A Non-asymptotic Analysis for Learning and Applying a Preconditioner in MCMC

Florian Maire; Jeffrey Negrea; Max Hird

REVIEW 3 major objections 5 minor 1 cited by

A Non-asymptotic Analysis for Learning and Applying a Preconditioner in MCMC

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read This paper shows that a Markov-chain sampler that learns its own linear preconditioner can have lower total computational cost than the same sampler without preconditioning, provided enough samples are drawn and the target is amenable to pr

desk verdict Useful AIID framework, but Theorem 5's FLOPS bounds omit a d, a 1/m_M, and the pushforward rescaling factor — the quantitative comparison as stated is not derived. read the letter →

arxiv 2602.10714 v2 pith:FR6NGGGK submitted 2026-02-11 stat.CO stat.ML

Max Hird , Florian Maire , Jeffrey Negrea This is my paper · ORCID

classification stat.COstat.ML MSC 65C0560J22

keywords MCMCpreconditioningadaptivenon-asymptoticanalysisWassersteincontractionunadjustedLangevinalgorithmcovariancepreconditionerFisher

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 asks whether an MCMC algorithm that learns its own linear preconditioner from the chain can beat its unpreconditioned counterpart in total floating-point cost, including the cost of learning. The answer, for the unadjusted Langevin algorithm on smooth, strongly convex targets, is yes: a preconditioner based on the target covariance or the Fisher matrix reduces the condition number and, once enough samples are needed, the one-time learning cost is amortised. To make this comparison rigorous, the authors introduce a √Nε-approximately IID condition that measures how far a thinned chain's output is from an IID sample in Wasserstein-2 distance, letting them split the total cost into a learning phase and a sampling phase. A central finding is that the preconditioner needs only be estimated to moderate constant accuracy (relative error 1/2) to retain essentially all of its benefit.

What carries the argument

The engine is the (Γ,γ,b)-W₂ contraction condition (Definition 2): a kernel K satisfies W₂(π, μK^k) ≤ Γe^{-γk}W₂(π,μ)+b, where b≥0 captures bias. This single condition lets the authors derive Theorem 3, which bounds the total number of iterations needed to output √Nε-AIID samples, and Theorem 4, which bounds the cost of estimating Σ_π or F from such an ensemble. Corollary 14 then shows that if the estimate satisfies ‖Σ^{-1/2}_π Σ̂_π Σ^{-1/2}_π - I‖ ≤ Δ, the resulting condition number is at most (1+Δ)/(1-Δ) times the ideal one, so Δ=1/2 suffices. These pieces assemble into the FLOPS totals in Theorem 5.

What would settle it

Run covariance-preconditioned ULA on a family of anisotropic Gaussian targets with known condition number κ, count the actual FLOPS to reach √Nε-AIID for a range of N, and compare with plain ULA; if the measured break-even N diverges from the amortization prediction of Theorem 5, or if preconditioning is never faster on any N for a target with κ_{Σ^{-1}_π}=κ, the central comparison is wrong for that class.

Watch

Extended reading notes

Core claim

The central claim, stated in Theorem 5, is an explicit total-FLOPS comparison for ULA: unpreconditioned ULA needs Õ(m^{-1}(d+G)κ²Nε^{-2}) FLOPS to produce N samples that are √Nε-approximately IID, while covariance-preconditioned ULA needs Õ(δ^{-2}d³(d+G)κ³max{δ^{-1},K³_cov} + (d²+G)κ²_{Σ^{-1}_π}Nε^{-2}) FLOPS with probability 1-δ (similarly for the Fisher preconditioner). The two terms are the cost to learn the preconditioner and the cost to generate the final ensemble; comparing them shows preconditioning wins when N is large enough to make the second term dominate and when the preconditioned condition number κ_{Σ^{-1}_π} is substantially smaller than κ. The authors also prove that learning

Load-bearing premise

The load-bearing premise is that the target is smooth and strongly convex and that the learned preconditioner lands within a constant relative error (Δ=1/2) of the true covariance or Fisher matrix; if either fails, or if the target is not actually preconditionable (the preconditioned condition number is close to the original), the bounds do not guarantee any benefit over plain sampling.

