Pith. sign in

REVIEW 64 references

Beyond Modern Asymptotics for Log-Likelihood Ratios in Logistic Regression

T0 review · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read In binary logistic regression, the worst-case (1−δ) quantile of the log-likelihood ratio is d log(en/d) + log(1/δ) for n ≥ d ≥ 3, up to universal constants.

desk verdict The d≥3 worst-case LLR characterization is a genuine finite-sample Wilks analogue and deserves a serious referee; the d=2 log log log n upper bound is not yet proven — it sits in an appendix as an AI-generated proof sketch with load-bearing gaps. read the letter →

arxiv 2608.02507 v1 pith:7OCFL2UO submitted 2026-08-03 math.ST cs.ITcs.LGmath.ITstat.TH

Hugo Chardon , Reese Pathak , Nikita Zhivotovskiy This is my paper
classification math.STcs.ITcs.LGmath.ITstat.TH MSC 62E2062F2562J12
keywords logisticregressionlog-likelihoodratioWilksphenomenonShtarkovsumworst-casedesignVandermondesubspacesconfidencesetsGaussianrandom
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 proves a finite-sample, worst-case characterization of the log-likelihood-ratio statistic in logistic regression. For n≥d≥3, the largest possible (1−δ) quantile over every fixed design matrix and every target parameter is, up to universal constants, d log(en/d) + log(1/δ); the d=2 worst case is only log log log n + log(1/δ), and d=1 is log(1/δ). This replaces the asymptotic Wilks chi-square calibration with a nonasymptotic guarantee that holds uniformly, with no regularity conditions on the design and no requirement that the maximum likelihood estimator exist. The payoff is a confidence set for the target parameter that covers with probability at least 1−δ for every design and every parameter. The proof turns the likelihood ratio into a supremum over a linear subspace, bounds its exponential moment by a Shtarkov sum, and shows the bound is tight via Vandermonde designs whose polynomial sign patterns force the logarithmic inflation; the d=2 upper bound is presented as a proof sketch in the appendix.

What carries the argument

The argument rests on three mechanisms. First, the log-likelihood ratio depends on the design only through the subspace W = ran(X) and on the target only through v = Xθ*, reducing the statistic to Γ_W(ε;v) = sup_{w∈W} log(p_w(ε)/p_v(ε)). Second, the exponential moment of Γ_W evaluated at λ=1 is exactly the Shtarkov sum S_log(W) = Σ_ε sup_{w∈W} p_w(ε), which is independent of v and bounded by the number of regions in an arrangement of at most n hyperplanes, at most Σ_{ℓ=0}^k C(n,ℓ) ≤ (en/k)^k; Markov's inequality then yields the upper tail. Third, matching lower bounds come from Vandermonde subspaces—ranges of polynomial-evaluation maps—whose sign patterns realize all patterns with few negati

What would settle it

For n = exp(exp(exp(20))), simulate the two-dimensional block design from Proposition 4.1 with k=20 and estimate the 90% quantile of the log-likelihood ratio; the paper's d=2 claim predicts it to be about 20 + log(1/δ) up to a universal constant, so a value outside a constant factor of that would refute the d=2 scale.

Watch

Extended reading notes

Core claim

The paper's central discovery is a nonasymptotic Wilks-type law for logistic regression that is uniform in the worst possible way. For n≥d≥3, no matter which fixed design vectors x_1,...,x_n in R^d and no matter which target parameter θ* in R^d, the (1−δ) quantile of the log-likelihood ratio Λ satisfies Q_{1−δ} ≈ d log(en/d) + log(1/δ). The same scale holds for random design when the supremum is taken over all distributions. In dimension 2 the worst case drops to log log log n + log(1/δ), and in dimension 1 to log(1/δ). Under i.i.d. Gaussian design the logarithmic factor disappears: with n ≳ d + log(1/δ), the bound is d + log(1/δ), uniformly over θ*. The upper bounds require no MLE existence

Load-bearing premise

Everything in the d≥3 theorem rests on standard inequalities, but the d=2 upper bound rests on the appendix's proof sketch—after fixing the common offset and conditioning on the number of mislabeled points, the remaining uncertainty is handled by a one-dimensional bound—and if that reduction or the planar Helly step has a gap, the log log log n scale for d=2 is not established.

