On the Universality of Simple Trust-Region Algorithms

Clemens Sirotenko

REVIEW 2 major objections 5 minor 65 references

On the Universality of Simple Trust-Region Algorithms

T0 review · 2 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read The classical quadratic trust-region method, given no information about the smoothness of the objective, provably attains the optimal iteration complexity in convex, nonconvex, and local regimes under Hölder-continuous Hessians.

desk verdict Sharp idea, one bad constant: Lemma 12's beta_tau invalidates the convex proof for the basic method as written, but the UniCAT analysis and the core mechanism look sound. read the letter →

arxiv 2607.19647 v1 pith:LRDHJNIC submitted 2026-07-22 math.OC

Clemens Sirotenko This is my paper

classification math.OC MSC 90C3090C2590C60

keywords trust-regionmethodsuniversalcomplexityHölder-continuousHessianconvexoptimizationnonconvexNewtonmethoditeration

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 claims that the classical trust-region mechanism is inherently adaptive. For the basic ratio-test method with inexact subproblem solves, it proves the universal convex complexity bound O(ε^{-1/(1+ν)}) under a ν-Hölder-continuous Hessian, with no knowledge of ν or the Hölder constant. For a variant with a modified acceptance ratio and exact solves (UniCAT), it proves the same convex bound together with the optimal nonconvex first-order complexity O(ε^{-(2+ν)/(1+ν)}) and local Newton rates. The key new ingredient is a function-gap model-decrease estimate that links the model decrease to the function gap rather than the gradient norm. If correct, this shows that trust-region methods automatically adapt to unknown smoothness across all three regimes, without cubic, gradient, or tensor regularization.

What carries the argument

The central mechanism is the function-gap model-decrease estimate (Lemma 7, property (MD_α)): for a convex objective with Hölder-continuous Hessian, the optimal trust-region model decrease at x_k is bounded below by (1+η_2)F_k/(2D) min{βF_k^α, Δ}, where F_k = f(x_k)-f* is the function gap, α = 1/(1+ν), and β, D are constants. This replaces the classical Cauchy-decrease (gradient-norm) estimate with one in terms of the function gap, and it supplies one of the two hypotheses of an abstract convex complexity theorem (Theorem 8); the other is that sufficiently small trust-region radii are automatically very successful. Together they yield the universal rate O(ε^{-1/(1+ν)}).

What would settle it

Implement the basic ratio-test trust-region method with a near-exact subproblem solver on convex functions with known Hölder exponent ν (for example f(x) = |x|^{2+ν} on ℝ) and check whether the number of iterations to reach f(x)-f* ≤ ε scales as ε^{-1/(1+ν)} without any ν-dependent tuning. An observed exponent different from 1/(1+ν), or a need to know ν to achieve it, would refute the universality claim.

Watch

Extended reading notes

Core claim

The paper proves that the classical quadratic trust-region method is universal under convexity: for every ν∈[0,1], under a ν-Hölder-continuous Hessian, the basic ratio-test method with inexact solves attains the global complexity O(ε^{-1/(1+ν)}) for an ε-approximate minimizer, with no knowledge of ν or its Hölder constant; the UniCAT variant—same quadratic model, slightly modified acceptance ratio, exact subproblem solves—attains the same convex bound, the optimal nonconvex first-order complexity O(ε^{-(2+ν)/(1+ν)}), and local Newton rates in one algorithm. The central new estimate is a function-gap model-decrease inequality that bounds the model decrease by the scaled function gap, and the

Load-bearing premise

The load-bearing premise is that the method can compute, at every iteration, a trust-region step that captures a fixed fraction of the optimal model decrease (for the basic variant) or the exact optimal step (for the faster variant); if only the weaker standard Cauchy-decrease condition is available, the universal convex analysis does not go through.

