper-triangular Cholesky factors obtained from independent replicates of a Gaussian process observed at many scattered locations. Under Assumptions 2.1 and 2.2, which model the precision operator as a local elliptic operator with zero Dirichlet boundary conditions and the observation points as homogeneously scattered, the main results (Theorems 2.3 and 2.7) assert that the relative spectral-norm error can be driven to zero with a sample size that is polylogarithmic in the number of locations, despite the polynomial growth of the condition number of the target matrices. The proof route is to match scattered locations to a lattice via Hall's theorem, estimate the lattice precision operator by a block-local inversion procedure (Theorem 2.8), and then estimate the Cholesky factor through a hierarchical block-Cholesky decomposition. The paper also gives an explicit computational cost analysis, states a lattice result with an explicit condition-number dependence, and openly discusses the boundary-conditioning limitation in Remark 2.5.

Significance. If correct, these results are significant: they show that ill-conditioned precision matrices with approximate sparsity can be estimated with polylogarithmic sample complexity, in contrast to the N growing linearly with the matrix size that generic precision estimation would suggest, and they provide the first such guarantees under the general operator Assumption 2.1. The lattice result Theorem 2.8 is a useful standalone contribution with explicit dependence on the condition number, and the Hall-matching reduction is an elegant technique. The paper is also careful to discuss computational cost, to connect the local estimator to Vecchia-type regression, and to flag the zero-Dirichlet boundary-conditioning assumption as an open issue for whole-space Matern models. The main caveat is a gap in the reduction from the scattered setting to the stated lattice theorem; it is repairable but, as written, the proof of the principal theorem is incomplete.

major comments (1)

[Section 3, proof of Theorem 2.3; eq. (3.3) and Section 2.3] The proof of Theorem 2.3 reduces the scattered problem to estimating the padded lattice precision matrix \bar{\Omega} by what is called a straightforward application of Theorem 2.8 and Remark 2.9. However, \bar{\Omega} is only shown in (3.3) to satisfy the row-tail bound C\|\bar{\Omega}\|e^{-ck} for k \ge 8\sqrt{d}/c_1, where the constant c may be smaller than 1 because c_1 is chosen small for the Hall-matching argument. The class F_{d,p} defined in Section 2.3 requires the tail \le \|\Omega\|e^{-k}, so \bar{\Omega} need not lie in F_{d,p}. This mismatch is load-bearing: Proposition 5.1's bias term (5.6) is \kappa(\Omega)e^{-3b} with b=\lceil\log(N\kappa(\Omega))\rceil; if the actual tail is C\|\Omega\|e^{-ck}, the bias becomes C^2\kappa(\Omega)e^{-3cb}, and the theorem's choice of b does not make this negligible unless c is bounded below by a universal constant, which is not established. The gap is repairable by stating Theorem 2.8 for a decay class C_0\|\Omega\|e^{-\alpha k} and taking b=O(\alpha^{-1}\log(N\kappa(\Omega))), which preserves the polylogarithmic sample complexity; but as written the proof of the main theorem is incomplete.

minor comments (5)

[Section 2.2, Theorem 2.7] The statement says 'Let \Omega = UU^\top be the upper-triangular Cholesky factorization', but for an upper-triangular factor the standard convention is \Omega = U^\top U. The proof later uses the latter convention, so this is a notation error that should be corrected.
[Remark 2.4] The phrase 'setting apart our work apart from existing high-dimensional results' contains a redundant 'apart'; it should read 'sets our work apart from existing high-dimensional results'.
[Section 5, proof of Theorem 2.8] In the final probability bound, the step from (5.9) to the displayed bound with (\log(N\kappa(\Omega))/p)^{rd} uses S=\lceil p/b\rceil and b=\lceil\log(N\kappa(\Omega))\rceil; the extra factor involving b^{rd} is absorbed into C_2 without comment. This is harmless, but stating the absorption explicitly would improve clarity.
[Section 4, proof of Theorem 2.7] The normalization step \|\hat{U}^\top - U^\top\|/\|U^\top\| \asymp h^{q(s-d/2)}\|\hat{U}^\top - U^\top\| relies on \|U^\top\|=\sqrt{\|\Omega\|} \asymp h^{q(d/2-s)} from Lemma 3.1; this should be stated explicitly at that point for readability.
[Section 1, notation] The paper uses the notation \gtrsim and related symbols inconsistently in the arXiv text, appearing as '>/greaterorsimilar' in several places; the published version should use standard symbols consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main theorems are proved from external decay and concentration results, and the only self-citation is not load-bearing.