Editorial extensions

If this is right

When the target is linearly preconditionable (κ_{Σ^{-1}_π} or κ_F much smaller than κ) and N is large, preconditioned ULA has lower total FLOPS than unpreconditioned ULA.
Learning the preconditioner costs Õ(d³(d+G)κ³δ^{-2}max{δ^{-1},K³_cov}) FLOPS, independent of N; this is a one-time expense that can be amortised over the output sample size.
A constant-accuracy estimate (Δ=1/2) of the covariance or Fisher matrix suffices; the condition number of the preconditioned chain grows by at most a factor of 3.
The √Nε-AIID framework gives non-asymptotic guarantees for both exact and biased (b>0) kernels, covering ULA and, in the supplement, the unadjusted underdamped Langevin algorithm.
The per-sample FLOPS after preconditioning is (d²+G)κ²_M Nε^{-2}, so the break-even point depends concretely on the dimension d and the gradient cost G.

Reading between the lines

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

A practical consequence the authors leave implicit: because Δ=1/2 suffices, a short adaptation run that gives a rough covariance estimate should already deliver most of the sampling speedup; one need not wait for precise adaptation.
The FLOPS comparison suggests preconditioning is most valuable when gradient evaluations are expensive (large G, e.g., inverse problems); for very large d, the added d² per-iteration matrix-vector product may outweigh the condition-number reduction unless κ/κ_M is large. This is a testable boundary worth probing numerically.
The √Nε-AIID condition is a general-purpose objective: it could be applied to other adaptive MCMC parameters (step size, HMC integration time) or to periodic re-adaptation, as the paper itself notes as future work.

Share X Bluesky LinkedIn Reddit HN

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

Colleague,

Two things to know up front. The √Nε-AIID condition and the two-phase cost decomposition are genuinely useful tools, and the paper is careful in most of its proofs. But the central quantitative claim in Theorem 5 does not follow from the proof: the displayed FLOPS bounds are missing at least three factors — a d, a 1/m_M, and the Lipschitz constant of the inverse preconditioning map.

The paper's new idea is to measure MCMC output by how close a whole ensemble is to an IID sample in W2 (the AIID condition). That bridges the gap between mixing time and effective sample size, and it behaves well under Lipschitz maps (Prop 7). The two-phase analysis — one-time learning cost plus per-sample cost — is exactly the right way to think about amortizing a preconditioner. The claim that a moderately accurate estimate (Δ=0.5) suffices is a useful insight, and the learning-phase analysis is detailed and mostly sound.

The problem is Theorem 5. In the unpreconditioned case the step size is h = c d^{-1} κ^{-2} ε^2, so γ = m h and the iteration count carries d/(m ε^2). Eq (11) omits the d. The preconditioned cases inherit this and add 1/m_M from the transformed target; the sampling-stage terms in Eq (13) and (15) omit both. More seriously, when Algorithm 2 runs the chain on the transformed target π̃ and outputs X_t = M^{-1/2} X̃_t, the AIID tolerance must be rescaled by ||M^{-1/2}|| (Prop 7). The proof verifies the ε conditions for the transformed chain using the same ε as the theorem, not ε/||M^{-1/2}||. Correcting this adds λ_max(Σ̂_π) in the covariance case and λ_min(F̂)^{-1} in the Fisher case — factors that can be arbitrarily large under the theorem's stated conditions. So the quantitative FLOPS comparison, which is the paper's headline result, is not derived as stated. The qualitative amortization story may survive a correction, but the actual rates could differ by problem-dependent constants that affect the crossover regime.

Smaller point: the abstract promises a proximal-sampler analysis that does not appear in the body. That is a quick fix.

Who should read this? Anyone working on adaptive MCMC or non-asymptotic sampling theory. The framework is worth engaging with, but the paper needs a major revision before its main theorem can be trusted. I would send it to peer review — the ideas deserve referee time — but the referee should require a corrected Theorem 5 that tracks the transformation of ε and the step-size constants carefully.