Editorial extensions

If this is right

The basic ratio-test trust-region method, with near-exact subproblem solves, achieves the universal convex rate O(ε^{-1/(1+ν)}) for every ν∈[0,1], improving on the standard O(ε^{-1}) bound whenever ν>0.
The UniCAT variant attains the optimal nonconvex first-order complexity O(ε^{-(2+ν)/(1+ν)}) while preserving the convex rate and local Newton rates—the first simple trust-region method universal across all three regimes.
Under vanishing inexactness, both variants recover Q-superlinear convergence (ν=0) and order 1+ν (ν∈(0,1]) near nondegenerate minimizers, matching classical Newton behavior adaptively.
Any trust-region variant that satisfies the two abstract hypotheses—function-gap model decrease and small-radii success—inherits the universal convex rate (Theorem 8), giving a template for further designs.
The nonconvex O(ε^{-2}) guarantee for the basic method is retained under bounded Hessians, so the new convex results do not weaken the classical first-order bound.

Reading between the lines

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

If the universal bound holds, the practical bottleneck shifts to subproblem solvers: making the guarantee effective at scale requires inexact solvers satisfying condition (11), which standard truncated-CG does not; designing such solvers for large-scale problems is a natural next step.
The abstract convex theorem suggests that the radius-update rule, not the specific model, is what drives adaptation to unknown smoothness; this may transfer to quasi-Newton, stochastic, or proximal trust-region variants.
The function-gap mechanism might also explain empirically observed fast convergence of trust-region methods on ill-conditioned machine-learning problems, though the exact-solve assumption limits direct transfer.
A testable extension is whether the UniCAT modification retains its three-regime universality under a more permissive inexactness condition than exact solves; the paper notes this as an open direction.

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

Here's my take. The core idea is real: Lemma 7's function-gap model-decrease estimate is the right lever to get Hölder-adaptive convex complexity for trust-region methods, and the abstract Theorem 8 cleanly isolates what you need. I checked the main chain for the UniCAT variant and it looks coherent — Lemma 19 works, and the nonconvex bound with the ψ_k measure is plausible and clearly disclosed. The paper is also honest about its limitations: Variant I needs near-exact subproblem solves, Variant II exact solves, and only first-order stationarity is treated. Those are disclosed and don't by themselves undermine the claims.

The problem is Lemma 12. The claim is that small radii are very successful for the basic method, and the proof needs 2(1+τ)C_fD β_τ^{1+ν} ≤ 1−η_2. But Eq. (49) defines β_τ with a square in the denominator: ((1−η_2)/((1+τ)^2 C_fD))^α. Plugging that in gives exactly 2(1−η_2)/(1+τ), which is larger than 1−η_2 whenever τ<1. So the inequality asserted at the end of the lemma doesn't follow. For τ=0 and η_2=0.1, it's 1.8 vs 0.9. This is exactly the range the algorithm allows. The rest of the proof structure is right, and replacing (1+τ)^2 in Eq. (49) by 2(1+τ) repairs the argument — but the submitted manuscript does not rigorously prove Theorem 13 as stated.

That's load-bearing because Theorem 13 is one of the two advertised headline results. It doesn't sink the paper's overall thesis — the same mechanism works for UniCAT, and the basic-method result is almost certainly true with a corrected constant — but it needs to be fixed before the result can be taken as established.

Worth a serious referee? Yes. The idea is good, the writing is careful, and the flaw looks like a genuine typo rather than a structural gap. A referee should push for the corrected constant and re-verification of the τ<1 range. I'd bring it to the reading group, and I'd cite it once the corrected version is out.

Referee Report

2 major / 5 minor

Summary. The paper studies two quadratic trust-region variants: the classical ratio-test method with inexact subproblem solves (Variant I) and an exact-solve variant with a modified acceptance ratio (UniCAT, Variant II). Under ν-Hölder-continuous Hessians and without knowledge of ν or the Hölder constant, it claims the following: Variant I achieves O(ε^{-1/(1+ν)}) for convex ε-approximate minimization, retains the classical O(ε^{-2}) nonconvex first-order bound, and recovers local Newton-type rates; Variant II achieves the Hölder-adaptive nonconvex first-order bound O(ε^{-(2+ν)/(1+ν)}), the universal convex bound O(ε^{-1/(1+ν)}), and local Q-superlinear/order-(1+ν) convergence. The convex analysis is organized around a function-gap model-decrease property (MD_α) that is isolated as the key new mechanism.