full rationale

The derivation chain is self-contained in the sense relevant to circularity. Theorem 2.3 is reduced, via Hall's marriage theorem, to the lattice precision-estimation problem of Theorem 2.8; Theorem 2.8 is proved in Section 5 from the external concentration bound of Koltchinskii and Lounici [30] and the definition of the class F_{d,p}. The screening-effect decay used in the reduction is quoted from Owhadi and Scovel [39, Theorem 9.6], not from the present authors, and the block-Cholesky structure used for Theorem 2.7 is quoted from Schafer, Sullivan, and Owhadi [46]. The estimators are explicitly constructed from local inverse sample covariances; no fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' own prior work. The only reference involving the present authors is [1], which appears in the final discussion of future directions and plays no role in the proofs. The skeptic's concern about the lattice reduction is real but is a proof gap, not circularity: equation (3.3) establishes a tail bound of the form C||Omega||e^{-ck}, while the class F_{d,p} in Section 2.3 requires a tail of the form ||Omega||e^{-k}, and c may be smaller than 1 depending on c1. The paper calls the application of Theorem 2.8 'straightforward' without stating a generalized theorem with an explicit decay parameter, so the written proof of Theorem 2.3 is incomplete at that step. This is a correctness or rigor issue, repairable by restating Theorem 2.8 for a decay class ||Omega||e^{-alpha k}, and it does not make any claimed result equal to its inputs by construction. Remark 2.5 also explicitly flags the boundary-conditioning limitation for Matérn models on bounded domains, but this is an acknowledged assumption boundary rather than a circular step. Overall, no circularity is present.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The central claims rest on the structural theory of operator-adapted wavelets, exponential decay of precision, and block-Cholesky decomposition imported from prior work, plus standard concentration and matching theorems. No fitted parameters or invented entities are introduced.

assumptions (7)

domain assumption Assumption 2.1: precision operator L is symmetric, positive definite, bounded, and local, with zero Dirichlet boundary conditions defining L^{-1}; the process is continuous with s > d/2.
Defines the class of Gaussian processes; the screening effect (Lemma 3.3) and eigenvalue bounds (Lemma 3.1) hold under it. Remark 2.5 notes that the zero Dirichlet boundary conditioning is an open question in practical Matern models.
domain assumption Assumption 2.2: observation locations are homogeneously scattered with fill distance h and separation delta*h.
Required for the volume argument M ~ h^{-d}, the screening decay in Lemma 3.3, and the Hall matching Claim I in the proof of Theorem 2.3.
domain assumption Assumption 2.6: observation locations are sorted by the maximin ordering.
Chosen so that the hierarchy of Example 4.1 holds; the block-Cholesky lemmas from Schafer et al. [46] are stated for this ordering.
domain assumption Screening effect exponential decay (Lemma 3.3, Theorem 9.6 in Owhadi-Scovel [39]): conditional correlations decay as e^{-C^{-1} ||xi - xj|| / h}.
Imported from prior work; it is the source of approximate sparsity that the local regression estimator exploits. The paper does not prove it.
domain assumption Block-Cholesky structure (Lemma 4.2) and uniform bounds on B(k) (Lemma 4.3) from Schafer et al. [46] and Owhadi-Scovel [39].
Imported external theorems; they give the hierarchical scale decomposition used to estimate the Cholesky factor in Theorem 2.7.
standard math Hall's marriage theorem (Lemma 3.4).
Used to prove the existence of a matching between scattered locations and lattice points (Claim I) in the proof of Theorem 2.3.
standard math Concentration inequality for sample covariance matrices (Koltchinskii and Lounici [30]).
Used to bound the local sample covariance inversion error in Proposition 5.1.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Precision and Cholesky Factor Estimation for Gaussian Processes." pith.science (2026). https://pith.science/paper/D56R6TLI

