Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

A Frank-Wolfe Algorithm for Oracle-based Robust Optimization

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper proves that a smoothed Frank-Wolfe method solves oracle-based robust optimization in O(1/epsilon^2) oracle calls, matching the best known bound and giving the first explicit oracle count for min-max-min problems.

desk verdict The fixed-smoothing Frank-Wolfe theorem is sound and the paper's core idea has value, but the adaptive-smoothing Proposition 4 is false, which undercuts a chunk of the experiments. read the letter →

arxiv 2411.19848 v2 pith:33LT7WJB submitted 2024-11-29 math.OC

Mathieu Besançon , Jannis Kurtz This is my paper · ORCID
classification math.OC MSC 90C2590C47
keywords robustoptimizationoracle-basedalgorithmFrank-Wolfesmoothingmin-max-minoraclecomplexitylinearminimizationprojection
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

This paper tries to establish that robust min-max problems of the form $\min_{x \in X} \max_{c \in U} c^{\top} x$, where the feasible set $X$ is available only through a linear minimization oracle, can be solved by Frank-Wolfe on a smoothed version of the objective. The smoothing adds a quadratic penalty $-\frac{\mu}{2}\|c-c_0\|^2$ to the inner maximization, making the piecewise-linear function differentiable, with a gradient that costs one Euclidean projection onto the uncertainty set $U$. The main theorem says that with $\mu = \varepsilon/M^2$, after $t \ge 4D^2M^2/\varepsilon^2$ Frank-Wolfe iterations the primal gap is at most $\varepsilon$, giving an oracle-polynomial algorithm (polynomial in the problem data when each oracle call is cheap) and the first explicit oracle-call bound for min-max-min robust optimization. A sympathetic reader would care because the result matches the best known complexity while working in the high-dimensional, large-uncertainty regime where earlier constraint-generation methods stall.

What carries the argument

The load-bearing object is the smoothed robust objective $f_\mu(x) = \max_{c\in U}\{c^{\top} x - \frac{\mu}{2}\|c-c_0\|^2\}$ with $\mu=\varepsilon/M^2$. The quadratic penalty is strongly concave in $c$, which makes $f_\mu$ differentiable and its gradient $1/\mu$-Lipschitz; evaluating the gradient is equivalent to projecting $c_0 + x/\mu$ onto $U$. Each Frank-Wolfe iteration therefore alternates a projection oracle on $U$ and a linear minimization oracle on $X$. The proof of Theorem 1 splits the true gap as $f(x_t)-f(x^*) \le [f(x_t)-f_\mu(x_t)] + [f_\mu(x_t)-f_\mu(x^*_\mu)]$, bounding the first term by $\varepsilon/2$ through the choice of $\mu$ and the second by the classical Frank-Wolfe convergence rate $2D^2/(\mu t)$; the schedule $\mu = \varepsilon/M^2$ and $t \ge 4D^2M^2/\varepsilon^2$ then force each term below $\varepsilon/2$.

What would settle it

Run Algorithm 1 on a small instance such as $X=[0,1]$ and $U=[-1,1]$, compute the exact optimum of (RO), and check at $t=\lceil 4D^2M^2/\varepsilon^2\rceil$ iterations for a range of $\varepsilon$ values whether $f(x_t)-f(x^*) \le \varepsilon$; a single violation would disprove Theorem 1.