Significance. If the results are correct, this is a valuable contribution: it shows that the classical trust-region mechanism is automatically adaptive to unknown Hölder smoothness across convex, nonconvex, and local regimes, without gradient/cubic regularization or gradient-dependent radii. The function-gap model-decrease estimate in Lemma 7 is a clean and reusable abstraction, and the paper is generally self-contained, carefully separating assumptions from conclusions and explicitly disclosing limitations (exact subproblem solves for Variant II, the strong inexactness condition for Variant I, and the use of ψ_k rather than ‖∇f(x_k)‖ in the nonconvex UniCAT bound). The abstract theorem framework in Section 4 is well designed. The main reservation is a localized but load-bearing proof error in Lemma 12 that affects the first main convex claim as submitted.

major comments (2)

[§6.1, Theorem 18 and Remark 8] The definition of βτ in Eq. (49) does not imply the claimed conclusion for τ<1. The proof derives |1−ρ_k| ≤ 2(1+τ)C_fD βτ^{1+ν}. With βτ as defined, this is at most 2(1−η2)/(1+τ), which exceeds 1−η2 whenever τ<1. For example, τ=0, η2=0.1, C_f=D=1 gives 1.8 > 0.9. Thus Lemma 12 is not established for the allowed range τ_k∈[0,\barτ) when \barτ<1, including the exact-solve case τ=0, and Theorem 13's universal convex claim for Variant I is therefore unproved as submitted. The likely fix is to replace (1+τ)^2 in (49) by 2(1+τ) (or equivalently to define βτ with the factor 2(1+τ) in the denominator); the rest of the convex proof appears sound. This is load-bearing for the paper's first main theorem and must be corrected or the theorem restated with a restricted inexactness range.
The nonconvex UniCAT guarantee is stated for ψ_k = min{‖∇f(x_k)‖, ‖∇f(x_k+s_k)‖}, not for the gradient at the accepted iterate. Remark 8 discloses this, but the abstract and introduction phrase the result as a bound on 'computing an ε-first-order stationary point.' This is a weaker output convention than the standard one, since the certificate may be at a rejected trial point and no explicit output-selection rule is given. The paper should state this convention prominently in the abstract/introduction, or add a simple rule such as returning the best of the iterates and trial points, so that the claimed nonconvex complexity is not overstated.

minor comments (5)

[§3.4, Lemma 6] The global parameters are stated as 0<η1≤η2<1, but for Variant II the analysis later requires η2<1/(1+ϑ) or η2<1/(1+ϑ/2). State these restrictions near Algorithm 1 to avoid an apparent inconsistency.
In the proof, the denominator |log(γ2)| is written inconsistently (once as logγ2 and once as |logγ2|). This is cosmetic but should be cleaned up.
After fixing the constant in Lemma 12, please re-check the related constants in Theorems 8 and 13 so that the dependencies on \barτ are consistent throughout.
The local convergence proof is standard but somewhat long; it could be streamlined by citing the classical Newton analysis more explicitly and moving some estimates to an appendix.
The phrase 'function-gap model-decrease estimate that appears to be new' is useful, but the relationship to the related estimate in [CGT12] should be spelled out more precisely in the introduction or in Remark 3.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the universal complexity proofs are self-contained and do not reduce to algorithmic inputs or self-citations.

full rationale

Walking the derivation chain, the paper's central claims are not circular. Variant I's convex result is reduced to the abstract Theorem 8, whose hypotheses — relative model decrease (Eq. (38)) and 'small radii are very successful' (Eq. (39)) — are verified from the standing assumptions via Lemma 7 (the MDα function-gap estimate) and Lemma 12, rather than assumed as the target rate. The MDα inequality is derived from convexity, the Hölder Taylor remainder (17), and the explicit constant β in (32); β is an analysis constant chosen to absorb other constants, and the algorithm is never given β, ν, Lν, D, or F0. The rate O(ε^{-1/(1+ν)}) emerges through the algebra α = 1/(1+ν) in Lemma 10, not by inserting the desired conclusion into the assumptions. Variant II similarly obtains a βCAT via Lemma 19 and then applies the same abstract theorem; its nonconvex bound follows from Lemmas 16–18 and the radius-update mechanism. Local convergence results are standard Newton-type arguments and are not used to force the global rates. The paper explicitly discloses its limitations — exact subproblem solves for Variant II and the strong relative-decrease condition (11) for Variant I — but these are structural assumptions, not hidden inputs or fitted predictions. The only same-author citation, [ADS26] in Remark 7, is a related-work remark and is not load-bearing. A possible numerical mismatch in Lemma 12's βτ constant for τ < 1 would be a proof defect, not circularity, because βτ is an analysis parameter rather than an algorithmic input or a fitted quantity renamed as a prediction.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters are fitted to data; algorithmic hyperparameters (η1, η2, γ1–γ3, ϑ, τ̄) are user-tunable and only subject to inequality constraints, and they do not encode ν or L_ν. No new entities are postulated. All axioms are either standard results or standard regularity conditions.