@misc{pith2026241208820,
  author       = {Pith},
  title        = {Pith review of: Precision and Cholesky Factor Estimation for Gaussian Processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D56R6TLI}},
  note         = {Machine review of arXiv:2412.08820}
}

read the original abstract

This paper studies the estimation of large precision matrices and Cholesky factors obtained by observing a Gaussian process at many locations. Under general assumptions on the precision and the observations, we show that the sample complexity scales poly-logarithmically with the size of the precision matrix and its Cholesky factor. The key challenge in these estimation tasks is the polynomial growth of the condition number of the target matrices with their size. For precision estimation, our theory hinges on an intuitive local regression technique on the lattice graph which exploits the approximate sparsity implied by the screening effect. For Cholesky factor estimation, we leverage a block-Cholesky decomposition recently used to establish complexity bounds for sparse Cholesky factorization.

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

Generative multi-scale modeling and downscaling via spatial autoregressive transport maps
stat.ME 2025-09 conditional novelty 6.0 of 10

A new multi-fidelity Bayesian transport map method learns non-Gaussian joint distributions across spatial scales and outperforms existing emulators in downscaling climate fields from small training sets.

Reference graph

Works this paper leans on

52 extracted references · 50 canonical work pages · cited by 1 Pith paper

[1]

Al-Ghattas, J

O. Al-Ghattas, J. Chen, D. Sanz-Alonso, and N. W aniorek , Optimal estimation of structured covariance operators, arXiv preprint arXiv:2408.02109, (2024)

arXiv 2024
[2]

P. J. Bickel and E. Levina , Covariance regularization by thresholding , The Annals of Statistics, 36 (2008), pp. 2577–2604

work page 2008
[3]

P. J. Bickel and E. Levina , Regularized estimation of large covariance matrices , The Annals of Sta- tistics, 36 (2008), pp. 199–227

work page 2008
[4]

D. L. Brown, J. Gedicke, and D. Peterseim , Numerical homogenization of heterogeneous fractional Laplacians, Multiscale Modeling & Simulation, 16 (2018), pp. 1305–133 2

work page 2018
[5]

Cai and W

T. Cai and W. Liu , Adaptive thresholding for sparse covariance matrix estima tion, Journal of the American Statistical Association, 106 (2011), pp. 672–684

work page 2011
[6]

T. Cai, W. Liu, and X. Luo , A constrained ℓ1 minimization approach to sparse precision matrix esti- mation, Journal of the American Statistical Association, 106 (201 1), pp. 594–607

work page
[7]

T. T. Cai, W. Liu, and H. H. Zhou , Estimating sparse precision matrix: Optimal rates of conve rgence and adaptive estimation , The Annals of Statistics, (2016), pp. 455–488

work page 2016
[8]

T. T. Cai, Z. Ren, and H. H. Zhou , Estimating structured high-dimensional covariance and pr ecision matrices: Optimal rates and adaptive estimation , Electronic Journal of Statistics, 10 (2016), pp. 1–59

work page 2016

Show all 52 references

[9]

T. T. Cai and M. Yuan , Minimax and adaptive estimation of covariance operator for random variables observed on a lattice graph , Journal of the American Statistical Association, 111 (201 6), pp. 253–265
[10]

T. T. Cai, C.-H. Zhang, and H. H. Zhou , Optimal rates of convergence for covariance matrix estima- tion, The Annals of Statistics, 38 (2010), pp. 2118–2144