Watch

Extended reading notes

Core claim

The central claim is that the nonsmooth problem (RO) can be approximated to accuracy $\varepsilon$ by a single smoothing parameter $\mu = \varepsilon/M^2$: the smoothed objective $f_\mu(x)=\max_{c\in U}\{c^{\top} x - \frac{\mu}{2}\|c-c_0\|^2\}$ differs from $f(x)=\max_{c\in U} c^{\top} x$ by at most $\varepsilon/2$ at every $x$, while the Frank-Wolfe gap on $f_\mu$ after $t$ steps is at most $2D^2/(\mu t)$. Balancing the two $\varepsilon/2$ terms gives the oracle bound $4D^2M^2/\varepsilon^2$. The proof works for any Frank-Wolfe variant, so the authors run a lazified blended pairwise conditional gradient in practice; it also extends, through a standard composite convex minimization argument, to an adaptive schedule $\mu_t = 2D/(M_{\max}\sqrt{t+1})$ with $O(1/\sqrt{T})$ convergence. A direct corollary is the first explicit oracle-call bound $4nM^2/\varepsilon^2$ for min-max-min robust combinatorial optimization when $U$ is the convex hull of scenarios.

Load-bearing premise

The entire complexity promise rests on the assumption that each linear minimization oracle call over $X$ and each Euclidean projection oracle call over $U$ takes constant time, so the bound is useful only when projection onto the uncertainty set is itself a cheap subroutine.

Editorial extensions

If this is right

  • Whenever the linear minimization oracle over $X$ and the projection oracle over $U$ are both polynomial-time subroutines, problem (RO) is solvable to accuracy $\varepsilon$ in at most $4D^2M^2/\varepsilon^2$ oracle calls, matching the previous best bound in the same model.
  • For min-max-min robust combinatorial optimization with $k \ge n+1$ solutions, the same result supplies the first explicit oracle-call bound, $4nM^2/\varepsilon^2$.
  • Because the convergence proof does not depend on the specific Frank-Wolfe variant, modern implementations such as lazified blended pairwise conditional gradient can be used without losing the theoretical guarantee, which is what makes the method competitive in the experiments.
  • The adaptive smoothing schedule $\mu_t = 2D/(M_{\max}\sqrt{t+1})$ gives an $O(1/\sqrt{T})$ primal-gap bound without knowing $\varepsilon$ in advance, and the analysis tolerates inexact linear minimization oracles with additive or multiplicative error.

Reading between the lines

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

  • The paper leaves implicit that the same $\varepsilon/2$ splitting should transfer to other projection-free variants such as away-step, pairwise, or fully-corrective Frank-Wolfe, potentially improving the constant $4$ in the oracle bound.
  • A testable extension is to replace the exact projection oracle with an approximate projection: if the approximation error propagates linearly through the bound, the $O(1/\varepsilon^2)$ rate should survive with a slightly larger constant, broadening the class of uncertainty sets.
  • The explicit $4nM^2/\varepsilon^2$ bound for min-max-min problems suggests that branch-and-bound solvers for two-stage robust optimization could certify node lower bounds with a predictable oracle budget, a connection the paper mentions only 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.

Referee Report

3 major / 6 minor

Summary. This paper studies the oracle model for robust optimization problems of the form min_{x in X} max_{c in U} c^T x, where X is accessed only through a linear minimization oracle and U through a projection oracle. It introduces a smoothed Frank-Wolfe method (Algorithm 1) with fixed smoothing parameter mu=epsilon/M^2 and proves in Theorem 1 an O(D^2 M^2 / epsilon^2) iteration bound; Corollaries 2 and 3 translate this into oracle-complexity bounds, including the first explicit bound for min-max-min robust combinatorial optimization. The paper also presents an adaptive-smoothing variant (Algorithm 2), claims an O(1/sqrt(T)) convergence in Proposition 4, and reports experiments on robust spanning tree and TSP instances comparing constraint generation, fixed smoothing, adaptive smoothing, and a convex-hull-accelerated variant.