assumptions (5)

domain assumption Global (L_ν,ν)-Hölder continuity of the Hessian (Assumption A2).
Used in Lemma 1 to bound Taylor remainders as C_f‖s‖^{2+ν} and C_g‖s‖^{1+ν}, and throughout all theorems.
domain assumption Convexity of f and finiteness of initial distance D (Assumptions A4–A5).
Needed for the function-gap projection argument in Lemma 7 and for f(x_k+s) ≤ f(x_k)-tF_k.
domain assumption Uniform boundedness of Hessians (Assumption A3) for the classical nonconvex O(ε^{-2}) result.
Invoked in Theorem 11 to obtain Lipschitz continuity of the gradient via the mean value theorem.
domain assumption Strong convexity / nonsingular Hessian at x* for local rates (Theorems 14 and 21).
Local results assume ∇²f(x*)≻0 and x_k→x*; gives µ-strong convexity of models near x*.
standard math Classical trust-region tools: Cauchy decrease (Lemma 3), Lagrange characterization (Theorem 2), radius-counting bound (Lemma 6).
Imported from CGT22 and NW06; used in Lemmas 4, 5, 16, 17, 18.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the Universality of Simple Trust-Region Algorithms." pith.science (2026). https://pith.science/paper/LRDHJNIC

@misc{pith2026260719647,
  author       = {Pith},
  title        = {Pith review of: On the Universality of Simple Trust-Region Algorithms},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LRDHJNIC}},
  note         = {Machine review of arXiv:2607.19647}
}

abstract

We establish universal complexity guarantees for quadratic trust-region methods and identify a common mechanism underlying their universal behavior under convexity, based on a function-gap model-decrease estimate that appears to be new in the trust-region literature. First, we prove that the basic trust-region method with inexact subproblem solves is universal under convexity. Under a $\nu$-H\"older-continuous Hessian, it attains the global complexity bound $\mathcal{O}(\varepsilon^{-1/(1+\nu)})$ for computing an $\varepsilon$-approximate minimizer, without knowledge of $\nu\in[0,1]$ or the corresponding H\"older constant. In the nonconvex regime, the method retains the classical $\mathcal{O}(\varepsilon^{-2})$ first-order complexity bound under the usual additional bounded-Hessian assumption. With suitably vanishing inexactness, it also recovers Q-superlinear local convergence for $\nu=0$ and convergence of order $1+\nu$ for $\nu\in(0,1]$. Second, we show that the same convex-universal mechanism applies to a trust-region variant with exact subproblem solves and a simple modification of the acceptance ratio. This variant is universal simultaneously in the nonconvex, convex, and local regimes: it attains the optimal nonconvex first-order complexity $\mathcal{O}(\varepsilon^{-(2+\nu)/(1+\nu)})$, while preserving the universal convex complexity and the local Newton rates. These guarantees require no knowledge of $\nu$ or its H\"older constant. Both methods use the usual quadratic trust-region model and the classical radius-update mechanism, without gradient-dependent radii or model modifications such as cubic, gradient, or tensor regularization. The results show that the trust-region mechanism is inherently adaptive across nonconvex, convex, and locally strongly convex regimes, providing further theoretical support for the practical success of trust-region methods.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

65 extracted references · 9 canonical work pages

[1]

Dennis, Jr., J. E. and Mor. A Characterization of Superlinear Convergence and Its Application to. Mathematics of Computation , volume =. 1974 , doi =

1974
[2]

, title =

Nesterov, Yurii and Polyak, Boris T. , title =. Mathematical Programming , volume =. 2006 , doi =

2006
[3]