2010
[11]

Y. Chen, H. Owhadi, and F. Sch ¨afer, Sparse Cholesky factorization for solving nonlinear PDEs v ia Gaussian processes, Mathematics of Computation, (2024)

2024
[12]

Cressie , Statistics for Spatial Data , John Wiley & Sons, 2015

N. Cressie , Statistics for Spatial Data , John Wiley & Sons, 2015

2015
[13]

Drma ˇc, M

Z. Drma ˇc, M. Omladi ˇc, and K. Veseli ´c, On the perturbation of the Cholesky factorization , SIAM Journal on Matrix Analysis and Applications, 15 (1994), pp. 1319–1332

1994
[14]

Edelman and W

A. Edelman and W. F. Mascarenhas , On Parlett’s matrix norm inequality for the Cholesky decom- position, Numerical Linear Algebra with Applications, 2 (1995), pp. 243–250

1995
[15]

Friedman, T

J. Friedman, T. Hastie, and R. Tibshirani , Sparse inverse covariance estimation with the graphical lasso, Biostatistics, 9 (2008), pp. 432–441

2008
[16]

R. B. Gramacy and D. W. Apley , Local Gaussian process approximation for large computer ex peri- ments, Journal of Computational and Graphical Statistics, 24 (20 15), pp. 561–578
[17]

R. B. Gramacy and B. Haaland , Speeding up neighborhood search in local Gaussian process p rediction, Technometrics, 58 (2016), pp. 294–303

2016
[18]

Guinness , Permutation and grouping methods for sharpening Gaussian p rocess approximations, Tech- nometrics, 60 (2018), pp

J. Guinness , Permutation and grouping methods for sharpening Gaussian p rocess approximations, Tech- nometrics, 60 (2018), pp. 415–429

2018
[19]

Hall , On representatives of subsets , Classic Papers in Combinatorics, (1987), pp

P. Hall , On representatives of subsets , Classic Papers in Combinatorics, (1987), pp. 58–62

1987
[20]

Harbrecht, L

H. Harbrecht, L. Herrmann, K. Kirchner, and C. Schw ab , Multilevel approximation of Gauss- ian random ﬁelds: Covariance compression, estimation and s patial prediction , arXiv preprint arXiv:2103.04424, (2021)

2021 arXiv
[21]

Herrmann, K

L. Herrmann, K. Kirchner, and C. Schw ab , Multilevel approximation of Gaussian random ﬁelds: fast simulation , Mathematical Models and Methods in Applied Sciences, 30 (2 020), pp. 181–223. PRECISION AND CHOLESKY FACTOR ESTIMATION 29
[22]

T. Y. Hou and P. Zhang , Sparse operator compression of higher-order elliptic oper ators with rough coeﬃcients , Research in the Mathematical Sciences, 4 (2017), pp. 1–49

2017
[23]

Hu and S

A. Hu and S. Negahban , Minimax estimation of bandable precision matrices , Advances in Neural Information Processing Systems, 30 (2017)

2017
[24]

M. Kang, F. Sch ¨afer, J. Guinness, and M. Katzfuss , Asymptotic properties of Vecchia approxima- tion for Gaussian processes , arXiv preprint arXiv:2401.15813, (2024)

2024 arXiv
[25]

Katzfuss and J

M. Katzfuss and J. Guinness , A general framework for Vecchia approximations of Gaussian processes, Statistical Science, 36 (2021), pp. 124–141

2021
[26]

Katzfuss and F

M. Katzfuss and F. Sch ¨afer, Scalable Bayesian transport maps for high-dimensional non -Gaussian spatial ﬁelds , Journal of the American Statistical Association, 119 (202 4), pp. 1409–1423
[27]

Kelner, F

J. Kelner, F. Koehler, R. Meka, and A. Moitra , Learning some popular Gaussian graphical models without condition number bounds , Advances in Neural Information Processing Systems, 33 (20 20), pp. 10986–10998
[28]