Referee Report

3 major / 5 minor

Summary. The paper proposes a non-asymptotic framework for comparing the total computational cost of Markov chain Monte Carlo algorithms that learn and apply a linear preconditioner with those that do not. It introduces a √Nε-approximate IID (AIID) condition based on Wasserstein-2 distance to the N-fold product target and proves a generic iteration-complexity bound (Theorem 3) for thinned samplers whose kernel satisfies a (Γ,γ,b)-W2 contraction. It then provides high-probability learning complexities for covariance-based and Fisher-information-based preconditioners (Theorem 4), and applies the results to ULA to obtain total FLOPS bounds (Theorem 5), with an extension to underdamped Langevin in the supplement. The central affirmative claim is that, for targets that are linearly preconditionable and when many samples are required, the one-time cost of learning the preconditioner is amortized and preconditioned ULA can have lower total complexity than unpreconditioned ULA.

Significance. The framework and the AIID condition are genuinely useful: they connect mixing-time bounds with effective-sample-size intuition, and the modular structure (generic contraction bound, learning bound, sampler-specific application) is elegant. The paper contains detailed proofs of several intermediate propositions and makes the constants in the learning bounds fairly explicit. If Theorem 5 were established as stated, it would fill a notable gap in the non-asymptotic theory of adaptive MCMC. However, the central quantitative comparison is not currently derived from the proof: the sampling stage of Algorithm 2 ignores the Lipschitz constant of the inverse preconditioning map, and the displayed FLOPS omit factors that follow from the step-size choice. These issues directly affect the claimed comparison, so the affirmative answer to the motivating question is not yet supported.

major comments (3)

[Sections B.5.2, B.6.2, Eq. (13), Eq. (15), Proposition 7] The proof of Theorem 5, parts 2 and 3, does not convert the AIID guarantee from the transformed target back to π. Algorithm 2 outputs X_t = M^{-1/2} X̃_t, with X̃_t produced from π̃ = M^{1/2}_♯π. By Proposition 7, if {X̃_t} is √N ε'-AIID from π̃, then {M^{-1/2} X̃_t} is only √N ||M^{-1/2}|| ε'-AIID from π. The proofs in B.5.2 and B.6.2 verify the ε-conditions for π̃ using the same ε as in the theorem statement and choose the step size h = ε²/(100 d κ_M²) from that ε; they never apply Proposition 7 with ε' = ε/||M^{-1/2}||. Consequently the computed sampling-stage FLOPS are those needed to obtain a √Nε-AIID sample from π̃, not from π. Correcting this adds a factor ||M^{-1/2}||² — i.e. 1/λ_min(Σ̂π) for the covariance preconditioner and 1/λ_min(F̂) for the Fisher preconditioner — to the sampling term. This factor is not controlled by the theorem's assumptions: for π = N(0, σ² I_d), κ = 1 bu
[Section 4.3.1, Section B.4, Eq. (11)] The step-size choice made in the proof, h = 100^{-1} d^{-1} κ^{-2} ε², gives γ = mh and hence γ^{-1} = 100 d m^{-1} κ² ε^{-2}. Since the per-iteration cost is (d+G), the iteration bound from Theorem 3 yields an additional factor d in the total FLOPS bound. The displayed unpreconditioned bound in Eq. (11) is Õ(m^{-1}(d+G)κ² N ε^{-2}), with no d. Similarly, the sampling-stage terms in Eq. (13) and Eq. (15) omit the d from γ^{-1} and the m_M^{-1} (equivalently κ_M/L_M) factor in the contraction rate. These factors are polynomial in d and κ, not logarithmic constants, so they cannot be absorbed into Õ without saying so. As written, the displayed FLOPS expressions in Theorem 5 do not follow from the proof.
[Section B.3.2, Eq. (87)–(88)] Theorem 4, part 2, and hence the Fisher-preconditioned FLOPS in Theorem 5, part 3, rely on the bound that F^{-1/2}∇logπ(Z) is sub-Gaussian with proxy sup_x λ_1(F^{-1/2}∇²U(x)F^{-1/2}), imported from Negrea [2022, Theorem 2.2]. This is a self-citation to a thesis and is load-bearing: it is what produces K_Fisher in Eq. (88). The manuscript gives only a one-sentence summary of the result. Please include the precise statement and a proof (or a clearly labelled reference with the theorem reproduced) in the appendix, so this step can be verified independently.