SIAM Journal on Optimization , volume =

Doikov, Nikita and Mishchenko, Konstantin and Nesterov, Yurii , title =. SIAM Journal on Optimization , volume =. 2024 , doi =

2024
[4]

SIAM Journal on Optimization , volume =

Grapiglia, Geovani Nunes and Nesterov, Yurii , title =. SIAM Journal on Optimization , volume =. 2020 , doi =

2020
[5]

Kelley, C. T. , title =. 1995 , doi =

1995
[6]

SIAM Journal on Imaging Sciences , volume =

Wang, Shida and Fadili, Jalal and Ochs, Peter , title =. SIAM Journal on Imaging Sciences , volume =. 2025 , doi =

2025
[7]

Global non-asymptotic super-linear convergence rates of regularized proximal quasi-Newton methods on non-smooth composite problems

Wang, Shida and Fadili, Jalal and Ochs, Peter , title =. 2024 , eprint =. doi:10.48550/arXiv.2410.11676 , url =

work page Pith review arXiv doi:10.48550/arxiv.2410.11676 2024
[8]

Advances in Neural Information Processing Systems , volume =

Doikov, Nikita and Nesterov, Yurii , title =. Advances in Neural Information Processing Systems , volume =

Show all 65 references

[9]

, title =

Nocedal, Jorge and Wright, Stephen J. , title =
[10]

2025 , eprint =

Wang, Shida and Fadili, Jalal and Ochs, Peter , title =. 2025 , eprint =. doi:10.48550/arXiv.2506.00521 , url =

2025 doi arXiv
[11]

Cartis, Coralia and Gould, Nicholas I. M. and Toint, Philippe L. , title =. Mathematical Programming , volume =. 2011 , doi =

2011
[12]

Mathematical Programming , volume =

Nesterov, Yurii , title =. Mathematical Programming , volume =. 2008 , doi =

2008
[13]

Grapiglia, G. N. and Nesterov, Yurii , title =. SIAM Journal on Optimization , volume =. 2017 , doi =

2017
[14]

SIAM Journal on Optimization , volume =

Mishchenko, Konstantin , title =. SIAM Journal on Optimization , volume =. 2023 , doi =

2023
[15]

Mathematical Programming , volume =

Doikov, Nikita and Nesterov, Yurii , title =. Mathematical Programming , volume =. 2024 , doi =

2024
[16]

2021 , isbn =

Ern, Alexandre and Guermond, Jean-Luc , title =. 2021 , isbn =

2021
[17]

Mathematical Programming , volume =

Rebjock, Quentin and Boumal, Nicolas , title =. Mathematical Programming , volume =. 2025 , doi =

2025
[18]

Mathematical Programming , volume =

Ouyang, Wenqing and Milzarek, Andre , title =. Mathematical Programming , volume =. 2025 , doi =

2025
[19]

Journal of Scientific Computing , volume =

Zhang, Chenyu and Xiao, Rufeng and Huang, Wen and Jiang, Rujun , title =. Journal of Scientific Computing , volume =. 2024 , doi =

2024
[20]

Journal of Optimization Theory and Applications , volume =

Jiang, Houyuan and Qi, Liqun , title =. Journal of Optimization Theory and Applications , volume =
[21]

SIAM Journal on Optimization , volume =

Qi, Liqun , title =. SIAM Journal on Optimization , volume =. 1995 , doi =

1995
[22]

Cartis, Coralia and Gould, Nicholas I. M. and Toint, Philippe L. , title =
[23]

and Gould, Nicholas I

Conn, Andrew R. and Gould, Nicholas I. M. and Toint, Philippe L. , title =
[24]

and Kouri, Drew P

Baraldi, Robert J. and Kouri, Drew P. , title =. Mathematical Programming , volume =. 2023 , doi =

2023
[25]

and Kouri, Drew P

Baraldi, Robert J. and Kouri, Drew P. , title =. Optimization Letters , volume =. 2024 , doi =

2024
[26]

and Funke, Simon W

Schwedes, Tobias and Ham, David A. and Funke, Simon W. and Piggott, Matthew D. , title =. 2017 , doi =

2017
[27]

Alphonse, Amal and Dvurechensky, Pavel and Papadopoulos, Ioannis P. A. and Sirotenko, Clemens , title =. 2025 , url =

2025
[28]