J. A. Kelner, F. Koehler, R. Meka, and D. Rohatgi , On the power of preconditioning in sparse lin- ear regression, in 2021 IEEE 62nd Annual Symposium on Foundations of Comput er Science (FOCS), IEEE, 2022, pp. 550–561

2021
[29]

Kidd and M

B. Kidd and M. Katzfuss , Bayesian nonstationary and nonparametric covariance esti mation for large spatial data (with discussion) , Bayesian Analysis, 17 (2022), pp. 291–351

2022
[30]

Koltchinskii and K

V. Koltchinskii and K. Lounici , Concentration inequalities and moment bounds for sample co variance operators, Bernoulli, 23 (2017), pp. 110–133

2017
[31]

Lindgren, H

F. Lindgren, H. Rue, and J. Lindstr ¨om, An explicit link between Gaussian ﬁelds and Gaussian Markov random ﬁelds: the stochastic partial diﬀerential equation a pproach, Journal of the Royal Statistical Society: Series B (Statistical Methodology), 73 (2011), pp . 423–498

2011
[32]

Liu and Z

Y. Liu and Z. Ren , Minimax estimation of large precision matrices with bandab le Cholesky factor , The Annals of Statistics, 48 (2020)

2020
[33]

Lu and S.-H

T.-T. Lu and S.-H. Shiou , Inverses of 2 × 2 block matrices , Computers & Mathematics with Applica- tions, 43 (2002), pp. 119–129

2002
[34]

M ˚alqvist and D

A. M ˚alqvist and D. Peterseim , Localization of elliptic multiscale problems , Mathematics of Compu- tation, 83 (2014), pp. 2583–2603

2014
[35]

Marzouk, T

Y. Marzouk, T. Moselhy, M. P arno, and A. Spantini , Sampling via measure transport: An intro- duction, Handbook of Uncertainty Quantiﬁcation, 1 (2016), pp. 1–41

2016
[36]

Misra, M

S. Misra, M. Vuffray, and A. Y. Lokhov , Information theoretic optimal learning of gaussian graphi cal models, in Conference on Learning Theory, PMLR, 2020, pp. 2888–290 9

2020
[37]

Owhadi, Multigrid with rough coeﬃcients and multiresolution opera tor decomposition from hierarchical information games , SIAM Review, 59 (2017), pp

H. Owhadi, Multigrid with rough coeﬃcients and multiresolution opera tor decomposition from hierarchical information games , SIAM Review, 59 (2017), pp. 99–149

2017
[38]

Owhadi and C

H. Owhadi and C. Scovel , Universal scalable robust solvers from computational info rmation games and fast eigenspace adapted multiresolution analysis , arXiv preprint arXiv:1703.10761, (2017)

2017 arXiv
[39]

Owhadi and C

H. Owhadi and C. Scovel , Operator-Adapted Wavelets, Fast Solvers, and Numerical Ho mogenization: From a Game Theoretic Approach to Numerical Approximation a nd Algorithm Design , vol. 35, Cam- bridge University Press, 2019

2019
[40]

Owhadi, C

H. Owhadi, C. Scovel, and F. Sch ¨afer, Statistical numerical approximation , Notices of the AMS, 66 (2019)

2019
[41]

Owhadi, L

H. Owhadi, L. Zhang, and L. Berlyand , Polyharmonic homogenization, rough polyharmonic splines and sparse super-localization , ESAIM: Mathematical Modelling and Numerical Analysis, 48 (2014), pp. 517–552

2014
[42]

Z. Ren, T. Sun, C.-H. Zhang, and H. H. Zhou , Asymptotic normality and optimalities in estimation of large Gaussian graphical models , The Annals of Statistics, (2015), pp. 991–1026

2015
[43]

Rue and L

H. Rue and L. Held , Gaussian Markov random ﬁelds: theory and applications , CRC Press, 2005

2005
[44]