minor comments (5)

[Section 1] 'one its un-preconditioned counterpart' should be 'one of its unpreconditioned counterparts'.
[Section 1] 'twice differential' should be 'twice differentiable'.
[Section 5] 'See 5 for further discussion' should be 'See Section 5'.
[Algorithm 2] In step 3, the notation K(M^{1/2}_♯π) is confusing because the pushforward is of the target, not of the kernel; a brief explanation of the transformation would help.
[Eq. (21) and elsewhere] C is introduced as an absolute constant but its value changes from line to line; state 'C denotes a universal constant, possibly changing on each use'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the complexity bounds are derived from contraction and concentration arguments, with an independent self-citation that does not assume the target result.

full rationale

I walked the derivation chain from Definition 2 and Theorem 3 through the ULA-specific bounds in Theorem 5. The central quantities are derived rather than fitted: Theorem 3 converts a (Γ,γ,b)-W2 contraction into an iteration bound for a √Nε-AIID output; Theorem 4 obtains preconditioner-learning costs from AIID concentration and Vershynin's empirical-covariance bound; Theorem 5 substitutes the known ULA contraction constants from Dalalyan and Karagulyan and the condition-number control of Corollary 14, which is proved from Ostrowski's theorem rather than assumed. The 'linearly preconditionable' conditions on κ_{Σ^{-1}_π} or κ_F are explicit assumptions, not conclusions smuggled into the derivation. The only author-overlapping citation that carries weight is Negrea [2022, Theorem 2.2], used to bound the sub-Gaussian norm of F^{-1/2}∇logπ(Z). This is a parameter-free lemma whose stated assumption (L-smoothness) does not include the present paper's target result, so under Rule 4 it counts as independent support rather than a circular premise. No fitted parameter is renamed as a prediction, and no uniqueness theorem is imported to force the authors' choice. Possible technical concerns about missing Lipschitz factors in the final FLOPS comparison are correctness issues, not circularity.

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

The central bounds rest on the strong-convexity/smoothness regime, the W2-contraction constants for ULA imported from Dalalyan-Karagulyan, and two concentration lemmas (Vershynin, Negrea). No empirical free parameters are fit; the user-chosen Δ=1/2 and δ only set the accuracy/probability of preconditioner learning.

assumptions (7)

domain assumption U is m-strongly convex and L-smooth: mI ⪯ ∇²U ⪯ LI.
Stated at start of Section 1 and used in Theorem 5 to invoke ULA contraction bounds.
standard math ULA kernel is a (1, mh, (33/20)κ√(dh))-W2 contraction when h≤2/(L+m).
Imported from Dalalyan and Karagulyan [2019, Thm 1a]; used in proofs of Theorem 5 and in learning-phase step-size choices.
domain assumption Preconditioned potential remains in a W2-contraction regime with constants L_M, m_M defined in Eq. (3), and Corollary 14 bounds κ_{\hat M} by 3κ_M with probability 1−δ.
The two-phase argument in Theorem 5 requires the preconditioned ULA to contract with the new constants; if the estimate fails the relative-error bound this step fails.
domain assumption β_{Σπ} I ⪯ Σπ and α_F I ⪯ F.
Theorem 4 requires lower bounds on the covariance and Fisher matrix to control Lipschitz constants and sub-Gaussian norms; the paper notes β≥L^{-1} under smoothness, α_F≥m under strong convexity.
standard math [Negrea 2022, Thm 2.2]: the score F^{-1/2}∇logπ(Z) of an L-smooth potential is sub-Gaussian with proxy L_F.
Used in B.3.2 to bound K_Fisher; self-cited from a co-author's thesis.
standard math Vershynin, Exercise 4.7.3: empirical covariance concentration for sub-Gaussian samples.
Used in B.2.2 and B.3.2 to bound the IID-sample term in preconditioner estimation.
domain assumption The target is linearly preconditionable: κ_{Σ^{-1}_π} or κ_F can be much smaller than κ.
Theorems 5 part 2/3 state rates in terms of these smaller constants; the benefit claim depends on this being true, which is not proved for general targets.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Non-asymptotic Analysis for Learning and Applying a Preconditioner in MCMC." pith.science (2026). https://pith.science/paper/FR6NGGGK