Editorial extensions

If this is right

  • The confidence set {θ : Λ(θ) ≤ d log(en/d) + log(1/δ)} has coverage at least 1−δ for every fixed design and every θ*, with no regularity and no need for the MLE to exist.
  • The logarithmic factor log(en/d) is an unavoidable price of uniformity: some fixed designs and target parameters (Vandermonde designs with θ* along a coordinate axis) have likelihood-ratio quantiles of this order, so the classical chi-square scale cannot hold in worst-case fixed design.
  • Under i.i.d. Gaussian design, the classical chi-square-type scale d + log(1/δ) is restored uniformly over θ* as soon as n ≳ d + log(1/δ).
  • The validity of the classical Wilks approximation at the origin is governed by d^{3/2}/n → 0, not by the aspect ratio d/n; the distribution can be far from chi-square even when d/n is small.
  • The worst-case scale is sharply dimension-sensitive: d=1 costs only log(1/δ), d=2 costs log log log n, and d≥3 costs d log(en/d).

Reading between the lines

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

  • The same subspace-and-Shtarkov machinery likely gives sharp finite-sample regret bounds for transductive sequential prediction of logistic labels; the paper's upper bound is effectively a worst-case regret of order d log(en/d) for the logistic class, and the lower bounds indicate this is minimax in the worst case.
  • The d=2 anomaly suggests that the plane has special geometry for this problem; one can ask whether other low-dimensional substructures inside higher-dimensional designs could create intermediate scales, a question the paper leaves open.
  • The Gaussian-design guarantee is proved only for Gaussian covariates, but the route through Hessian concentration suggests a testable extension to other random designs with well-conditioned covariance and sub-Gaussian tails; this is an editorial conjecture, not a paper claim.
  • The d=2 upper bound rests on an appendix proof sketch whose key reductions are asserted rather than fully fleshed out; until that sketch is completed, the log log log n scale should be treated as conditional, while the d≥3 statement does not inherit this caveat.
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.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the main d>=3 worst-case derivation is self-contained; the d=2 upper bound rests on an unverified AI-assisted proof sketch, which is a correctness risk, not a circularity.

full rationale

The central result, Theorem 2.1 for d>=3, is derived without fitting or renaming inputs as outputs. The upper bound comes from the Shtarkov-sum/hyperplane-arrangement count of Lemma 2.2 / Theorem 4.1(i) plus Proposition 2.1, and the lower bound comes from an explicit Vandermonde subspace construction in Proposition 2.4. These arguments are self-contained and do not assume the target quantile scale. No parameter is fitted to a subset of data and then called a prediction; the worst-case quantile is derived uniformly over all designs and all target parameters. The paper does contain self-citations: [14] is used for the Gaussian-design analysis, with one overlapping author, but [14] is an independent prior result with its own stated assumptions and proofs and is not used for the d>=3 worst-case theorem. The transductive regret paper [45] is mentioned but explicitly not relied upon. The main caveat is Appendix A: Theorem A.1, the d=2 upper bound, is credited to GPT-5.6-Sol, and the proof is only a sketch. The critical scalar reduction (58) and the pathwise planar quantitative Helly aggregation are asserted rather than fully proved. This is a genuine verification gap for the advertised log log log n scale, but it is not circularity: the asserted steps reduce the problem to univariate bounds and geometric covering arguments, not to the theorem being proved. Thus the paper does not exhibit definitional circularity, fitted-prediction circularity, or a load-bearing self-citation loop; the low score reflects only minor self-citations and the verification risk in the appendix.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The central theorem rests on standard combinatorial geometry (hyperplane arrangements, Cover's formula, conic kinematics) and, for the Gaussian-design section, on externally cited MLE bounds from [14]. No free parameters are fitted; universal constants are existential, not tuned to data.