Significance. The fixed-smoothing part of the paper is clean and mostly self-contained: the proof of Theorem 1 balances the smoothing bias mu M^2 / 2 against the Frank-Wolfe gap 2 D^2 / (mu t), the constants are specified, and no fitted parameters enter. This gives a parameter-free oracle-complexity bound that matches the best known order and provides the first explicit oracle-call count for min-max-min problems. The paper also makes its code publicly available. The main weakness is that the adaptive-smoothing Proposition 4, on which Algorithm 2 and the A-FW experiments rest, is false as stated; a one-dimensional example violates its bound. The contribution is therefore currently conditional on the fixed-smoothing theorems, and the adaptive claims need repair before the paper can be accepted.

major comments (3)
  1. [2.2, Proposition 4] Proposition 4 is false as stated. Consider X=[0,1], U=[-1,1], c0=-1, so D=1, Mmax=1, f(x)=x, and f*=0. Running Algorithm 2 with mu_t=2/sqrt(t+1), gamma_t=2/(t+1), and x0=0 gives: x1=1 at t=1; x2=1 at t=2 (gradient c=-1+1/mu2 approximately -0.134); at t=3 the gradient at x=1 is zero and the LMO tie-break v=1 gives x3=1 (the alternative v=0 leads to an even worse final value); at t=4 the gradient is c=-1+1/mu4 approximately 0.118, so the LMO returns v=0 and x4=0.6. Thus f(x4)-f*=0.6, while the proposition promises D Mmax / (2 sqrt(T)) = 0.25. This is not an oracle-model caveat or a tie-break artifact; it is a violation of the stated bound by the algorithm's own update rule. The proof by direct reference to Yurtsever et al. [2018, Theorem 3.2] is therefore not valid for Algorithm 2 as written, since the hypotheses and step-size/smoothing conditions of that theorem are not verified. Because Proposition 4 is the only theoretical support for Algorithm 2 and for the A-FW experiments, the adaptive-smoothing convergence claim needs to be corrected or removed.
  2. [2.2, Algorithm 2] Algorithm 2 is underspecified in a way that is directly relevant to Proposition 4: the Data line lists D and L_U, but L_U is never defined, while Proposition 4 defines the schedule with Mmax (max_{c in U} ||c||). If L_U is intended to be Mmax, this should be stated; otherwise the schedule in the algorithm box is not the schedule analyzed in the proposition. The notation M, Mmax, D, and Dmax should also be kept distinct, since D is the diameter of X, Dmax is the maximum norm of a point in X, M is the diameter of U, and Mmax is the maximum norm of a point in U.
  3. [3, Experiments] The experimental section presents A-FW as an implementation of Algorithm 2 and uses it to draw conclusions about adaptive smoothing (e.g., better performance on TSP and larger uncertainty sets). Since Proposition 4 is false, A-FW has no proven convergence guarantee in the current manuscript; the empirical observations may still be valid as heuristics, but the claims that adaptive smoothing is a theoretically supported instance of the proposed framework should be revised. The fixed-smoothing FW and FW-convhull results remain informative, but the paper should make clear which experimental conclusions depend only on Theorem 1 and which depend on the unproven adaptive variant.
minor comments (6)
  1. [2.2, Eq. (9)] The Frank-Wolfe bound should be stated with the iteration index matching the step-size rule gamma_t=2/(t+1) in Algorithm 1; as written the denominator t+3 and the later replacement by t are safe, but the indexing should be made explicit.
  2. [Table 1 and Corollary 2] 'Oracle calls' is used for a bound that counts each iteration, although Algorithm 1 makes one LMO call and one projection call per iteration; please state the convention (LMO calls only, or both oracles with the factor 2).
  3. [Abstract] The 'best known oracle complexity' claim should be qualified: Ben-Tal et al. [2015] already gives a bound of order D_max^2 M^2 / epsilon^2 in the same oracle model, and since D^2 can be up to 4 D_max^2, the present constant is not uniformly better.
  4. [Corollary 3] Corollary 3 should state explicitly that D^2 is at most n when X=conv(Z) and Z is a subset of {0,1}^n; otherwise the appearance of 4 n M^2 / epsilon^2 needs a derivation.
  5. [3, Experiments] The experiments use the BPCG variant with lazification, but the theoretical analysis is for plain Frank-Wolfe with the step-size rule in Algorithm 1; the text should explain how the convergence guarantee transfers to the implemented variant, or label the variant as a heuristic.
  6. [Table 1] There is a typo: 'sknapsack constraints' should be 'knapsack constraints'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 1 is a standard Frank-Wolfe smoothing bound with prescribed parameters, and the self-citations are not load-bearing.