@misc{pith2026260210714,
  author       = {Pith},
  title        = {Pith review of: A Non-asymptotic Analysis for Learning and Applying a Preconditioner in MCMC},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FR6NGGGK}},
  note         = {Machine review of arXiv:2602.10714}
}

abstract

Preconditioning is a common method applied to modify Markov chain Monte Carlo algorithms with the goal of making them more efficient. In practice it is often extremely effective, even when the preconditioner is learned from the chain. We analyse and compare the finite-time computational costs of schemes which learn a preconditioner based on the target covariance or the expected Hessian of the target potential with that of a corresponding scheme that does not use preconditioning. We apply our results to various algorithms including the Unadjusted Langevin Algorithm (ULA) and the proximal sampler for an appropriately regular target, establishing non-asymptotic guarantees for versions of these algorithms that learn and use preconditioners. To do so, we establish non-asymptotic guarantees on the time taken to collect $N$ approximately independent samples from the target for schemes that learn their preconditioners under the assumption that the underlying Markov chain satisfies a contraction condition in the Wasserstein-2 distance. This approximate independence condition, that we formalize, allows us to bridge the non-asymptotic bounds of modern MCMC theory and classical heuristics of effective sample size and mixing time, and is needed to amortise the costs of learning a preconditioner across the many samples it will be used to produce.

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

Error bounds for simultaneous Wasserstein contractive adaptive increasingly rare MCMC
math.ST 2026-06 unverdicted novelty 6.0 of 10

Explicit MSE bounds derived for time-average estimators in adaptive increasingly rare MCMC under simultaneous Wasserstein contraction.

Reference graph

Works this paper leans on

26 extracted references · cited by 1 Pith paper

[1]

Christophe Andrieu, Anthony Lee, Sam Power, and Andi Q. Wang. Explicit convergence bounds for Metropolis Markov chains: Isoperimetry , spectral gaps and profiles. The Annals of Applied Probability, 34 0 (4): 0 4022--4071, August 2024. ISSN 1050-5164, 2168-8737. doi:10.1214/24-AAP2058

doi:10.1214/24-aap2058 2024
[2]

Unadjusted Hamiltonian MCMC with stratified Monte Carlo time integration

Nawaf Bou-Rabee and Milo Marsden. Unadjusted Hamiltonian MCMC with stratified Monte Carlo time integration. The Annals of Applied Probability, 35 0 (1): 0 360--392, February 2025. ISSN 1050-5164, 2168-8737

2025
[3]

Polar factorization and monotone rearrangement of vector-valued functions

Yann Brenier. Polar factorization and monotone rearrangement of vector-valued functions. Communications on Pure and Applied Mathematics, 44 0 (4): 0 375--417, 1991

1991
[4]

Carlier, A

G. Carlier, A. Galichon, and F. Santambrogio. From Knothe 's transport to Brenier 's map and a continuation method for optimal transport. SIAM Journal on Mathematical Analysis, 41 0 (6): 0 2554--2576, January 2010

2010
[5]

Hoffman, Daniel Lee, Ben Goodrich, Michael Betancourt, Marcus Brubaker, Jiqiang Guo, Peter Li, and Allen Riddell

Bob Carpenter, Andrew Gelman, Matthew D. Hoffman, Daniel Lee, Ben Goodrich, Michael Betancourt, Marcus Brubaker, Jiqiang Guo, Peter Li, and Allen Riddell. Stan: A Probabilistic Programming Language . Journal of Statistical Software, 76: 0 1--32, January 2017. ISSN 1548-7660

2017
[6]