assumptions (8)
  • standard math An arrangement of at most n affine hyperplanes in R^d has at most sum_{ℓ=0}^d C(n,ℓ) cells.
    Used in Theorem 4.1(i)/Lemma 2.2 to bound the logistic Shtarkov sum by counting realizable sign patterns.
  • standard math Hoeffding's extremal theorem [27, Theorem 4]: among Poisson-binomial laws with mean k, Bin(n,k/n) maximizes upper-tail probabilities.
    Used in Theorem 4.1(ii) to lower-bound the one-dimensional Shtarkov sum by the Bernoulli Shtarkov constant.
  • standard math Cover's formula: for n points in general position in R^d, the number of dichotomies induced by homogeneous halfspaces is 2 sum_{ℓ=0}^{d−1} C(n−1,ℓ).
    Used in Theorem 4.1(iv) to lower-bound Shtarkov sums for designs in general position.
  • standard math Conic kinematic estimate of Amelunxen–Lotz–McCoy–Tropp [4, Thm 7.1] for random subspaces and cones, and the statistical-dimension bound for the cones C_r.
    Used in Lemma 3.4 to control the exponential moment of the critical radius for rotationally invariant designs.
  • standard math Planar quantitative Helly theorem [8]: for convex sets in R^2, a small subfamily controls the intersection area, used to reduce the number of dyadic blocks to at most 4.
    Invoked in Appendix A proof sketch of Theorem A.1; the pathwise application is asserted, not fully proved.
  • domain assumption Nonasymptotic bounds on the logistic MLE and Hessian under Gaussian design from Chardon–Lerasle–Mourtada [14, Prop 5, Thm 6, Lemma 3, Sec 8.2].
    Imported in Theorem 3.1 to provide concentration of the MLE and ∇-Hessian in the regular regime of Gaussian design.
  • domain assumption The observational model is well-specified logistic regression with deterministic covariates and labels in {−1,1}.
    The whole paper works under P_θ*(Y_i=1)=σ(⟨x_i,θ*⟩) for fixed x_i; all bounds are for this model.
  • standard math Okamoto's lower-tail bound for binomial tails [41, Theorem 2].
    Used in Proposition 2.4(iii) to show E[N ε 1_{N≤r}] ≥ c r for the Vandermonde lower bound.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Beyond Modern Asymptotics for Log-Likelihood Ratios in Logistic Regression." pith.science (2026). https://pith.science/paper/7OCFL2UO

@misc{pith2026260802507,
  author       = {Pith},
  title        = {Pith review of: Beyond Modern Asymptotics for Log-Likelihood Ratios in Logistic Regression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7OCFL2UO}},
  note         = {Machine review of arXiv:2608.02507}
}
abstract

We characterize the finite sample behavior of the log-likelihood ratio statistic in binary logistic regression, uniformly over both the design and the target parameter. For $n\geq d\geq 3$, we determine, up to universal constants, its worst case $(1-\delta)$ quantile over all fixed collections of design vectors and all target parameters: \[ d\log\left(\frac{e n}{d}\right)+\log\left(\frac{1}{\delta}\right). \] This is a nonasymptotic analogue of the Wilks $\chi^2_d$ phenomenon and requires no regularity assumptions on the design. The low dimensional cases exhibit unusual behavior. The worst case quantile in dimension $d=2$ is sharply of order \[ \log\log\log n+\log\left(\frac{1}{\delta}\right). \] The worst case quantile in dimension $d=1$ is of order $\log(1/\delta)$, with no dependence on $n$. Finally, i.i.d. Gaussian design vectors recover the classical Wilks scale. In the regime $n\gtrsim d+\log(1/\delta)$, we prove the sharp bound \[ d+\log\left(\frac{1}{\delta}\right). \] Unlike existing asymptotic results, our bounds are uniform over the target parameter, which may depend on $n$, $d$, and $\delta$.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

