Pith. sign in

REVIEW 1 major objections 20 references

A unified complexity bound for logconcave sampling

T0 review · 1 major / 0 minor · reviewed 2026-06-27 · grok-4.3

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

Pith's one-line read In-and-Out with exponential lifting yields a nearly tight unified bound for sampling any logconcave distribution from a warm start.

desk verdict Kook and Vempala unify logconcave sampling rates via an improved Poincaré bound on lifted distributions, but the unification stands or falls on whether that bound avoids hidden case splits between constrained and smooth regimes. read the letter →

arxiv 2606.12694 v1 pith:PP7ELGAR submitted 2026-06-10 cs.DS cs.LGmath.PRstat.ML

Yunbum Kook , Santosh S. Vempala This is my paper · ORCID
classification cs.DScs.LGmath.PRstat.ML
keywords logconcavesamplingIn-and-OutalgorithmexponentialliftingPoincaréconstantcomplexityboundsMarkovchainMonteCarloconvexbodieswarmstart
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 establishes a single complexity bound that covers sampling from arbitrary logconcave distributions by running the In-and-Out algorithm after an exponential lift. The central new step is a stronger uniform bound on the Poincaré constant of the lifted distribution. This bound produces convergence rates that are nearly tight both when the support is a convex body and when the density is strongly logconcave and smooth. A reader would care because earlier results required separate arguments for each regime and often lost tightness. The approach therefore simplifies analysis while preserving near-optimality across the full class of logconcave targets.

What carries the argument

Improved bound on the Poincaré constant of the exponentially lifted distribution, which supplies a uniform mixing guarantee across constrained and well-conditioned logconcave targets.

What would settle it

A concrete logconcave distribution whose exponentially lifted version has a Poincaré constant larger than the claimed uniform bound, causing the In-and-Out chain to mix slower than the stated rate.

Watch

Extended reading notes

Core claim

We give a simple, unified, and nearly tight bound for sampling arbitrary logconcave distributions from a warm start using the In-and-Out algorithm along with exponential lifting. The main new ingredient in the analysis is an improved bound on the Poincaré constant of a lifted distribution. As a consequence, the resulting convergence rate is nearly tight for both constrained settings (e.g., Gaussian restricted to a convex body) and well-conditioned settings (e.g., strongly logconcave and smooth densities).

Load-bearing premise

The analysis requires that an improved Poincaré-constant bound for the exponentially lifted distribution holds uniformly for both constrained and well-conditioned logconcave cases.

Editorial extensions

If this is right

  • The convergence rate becomes nearly tight for a Gaussian restricted to any convex body.
  • The convergence rate becomes nearly tight for strongly logconcave and smooth densities.
  • A single algorithm and analysis now cover sampling from arbitrary logconcave distributions starting from a warm distribution.
  • The In-and-Out method after lifting achieves the unified bound without needing case-specific adjustments.

Reading between the lines

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

  • The lifting technique may simplify mixing analysis for other random-walk samplers on logconcave targets.
  • Implementations could use the same lifted chain for both polytope-constrained and smooth posterior sampling tasks.
  • Empirical checks on high-dimensional polytopes could confirm whether the predicted step count matches observed mixing times.
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, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 0 minor

Summary. The manuscript claims to provide a simple, unified, and nearly tight complexity bound for sampling arbitrary logconcave distributions from a warm start using the In-and-Out algorithm with exponential lifting. The key contribution is an improved bound on the Poincaré constant of the exponentially lifted distribution that enables the unification for both constrained and well-conditioned settings.

Significance. If the improved Poincaré bound holds uniformly without case distinctions, this work would offer a streamlined analysis that achieves nearly tight rates across different logconcave sampling regimes, potentially advancing the field by reducing the need for separate proofs.

major comments (1)
  1. [Main new ingredient (improved Poincaré bound for lifted distribution)] The unification rests on a single improved Poincaré constant bound for the exponentially lifted measure that applies uniformly to both the constrained case (logconcave restricted to convex body) and the well-conditioned case (strongly logconcave + smooth). The proof of this bound (the “main new ingredient”) must be checked to ensure it does not invoke smoothness or strong convexity in any step unavailable for general constrained logconcave densities; otherwise the single-bound claim collapses.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We are grateful to the referee for their thoughtful review and for identifying the key point that requires verification for the validity of our unified bound. Below we provide a point-by-point response to the major comment.

read point-by-point responses

  1. Referee: [Main new ingredient (improved Poincaré bound for lifted distribution)] The unification rests on a single improved Poincaré constant bound for the exponentially lifted measure that applies uniformly to both the constrained case (logconcave restricted to convex body) and the well-conditioned case (strongly logconcave + smooth). The proof of this bound (the “main new ingredient”) must be checked to ensure it does not invoke smoothness or strong convexity in any step unavailable for general constrained logconcave densities; otherwise the single-bound claim collapses.