Chatterji, Peter L

Xiang Cheng, Niladri S. Chatterji, Peter L. Bartlett, and Michael I. Jordan. Underdamped Langevin MCMC : A non-asymptotic analysis. In Proceedings of the 31st Conference On Learning Theory , pages 300--323. PMLR, July 2018

2018
[7]

Log-concave sampling

Sinho Chewi. Log-concave sampling. Book draft: https://chewisinho. github. io, 2023

2023
[8]

Air Markov Chain Monte Carlo , January 2018

Cyril Chimisov, Krzysztof Latuszynski, and Gareth Roberts. Air Markov Chain Monte Carlo , January 2018

2018

Show all 26 references

[9]

Dalalyan and Avetik G

Arnak S. Dalalyan and Avetik G. Karagulyan. User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient. Stochastic Processes and their Applications, 129 0 (12): 0 5278--5311, December 2019. ISSN 03044149. doi:10.1016/j.spa.2019.02.016

2019 doi
[10]

Analysis of langevin monte carlo via convex optimization

Alain Durmus, Szymon Majewski, and B a \.z ej Miasojedow. Analysis of langevin monte carlo via convex optimization. Journal of Machine Learning Research, 20 0 (73): 0 1--46, 2019

2019
[11]

An Adaptive Metropolis Algorithm

Heikki Haario, Eero Saksman, and Johanna Tamminen. An Adaptive Metropolis Algorithm . Bernoulli, 7 0 (2): 0 223--242, 2001. ISSN 1350-7265. doi:10.2307/3318737

2001 doi
[12]

Beyond Canonical MCMC : Preconditioning , Adaptivity , and Variational Approximations

Max Harwood Hird. Beyond Canonical MCMC : Preconditioning , Adaptivity , and Variational Approximations . PhD thesis, UCL (University College London), 2025

2025
[13]

Roberts, and Daniel Rudolf

Julian Hofstadler, Krzysztof Latuszynski, Gareth O. Roberts, and Daniel Rudolf. Almost sure convergence rates of adaptive increasingly rare Markov chain Monte Carlo , October 2025

2025
[14]

Yunbum Kook and Matthew S. Zhang. Covariance estimation using Markov chain Monte Carlo , October 2024

2024
[15]

Concentration of measure and logarithmic Sobolev inequalities

Michel Ledoux. Concentration of measure and logarithmic Sobolev inequalities. In Jacques Az \'e ma, Michel \'E mery, Michel Ledoux, and Marc Yor, editors, S\'eminaire de Probabilit\'es XXXIII , pages 120--216. Springer, 1999

1999
[16]

Structured Logconcave Sampling with a Restricted Gaussian Oracle

Yin Tat Lee, Ruoqi Shen, and Kevin Tian. Structured Logconcave Sampling with a Restricted Gaussian Oracle . In Proceedings of Thirty Fourth Conference on Learning Theory , pages 2993--3050. PMLR, July 2021

2021
[17]

Characterizing Dependence of Samples along the Langevin Dynamics and Algorithms via Contraction of \ \ - Mutual Information , June 2025

Jiaming Liang, Siddharth Mitra, and Andre Wibisono. Characterizing Dependence of Samples along the Langevin Dynamics and Algorithms via Contraction of \ \ - Mutual Information , June 2025

2025
[18]

Fast convergence of \ \ -divergence along the unadjusted langevin algorithm and proximal sampler

Siddharth Mitra and Andre Wibisono. Fast convergence of \ \ -divergence along the unadjusted langevin algorithm and proximal sampler. In Proceedings of The 36th International Conference on Algorithmic Learning Theory , pages 846--869. PMLR, February 2025

2025
[19]

On sample complexity for covariance estimation via the unadjusted Langevin algorithm, January 2026

Shogo Nakakita. On sample complexity for covariance estimation via the unadjusted Langevin algorithm, January 2026

2026
[20]

Optimal Scaling and Shaping of Random Walk Metropolis via Diffusion Limits of Block-I

Jeffrey Negrea. Optimal Scaling and Shaping of Random Walk Metropolis via Diffusion Limits of Block-I . I . D . Targets , February 2019

2019
[21]