2022 , eprint =

Kanzow, Christian and Lechner, Theresa , title =. 2022 , eprint =. doi:10.48550/arXiv.2210.07644 , url =

2022 doi arXiv
[29]

Computational Optimization and Applications , volume =

Liu, Ruyu and Pan, Shaohua and Wu, Yuqia and Yang, Xiaoqi , title =. Computational Optimization and Applications , volume =
[30]

Computational Optimization and Applications , volume =

vom Dahl, Simeon and Kanzow, Christian , title =. Computational Optimization and Applications , volume =
[31]

and Kouri, Drew P

Baraldi, Robert J. and Kouri, Drew P. , title =. 2023 , url =

2023
[32]

and Yuan, Y

Grapiglia, Geovani Nunes and Yuan, J. and Yuan, Y. , title =. Optimization Methods and Software , volume =
[33]

Journal of Scientific Computing , volume =

Jiang, Yuntian and He, Chang and Zhang, Chuwen and Ge, Dongdong and Jiang, Bo and Ye, Yinyu , title =. Journal of Scientific Computing , volume =. 2026 , doi =

2026
[34]

and Kouri, Drew P

Baraldi, Robert J. and Kouri, Drew P. , title =. Computational Optimization and Applications , volume =. 2025 , doi =

2025
[35]

and Lubberts, Zachary and Robinson, Daniel P

Curtis, Frank E. and Lubberts, Zachary and Robinson, Daniel P. , title =. Optimization Letters , volume =
[36]

and Robinson, Daniel P

Curtis, Frank E. and Robinson, Daniel P. , title =. Mathematical Programming , volume =
[37]

A multilevel proximal trust-region method for nonsmooth optimization with applications , year =

Baraldi, Robert and Hinterm. A multilevel proximal trust-region method for nonsmooth optimization with applications , year =. doi:10.48550/arXiv.2512.00538 , url =. 2512.00538 , archivePrefix =

doi
[38]

Cartis, Coralia and Gould, Nicholas I. M. and Toint, Philippe L. , title =. Optimization Methods and Software , volume =
[39]

Alphonse, Amal and Dvurechensky, Pavel and Papadopoulos, Ioannis P. A. and Sirotenko, Clemens , title =. 2025 , eprint =. doi:10.48550/arXiv.2508.16468 , url =

2025 doi arXiv
[40]

2026 , eprint =

Alphonse, Amal and Dvurechensky, Pavel and Sirotenko, Clemens , title =. 2026 , eprint =. doi:10.48550/arXiv.2602.08069 , url =

2026 doi
[41]

2024 , eprint =

Hamad, Fadi and Hinder, Oliver , title =. 2024 , eprint =. doi:10.48550/arXiv.2412.02079 , url =

2024 doi arXiv
[42]

INFORMS Journal on Optimization , volume =

Blanchet, Jose and Cartis, Coralia and Menickelly, Matt and Scheinberg, Katya , title =. INFORMS Journal on Optimization , volume =
[43]

Mathematical Programming , year =

Doikov, Nikita , title =. Mathematical Programming , year =
[44]

Neural Networks , volume =

Zhang, Zhen and Che, Keqin and Yang, Shaofu and Xu, Wenying , title =. Neural Networks , volume =. 2024 , doi =

2024
[45]

Computing a Trust-Region Step , journal =

Mor. Computing a Trust-Region Step , journal =. 1983 , doi =

1983
[46]

Gould, Nicholas I. M. and Lucidi, Stefano and Roma, Massimo and Toint, Philippe L. , title =. SIAM Journal on Optimization , volume =. 1999 , doi =

1999
[47]

Optimization Methods and Software , volume =

Lenders, Felix and Kirches, Christian and Potschka, Andreas , title =. Optimization Methods and Software , volume =. 2018 , doi =

2018
[48]

Gould, Nicholas I. M. and Robinson, Daniel P. and Thorne, H. Sue , title =. Mathematical Programming Computation , volume =. 2010 , doi =

2010
[49]

Cartis, Coralia and Gould, Nicholas I. M. and Toint, Philippe L. , title =. SIAM Journal on Optimization , volume =
[50]

Mathematical Programming , volume =

Doikov, Nikita and Nesterov, Yurii , title =. Mathematical Programming , volume =. 2023 , doi =