64 extracted references · 5 linked inside Pith

  1. [14]

    Chardon, M

    H. Chardon, M. Lerasle, and J. Mourtada. Finite-sample performance of the maximum likelihood estimator in logistic regression.arXiv preprint arXiv:2411.02137v3, 2024

    arXiv 2024

  2. [45]

    J. Qian, A. Rakhlin, and N. Zhivotovskiy. Refined risk bounds for unbounded losses via transductive priors. Journal of Machine Learning Research, 27(26):1–64, 2026

    2026

  3. [1]

    Abbasi-Yadkori, D

    Y. Abbasi-Yadkori, D. P´ al, and C. Szepesv´ ari. Online-to-confidence-set conversions and application to sparse stochastic bandits. InProceedings of the Fifteenth International Conference on Artificial Intelligence and Statis- tics, volume 22, pages 1–9. PMLR, 2012

    2012

  4. [2]

    R. Agrawal. Finite-sample concentration of the multinomial in relative entropy.IEEE Transactions on Infor- mation Theory, 66(10):6297–6302, 2020

    2020

  5. [3]

    Albert and J

    A. Albert and J. A. Anderson. On the existence of maximum likelihood estimates in logistic regression models. Biometrika, 71(1):1–10, 1984

    1984

  6. [4]

    Amelunxen, M

    D. Amelunxen, M. Lotz, M. B. McCoy, and J. A. Tropp. Living on the edge: Phase transitions in convex programs with random data.Information and Inference: A Journal of the IMA, 3(3):224–294, 2014

    2014

  7. [5]

    Andresen and V

    A. Andresen and V. Spokoiny. Critical dimension in profile semiparametric estimation.Electronic Journal of Statistics, 8(2):3077–3125, 2014

    2014

  8. [6]

    F. Bach. Self-concordant analysis for logistic regression.Electronic Journal of Statistics, 4:384–414, 2010

    2010

Show all 64 references
  1. [7]

    Baillon, R

    J.-B. Baillon, R. Cominetti, and J. Vaisman. A sharp uniform bound for the distribution of sums of Bernoulli trials.Combinatorics, Probability and Computing, 25(3):352–361, 2016

    2016

  2. [8]

    B´ ar´ any, M

    I. B´ ar´ any, M. Katchalski, and J. Pach. Quantitative Helly-type theorems.Proceedings of the American Mathe- matical Society, 86(1):109–114, 1982

    1982

  3. [9]

    A. R. Barron. Information-theoretic characterization of Bayes performance and the choice of priors in parametric and nonparametric problems. In J. M. Bernardo, J. O. Berger, A. P. Dawid, and A. F. M. Smith, editors, Bayesian Statistics 6: Proceedings of the Sixth Valencia Inte...

    1999

  4. [10]

    P. J. Bickel and K. A. Doksum.Mathematical Statistics: Basic Ideas and Selected Topics, volume 1. Pearson Prentice Hall, Upper Saddle River, NJ, 2nd edition, 2007

    2007

  5. [11]

    Boucheron, G

    S. Boucheron, G. Lugosi, and P. Massart.Concentration Inequalities: A Nonasymptotic Theory of Independence. Oxford University Press, Oxford, 2013

    2013

  6. [12]

    E. J. Cand` es and P. Sur. The phase transition for the existence of the maximum likelihood estimate in high- dimensional logistic regression.The Annals of Statistics, 48(1):27–42, 2020

    2020

  7. [13]

    Cesa-Bianchi and G

    N. Cesa-Bianchi and G. Lugosi.Prediction, Learning, and Games. Cambridge University Press, Cambridge, 2006

    2006

  8. [15]

    Chernoff

    H. Chernoff. On the distribution of the likelihood ratio.Ann. Math. Statistics, 25(3):573–578, 1954

    1954

  9. [16]