    Authors: We thank the referee for this important observation. The proof of the improved bound on the Poincaré constant, presented in Section 3 of the manuscript, is designed to apply to general logconcave distributions without requiring strong convexity or smoothness. It uses only the logconcavity to establish the necessary variance bounds via the properties of the exponential lift, which are valid for densities restricted to convex bodies. No steps in the proof rely on differentiability or strong logconcavity. This ensures the unification holds as claimed. We are prepared to include additional explanatory remarks in a revised version if the referee believes it would strengthen the presentation. revision: partial

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; unified bound rests on independent Poincaré analysis

full rationale

The paper presents its main result as following from a new, improved bound on the Poincaré constant of the exponentially lifted distribution, which is introduced as the central technical contribution and claimed to hold uniformly for both constrained logconcave and well-conditioned settings. No equations or steps are shown reducing a prediction to a fitted parameter by construction, no self-citation is invoked as the sole justification for a uniqueness or ansatz claim, and the derivation chain does not rename a known empirical pattern. The abstract and description indicate a self-contained analysis whose key step is externally verifiable rather than tautological.

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

Based solely on the abstract, no free parameters, axioms, or invented entities are identifiable; the logconcavity assumption is standard in the domain.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A unified complexity bound for logconcave sampling." pith.science (2026). https://pith.science/paper/PP7ELGAR

@misc{pith2026260612694,
  author       = {Pith},
  title        = {Pith review of: A unified complexity bound for logconcave sampling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PP7ELGAR}},
  note         = {Machine review of arXiv:2606.12694}
}
read the original abstract

We give a simple, unified, and nearly tight bound for sampling arbitrary logconcave distributions from a warm start using the In-and-Out algorithm along with exponential lifting. The main new ingredient in the analysis is an improved bound on the Poincar\'e constant of a lifted distribution. As a consequence, the resulting convergence rate is nearly tight for both constrained settings (e.g., Gaussian restricted to a convex body) and well-conditioned settings (e.g., strongly logconcave and smooth densities).

Figures

Figures reproduced from arXiv: 2606.12694 by the authors.

Figure 1.1
Figure 1.1. Exponential lifting and the proximal sampler ( [PITH_FULL_IMAGE:figures/full_fig_p002_1_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references

  1. [1]

    Tyrrell and Wets, Roger J.-B

    Rockafellar, R. Tyrrell and Wets, Roger J.-B. , publisher =. Variational analysis , volume =

  2. [2]

    The geometry of logconcave functions and sampling algorithms , volume =

    Lov\'. The geometry of logconcave functions and sampling algorithms , volume =. Random Structures & Algorithms , number =

  3. [3]

    The entropic barrier is n -self-concordant , year =

    Chewi, Sinho , booktitle =. The entropic barrier is n -self-concordant , year =

  4. [4]

    , journal =

    Andrieu, Christophe and Lee, Anthony and Power, Sam and Wang, Andi Q. , journal =. Explicit convergence bounds for

  5. [5]

    Entropy contraction of the

    Filippo Ascolani and Hugo Lavenant and Giacomo Zanella , journal =. Entropy contraction of the

  6. [6]

    , booktitle =

    Kook, Yunbum and Zhang, Matthew S. , booktitle =. R\'

  7. [7]

    A random polynomial-time algorithm for approximating the volume of convex bodies , volume =

    Dyer, Martin and Frieze, Alan and Kannan, Ravi , journal =. A random polynomial-time algorithm for approximating the volume of convex bodies , volume =

  8. [8]

    Structured logconcave sampling with a restricted

    Lee, Yin Tat and Shen, Ruoqi and Tian, Kevin , booktitle =. Structured logconcave sampling with a restricted

Show all 20 references
  1. [9]

    The strong data processing inequality under the heat flow , volume =

    Klartag, Bo'az and Ordentlich, Or , journal =. The strong data processing inequality under the heat flow , volume =

  2. [10]

    Improved analysis for a proximal algorithm for sampling , volume =

    Chen, Yongxin and Chewi, Sinho and Salim, Adil and Wibisono, Andre , booktitle =. Improved analysis for a proximal algorithm for sampling , volume =

  3. [11]

    and Zhang, Matthew S

    Kook, Yunbum and Vempala, Santosh S. and Zhang, Matthew S. , journal =. In-and-

  4. [12]

    Concentration of the information in data with log-concave distributions , volume =

    Bobkov, Sergey and Madiman, Mokshay , journal =. Concentration of the information in data with log-concave distributions , volume =

  5. [13]

    Optimal Concentration of Information Content for Log-Concave Densities , year =

    Fradelizi, Matthieu and Madiman, Mokshay and Wang, Liyao , pages =. Optimal Concentration of Information Content for Log-Concave Densities , year =

  6. [14]

    Logarithmic bounds for isoperimetry and slices of convex sets , year =

    Klartag, Bo'az , journal =. Logarithmic bounds for isoperimetry and slices of convex sets , year =

  7. [15]

    Nguyen, Van Hoang , date-added =. In

  8. [16]

    Heat capacity bound, energy fluctuations and convexity , year =

    Wang, Liyao , date-added =. Heat capacity bound, energy fluctuations and convexity , year =

  9. [17]

    Tyrrell , publisher =

    Rockafellar, R. Tyrrell , publisher =. Convex analysis , year =

  10. [18]

    , booktitle =

    Kook, Yunbum and Vempala, Santosh S. , booktitle =. Faster logconcave sampling from a cold start in high dimension , year =

  11. [19]

    , booktitle =

    Kook, Yunbum and Vempala, Santosh S. , booktitle =. Sampling and integration of logconcave functions by algorithmic diffusion , year =

  12. [20]

    , journal =

    Kook, Yunbum and Vempala, Santosh S. , journal =. Zeroth-order logconcave sampling , year =

Pith tools

Reviewed June 27, 2026 · model on record in the stance chip above.