B. A. Schmitt , Perturbation bounds for matrix square roots and Pythagorea n sums , Linear Algebra and its Applications, 174 (1992), pp. 215–227

1992
[45]

Sch ¨afer, M

F. Sch ¨afer, M. Katzfuss, and H. Owhadi , Sparse Cholesky factorization by Kullback–Leibler mini- mization, SIAM Journal on Scientiﬁc Computing, 43 (2021), pp. A2019– A2046

2021
[46]

Sch ¨afer, T

F. Sch ¨afer, T. J. Sullivan, and H. Owhadi , Compression, inversion, and approximate PCA of dense kernel matrices at near-linear computational complexity , Multiscale Modeling & Simulation, 19 (2021), 30 J.CHEN AND D.SANZ-ALONSO pp. 688–730

2021
[47]

M. L. Stein , The screening eﬀect in kriging , The Annals of Statistics, 30 (2002), pp. 298–323

2002
[48]

M. L. Stein , 2010 Rietz lecture: When does the screening eﬀect hold? , The Annals of Statistics, 39 (2011), pp. 2795 – 2819

2011
[49]

M. L. Stein , Interpolation of Spatial Data: Some Theory for Kriging , Springer, 2012

2012
[50]

A. V. Vecchia , Estimation and model identiﬁcation for continuous spatial processes, Journal of the Royal Statistical Society Series B: Statistical Methodology, 50 (1988), pp. 297–312

1988
[51]

Whittle , On stationary processes in the plane , Biometrika, (1954), pp

P. Whittle , On stationary processes in the plane , Biometrika, (1954), pp. 434–449

1954
[52]

normalized

C. K. I. Williams and C. E. Rasmussen , Gaussian Processes for Machine Learning , vol. 2, MIT Press Cambridge, MA, 2006. Appendix A. Proofs in Section 4. Proof of Lemma 4.5. Note that ~εB := ‖(ˆB)−1 − B−1‖ ‖B−1‖ = ‖(ˆB)−1(ˆB − B)B−1‖ ‖B−1‖ ≤ ‖(ˆB)−1‖‖ˆB − B‖ ≤ ‖ (ˆB)−1‖‖B‖εB, ...

2006

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Al-Ghattas, J

O. Al-Ghattas, J. Chen, D. Sanz-Alonso, and N. W aniorek , Optimal estimation of structured covariance operators, arXiv preprint arXiv:2408.02109, (2024)

arXiv 2024

[2] [2]

P. J. Bickel and E. Levina , Covariance regularization by thresholding , The Annals of Statistics, 36 (2008), pp. 2577–2604

work page 2008

[3] [3]

P. J. Bickel and E. Levina , Regularized estimation of large covariance matrices , The Annals of Sta- tistics, 36 (2008), pp. 199–227

work page 2008

[4] [4]

D. L. Brown, J. Gedicke, and D. Peterseim , Numerical homogenization of heterogeneous fractional Laplacians, Multiscale Modeling & Simulation, 16 (2018), pp. 1305–133 2

work page 2018

[5] [5]

Cai and W

T. Cai and W. Liu , Adaptive thresholding for sparse covariance matrix estima tion, Journal of the American Statistical Association, 106 (2011), pp. 672–684

work page 2011

[6] [6]

T. Cai, W. Liu, and X. Luo , A constrained ℓ1 minimization approach to sparse precision matrix esti- mation, Journal of the American Statistical Association, 106 (201 1), pp. 594–607

work page

[7] [7]

T. T. Cai, W. Liu, and H. H. Zhou , Estimating sparse precision matrix: Optimal rates of conve rgence and adaptive estimation , The Annals of Statistics, (2016), pp. 455–488

work page 2016

[8] [8]

T. T. Cai, Z. Ren, and H. H. Zhou , Estimating structured high-dimensional covariance and pr ecision matrices: Optimal rates and adaptive estimation , Electronic Journal of Statistics, 10 (2016), pp. 1–59

work page 2016