    Clerico, H

    E. Clerico, H. Flynn, W. Kot lowski, and G. Neu. Confidence sequences for generalized linear models via regret analysis.arXiv preprint arXiv:2504.16555, 2025. 51

    2025 arXiv

  10. [17]

    T. M. Cover. Geometrical and statistical properties of systems of linear inequalities with applications in pattern recognition.IEEE Transactions on Electronic Computers, EC-14(3):326–334, 1965

    1965

  11. [18]

    J. N. Darroch. On the distribution of the number of successes in independent trials.The Annals of Mathematical Statistics, 35(3):1317–1321, 1964

    1964

  12. [19]

    Dembo and O

    A. Dembo and O. Zeitouni.Large deviations techniques and applications, volume 38 ofStochastic Modelling and Applied Probability. Springer-Verlag, Berlin, 2010. Corrected reprint of the second (1998) edition

    2010

  13. [20]

    J. G. Dowty. Volumes of logistic regression models with applications to model selection.arXiv preprint arXiv:1408.0881, 2014

    2014 arXiv

  14. [21]

    Drmota, P

    M. Drmota, P. Jacquet, C. Wu, and W. Szpankowski. Minimax regret with unbounded weights. In2024 IEEE International Symposium on Information Theory (ISIT), pages 2305–2310. IEEE, 2024

    2024

  15. [22]

    Drmota, P

    M. Drmota, P. Jacquet, C. Wu, and W. Szpankowski. Precise regularized minimax regret with unbounded weights.IEEE Transactions on Information Theory, 71(11):8753–8776, 2025

    2025

  16. [23]

    Drmota, P

    M. Drmota, P. Jacquet, C. Wu, and W. Szpankowski. Phase transition of regret for logistic regression with large weights. In M. Telgarsky and J. Ullman, editors,Proceedings of the 37th International Conference on Algorithmic Learning Theory, volume 313 ofProceedings of Machine ...

    2026

  17. [24]

    J. Fan, C. Zhang, and J. Zhang. Generalized likelihood ratio statistics and Wilks phenomenon.The Annals of Statistics, 29(1):153–193, 2001

    2001

  18. [25]

    D. J. Foster, S. Kale, H. Luo, M. Mohri, and K. Sridharan. Logistic regression: The importance of being improper. InProceedings of the 31st Conference On Learning Theory, volume 75, pages 167–208, 2018

    2018

  19. [26]

    F. R. Guo and T. S. Richardson. Chernoff-type concentration of empirical probabilities in relative entropy.IEEE Transactions on Information Theory, 67(1):549–558, 2021

    2021

  20. [27]

    Hoeffding

    W. Hoeffding. On the distribution of the number of successes in independent trials.The Annals of Mathematical Statistics, 27(3):713–721, 1956

    1956

  21. [28]

    Hsu and A

    D. Hsu and A. Mazumdar. On the sample complexity of parameter estimation in logistic regression with normal design. InProceedings of Thirty Seventh Conference on Learning Theory, volume 247 ofProceedings of Machine Learning Research, pages 2418–2437. PMLR, 2024

    2024

  22. [29]

    I. A. Ibragimov and R. Z. Khas’minskii.Statistical Estimation: Asymptotic Theory, volume 16 ofApplications of Mathematics. Springer-Verlag, New York, 1981

    1981

  23. [30]

    Jacquet, G

    P. Jacquet, G. Shamir, and W. Szpankowski. Precise minimax regret for logistic regression with categorical feature values. In V. Feldman, K. Ligett, and S. Sabato, editors,Proceedings of the 32nd International Conference on Algorithmic Learning Theory, volume 132 ofProceedings...