2023
[51]

Mathematical Programming , volume =

Nesterov, Yurii , title =. Mathematical Programming , volume =
[52]

Mathematical Programming , year =

Diouane, Youssef and Laghdaf Habiboullah, Mohamed and Orban, Dominique , title =. Mathematical Programming , year =
[53]

, title =

Toint, Philippe L. , title =. Sparse Matrices and Their Uses , editor =
[54]

SIAM Journal on Numerical Analysis , volume =

Steihaug, Trond , title =. SIAM Journal on Numerical Analysis , volume =. 1983 , doi =

1983
[55]

Mathematical Programming , volume =

Yuan, Ya-xiang , title =. Mathematical Programming , volume =. 2000 , doi =

2000
[56]

and Robinson, Daniel P

Curtis, Frank E. and Robinson, Daniel P. and Samadi, Mohammadreza , title =. Mathematical Programming , volume =. 2017 , doi =

2017
[57]

2018 , isbn =

Nesterov, Yurii , title =. 2018 , isbn =

2018
[58]

Advances in Neural Information Processing Systems , volume =

Hamad, Fadi and Hinder, Oliver , title =. Advances in Neural Information Processing Systems , volume =
[59]

2026 , eprint =

Zeng, Ziyang and Zhang, Junyu and He, Chuan , title =. 2026 , eprint =. doi:10.48550/arXiv.2604.02763 , url =

2026 doi arXiv
[60]

Advances in Neural Information Processing Systems , volume =

Zhou, Yuhao and Xu, Jintao and Li, Bingrui and Bao, Chenglong and Ding, Chao and Zhu, Jun , title =. Advances in Neural Information Processing Systems , volume =
[61]

Royer, Cl. A. Mathematical Programming , volume =
[62]

, title =

Gratton, Serge and Sartenaer, Annick and Toint, Philippe L. , title =. SIAM Journal on Optimization , volume =
[63]

2025 , eprint =

Jiang, Yuntian and Zhang, Chuwen and Jiang, Bo and Ye, Yinyu , title =. 2025 , eprint =. doi:10.48550/arXiv.2511.00680 , url =

2025 doi arXiv
[64]

SIAM Journal on Optimization , year =

Jin, Billy and Scheinberg, Katya and Xie, Miaolan , title =. SIAM Journal on Optimization , year =
[65]

Cartis, Coralia and Gould, Nicholas I. M. and Toint, Philippe L. , title =. Proceedings of the International Congress of Mathematicians (ICM 2018): Invited Lectures , editor =. 2018 , isbn =. 1709.07180 , archivePrefix =

2018 arXiv

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

Dennis, Jr., J. E. and Mor. A Characterization of Superlinear Convergence and Its Application to. Mathematics of Computation , volume =. 1974 , doi =

1974

[2] [2]

, title =

Nesterov, Yurii and Polyak, Boris T. , title =. Mathematical Programming , volume =. 2006 , doi =

2006

[3] [3]

SIAM Journal on Optimization , volume =

Doikov, Nikita and Mishchenko, Konstantin and Nesterov, Yurii , title =. SIAM Journal on Optimization , volume =. 2024 , doi =

2024

[4] [4]

SIAM Journal on Optimization , volume =

Grapiglia, Geovani Nunes and Nesterov, Yurii , title =. SIAM Journal on Optimization , volume =. 2020 , doi =

2020

[5] [5]

Kelley, C. T. , title =. 1995 , doi =

1995

[6] [6]

SIAM Journal on Imaging Sciences , volume =

Wang, Shida and Fadili, Jalal and Ochs, Peter , title =. SIAM Journal on Imaging Sciences , volume =. 2025 , doi =

2025

[7] [7]

Global non-asymptotic super-linear convergence rates of regularized proximal quasi-Newton methods on non-smooth composite problems

Wang, Shida and Fadili, Jalal and Ochs, Peter , title =. 2024 , eprint =. doi:10.48550/arXiv.2410.11676 , url =

work page Pith review arXiv doi:10.48550/arxiv.2410.11676 2024

[8] [8]

Advances in Neural Information Processing Systems , volume =

Doikov, Nikita and Nesterov, Yurii , title =. Advances in Neural Information Processing Systems , volume =