Approximations and Scaling Limits of Markov Chains with Applications to MCMC and Approximate Inference

Jeffrey Negrea. Approximations and Scaling Limits of Markov Chains with Applications to MCMC and Approximate Inference . PhD thesis, University of Toronto (Canada), 2022

2022
[22]

A. M. Ostrowski. A quantitative formulation of sylvester's law of inertia*. Proceedings of the National Academy of Sciences, 45 0 (5): 0 740--744, May 1959

1959
[23]

A. M. Stuart. Inverse problems: A bayesian perspective. Acta Numerica, 19: 0 451--559, May 2010

2010
[24]

Optimal Preconditioning and Fisher Adaptive Langevin Sampling

Michalis Titsias. Optimal Preconditioning and Fisher Adaptive Langevin Sampling . Advances in Neural Information Processing Systems, 36: 0 29449--29460, December 2023

2023
[25]

High-Dimensional Probability: An Introduction with Applications in Data Science , volume 47

Roman Vershynin. High-Dimensional Probability: An Introduction with Applications in Data Science , volume 47. Cambridge university press, 2018

2018
[26]

Minimax Mixing Time of the Metropolis-Adjusted Langevin Algorithm for Log-Concave Sampling

Keru Wu, Scott Schmidler, and Yuansi Chen. Minimax Mixing Time of the Metropolis-Adjusted Langevin Algorithm for Log-Concave Sampling . Journal of Machine Learning Research, 23 0 (270): 0 1--63, 2022. ISSN 1533-7928

2022

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [1]

Christophe Andrieu, Anthony Lee, Sam Power, and Andi Q. Wang. Explicit convergence bounds for Metropolis Markov chains: Isoperimetry , spectral gaps and profiles. The Annals of Applied Probability, 34 0 (4): 0 4022--4071, August 2024. ISSN 1050-5164, 2168-8737. doi:10.1214/24-AAP2058

doi:10.1214/24-aap2058 2024

[2] [2]

Unadjusted Hamiltonian MCMC with stratified Monte Carlo time integration

Nawaf Bou-Rabee and Milo Marsden. Unadjusted Hamiltonian MCMC with stratified Monte Carlo time integration. The Annals of Applied Probability, 35 0 (1): 0 360--392, February 2025. ISSN 1050-5164, 2168-8737

2025

[3] [3]

Polar factorization and monotone rearrangement of vector-valued functions

Yann Brenier. Polar factorization and monotone rearrangement of vector-valued functions. Communications on Pure and Applied Mathematics, 44 0 (4): 0 375--417, 1991

1991

[4] [4]

Carlier, A

G. Carlier, A. Galichon, and F. Santambrogio. From Knothe 's transport to Brenier 's map and a continuation method for optimal transport. SIAM Journal on Mathematical Analysis, 41 0 (6): 0 2554--2576, January 2010

2010

[5] [5]

Hoffman, Daniel Lee, Ben Goodrich, Michael Betancourt, Marcus Brubaker, Jiqiang Guo, Peter Li, and Allen Riddell

Bob Carpenter, Andrew Gelman, Matthew D. Hoffman, Daniel Lee, Ben Goodrich, Michael Betancourt, Marcus Brubaker, Jiqiang Guo, Peter Li, and Allen Riddell. Stan: A Probabilistic Programming Language . Journal of Statistical Software, 76: 0 1--32, January 2017. ISSN 1548-7660

2017

[6] [6]

Chatterji, Peter L

Xiang Cheng, Niladri S. Chatterji, Peter L. Bartlett, and Michael I. Jordan. Underdamped Langevin MCMC : A non-asymptotic analysis. In Proceedings of the 31st Conference On Learning Theory , pages 300--323. PMLR, July 2018

2018

[7] [7]

Log-concave sampling

Sinho Chewi. Log-concave sampling. Book draft: https://chewisinho. github. io, 2023

2023

[8] [8]

Air Markov Chain Monte Carlo , January 2018

Cyril Chimisov, Krzysztof Latuszynski, and Gareth Roberts. Air Markov Chain Monte Carlo , January 2018

2018