    2021

  24. [31]

    Jacquet, G

    P. Jacquet, G. I. Shamir, and W. Szpankowski. Precise minimax regret for logistic regression. In2022 IEEE International Symposium on Information Theory (ISIT), pages 444–449, 2022

    2022

  25. [32]

    S. M. Kakade and A. Y. Ng. Online bounds for Bayesian algorithms. In L. K. Saul, Y. Weiss, and L. Bottou, editors,Advances in Neural Information Processing Systems 17, pages 641–648. MIT Press, 2005

    2005

  26. [33]

    R. W. Keener.Theoretical Statistics: Topics for a Core Course. Springer Science & Business Media, 2010

    2010

  27. [34]

    Kirschner, A

    J. Kirschner, A. Krause, M. Meziu, and M. Mutn´ y. Confidence estimation via sequential likelihood mixing. arXiv preprint arXiv:2502.14689, 2025

    2025 arXiv

  28. [35]

    Kuchelmeister and S

    F. Kuchelmeister and S. van de Geer. Finite sample rates for logistic regression with small noise or few samples. Sankhya A, 2024

    2024

  29. [36]

    Laurent and P

    B. Laurent and P. Massart. Adaptive estimation of a quadratic functional by model selection.Ann. Statist., 28(5):1302–1338, 2000

    2000

  30. [37]

    L. Le Cam. Locally asymptotically normal families of distributions.University of California Publications in Statistics, 3(2):37–98, 1960

    1960

  31. [38]

    Le Cam and G

    L. Le Cam and G. L. Yang.Asymptotics in Statistics: Some Basic Concepts. Springer Series in Statistics. Springer, New York, 2nd edition, 2000

    2000

  32. [39]

    Lee, S.-Y

    J. Lee, S.-Y. Yun, and K.-S. Jun. Improved regret bounds of (multinomial) logistic bandits via regret-to- confidence-set conversion. InProceedings of the 27th International Conference on Artificial Intelligence and Statistics, volume 238 ofProceedings of Machine Learning Resea...

    2024

  33. [40]

    A. Maurer. A note on the PAC Bayesian theorem.arXiv preprint cs/0411099, 2004

    2004 arXiv

  34. [41]

    M. Okamoto. Some inequalities relating to the partial sum of binomial probabilities.Annals of the Institute of Statistical Mathematics, 10(1):29–35, 1959

    1959

  35. [42]

    Orlitsky and N

    A. Orlitsky and N. P. Santhanam. Speaking of infinity [iid strings].IEEE Transactions on Information Theory, 50(10):2215–2230, 2004

    2004

  36. [43]

    D. M. Ostrovskii and F. Bach. Finite-sample analysis of M-estimators using self-concordance.Electronic Journal of Statistics, 15(1):326–391, 2021

    2021

  37. [44]

    S. Portnoy. Asymptotic behavior of likelihood methods for exponential families when the number of parameters tends to infinity.The Annals of Statistics, 16(1):356–366, 1988

    1988

  38. [46]

    J. J. Rissanen. Fisher information and stochastic complexity.IEEE Transactions on Information Theory, 42(1):40–47, 1996

    1996

  39. [47]

    R. T. Rockafellar.Convex analysis. Princeton Mathematical Series, No. 28. Princeton University Press, Prince- ton, NJ, 1970

    1970

  40. [48]

    G. I. Shamir and W. Szpankowski. A lower bound for regret in logistic regression. In2021 IEEE International Symposium on Information Theory (ISIT), pages 2507–2512, 2021

    2021

  41. [49]

    Y. M. Shtarkov. Universal sequential coding of single messages.Problems of Information Transmission, 23(3):175–186, 1987

    1987

  42. [50]

    Spokoiny

    V. Spokoiny. Parametric estimation. Finite sample theory.The Annals of Statistics, 40(6):2877–2909, 2012

    2012

  43. [51]

    Spokoiny

    V. Spokoiny. Finite sample expansions and risk bounds in high-dimensional SLS models.arXiv preprint arXiv:2404.14227v3, 2025

    2025 arXiv

  44. [52]

    Spokoiny and M

    V. Spokoiny and M. Zhilova. Bootstrap confidence sets under model misspecification.The Annals of Statistics, 43(6):2653–2675, 2015

    2015

  45. [53]

    Sur and E

    P. Sur and E. J. Cand` es. A modern maximum-likelihood theory for high-dimensional logistic regression.Pro- ceedings of the National Academy of Sciences, 116(29):14516–14525, 2019

    2019

  46. [54]

    P. Sur, Y. Chen, and E. J. Cand` es. The likelihood ratio test in high-dimensional logistic regression is asymp- totically a rescaled chi-square.Probability Theory and Related Fields, 175(1–2):487–558, 2019

    2019

  47. [55]

    A. W. van der Vaart.Asymptotic statistics, volume 3 ofCambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, Cambridge, 1998

    1998

  48. [56]

    J. G. Wendel. A problem in geometric probability.Mathematica Scandinavica, 11:109–112, 1962

    1962

  49. [57]

    S. S. Wilks. The large-sample distribution of the likelihood ratio for testing composite hypotheses.The Annals of Mathematical Statistics, 9(1):60–62, 1938

    1938

  50. [58]

    0 is trivial. Assume thatHě1, fixRą0, and defineB R“ ␣ uPR 2 :}u} 2ďR ( . ForPĂt1, . . . , Hu, setL∅

    Q. Zhao, P. Sur, and E. J. Cand` es. The asymptotic distribution of the MLE in high-dimensional logistic models: Arbitrary covariance.Bernoulli, 28(3):1835–1861, 2022. A AI disclosure and the proof of thed“2upper bound The results in the main body of the paper were developed b...

    2022

  51. [59]

    Extreme block: modal-recode and set \(a_i=|v_i|\). For \(E=\{i:a_i>2\log n\}\) and the true product law \(P_E\) of their modal-error vector \(B_E\), prove \[ \Gamma_E(B_E)\le-\log P_E(B_E),\qquad \mathbb E e^{\eta[-\log P_E(B_E)]}\le C \] for universal \(\eta,C>0\)

  52. [60]

    Treat all-equal fibers and endpoint suprema directly

    Offset profile: for arbitrary fixed logistic offsets and a common intercept, profile the intercept, condition on total error count, regularize the scalar slope, and prove a universal fractional exponential moment for the extended gain. Treat all-equal fibers and endpoint supre...

  53. [61]

    Factor-two block: normalize by its true margins; use concavity to compare the rescaled likelihoods; then reduce its gain, including fractional rescaling through Bernoulli thinning, to a bounded number of arbitrary-offset common-intercept profiles

  54. [62]

    \] On a shell \(s\le\Gamma<2s\), the ridge/KL identity gives \(Q(h_\varepsilon)\le Cs\)

    Central geometry: use uniform local logistic curvature and \[ Q(h)=\sum_i\min\{z_i(h)^2,1\}. \] On a shell \(s\le\Gamma<2s\), the ridge/KL identity gives \(Q(h_\varepsilon)\le Cs\). For \(u\ge1\), prove by planar rank-two packing \[ \#\{i:\exists h,\ Q(h)\le u,\ |z_i(h)|\ge1\}...

  55. [63]

    Central gain: control the exceptional rows by a finite planar ray profile and the rest by a rank-two quadratic profile, obtaining a universal exponential tail and fractional exponential moment for the central-block gain

  56. [64]

    With \(\Delta=\sum_{s=0}^H\Gamma_s-\Gamma_\varepsilon\), prove that planar quantitative Helly selects \(|P|\le4\) with \[ \frac{Z_P}{Z}\le C(H+1)^2(\Delta+1)^2

    Aggregation: use convex deficits and the sublevel-set/partition-function comparison. With \(\Delta=\sum_{s=0}^H\Gamma_s-\Gamma_\varepsilon\), prove that planar quantitative Helly selects \(|P|\le4\) with \[ \frac{Z_P}{Z}\le C(H+1)^2(\Delta+1)^2. \] 62

Pith tools

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