full rationale

The central derivation is self-contained. The smoothed objective in (7), the gradient formula in (8), and the proof of Theorem 1 chain the standard Frank-Wolfe convergence rate f_mu(x_t)-f_mu(x*_mu) <= 2D^2/(mu(t+3)) to the original robust gap using the explicit choice mu = eps/M^2 and t >= 4D^2M^2/eps^2. No parameter is fitted to data and the bound is not defined in terms of the claimed output. The step sizes and smoothing parameter are prescribed by the accuracy analysis rather than calibrated to the target. Corollary 3 transfers this bound to min-max-min robust optimization through the external equivalence result of Buchheim and Kurtz (2017), but that citation supplies only the problem reduction, not the oracle-complexity bound itself. Proposition 4 invokes Yurtsever et al. Theorem 3.2 as an external convergence theorem; even if the application of that theorem were later found to be mathematically incorrect, that would be a correctness issue rather than circularity, since the proposition does not assume its own conclusion. The self-citations in the paper, namely Buchheim/Kurtz reductions and the FrankWolfe.jl implementation, are motivational or experimental tooling and do not carry the theoretical claim. The paper is therefore self-contained against external benchmarks for its oracle-complexity contribution.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No parameters are fitted to data: the smoothing parameter mu = eps / M^2 and the adaptive schedule mu_t = 2 D / (Mmax sqrt(t + 1)) are prescribed by the accuracy analysis, and the step size gamma_t = 2 / (t + 1) is the standard agnostic Frank-Wolfe choice. The paper relies on three external theorems for smoothing and Frank-Wolfe convergence, and on the oracle model.

assumptions (4)
  • standard math Frank-Wolfe primal convergence gap f_mu(x_t) - f_mu(x*_mu) <= 2 D^2 (1/mu) / (t + 3) for smooth convex objectives with agnostic step sizes.
    Invoked in the proof of Theorem 1 (Eq. 9) and attributed to Braun et al. 2022; not re-derived.
  • standard math Nesterov's smoothing theorem: the smoothed function f_mu(x) = max_{c in U} (c^T x - mu/2 ||c - c0||^2) has gradient given by the unique maximizer in (8) and is 1/mu-Lipschitz smooth.
    Used in Section 2.1 to justify applying Frank-Wolfe to f_mu; cited to Nesterov 2005, Theorem 1.
  • standard math Adaptive smoothing convergence result in Yurtsever et al. (2018, Theorem 3.2) applies with phi = 0, A = I, and g(x) = max_{c in U} c^T x, which is Mmax-Lipschitz.
    Proposition 4 is proved by direct reference to this theorem; the paper does not reproduce the conditions in full.
  • domain assumption The feasible set X and uncertainty set U are compact and convex, and both the LMO over X and the projection onto U can be evaluated, each call counting as constant runtime in the oracle model.
    Defined in Section 1 and used in Algorithm 1 and Theorem 1; without it the oracle-polynomial conclusion in Corollary 2 does not follow.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Frank-Wolfe Algorithm for Oracle-based Robust Optimization." pith.science (2026). https://pith.science/paper/33LT7WJB

@misc{pith2026241119848,
  author       = {Pith},
  title        = {Pith review of: A Frank-Wolfe Algorithm for Oracle-based Robust Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/33LT7WJB}},
  note         = {Machine review of arXiv:2411.19848}
}
read the original abstract

We tackle robust optimization problems under objective uncertainty in the oracle model, i.e., when the deterministic problem is solved by an oracle. The oracle-based setup is favorable in many situations, e.g., when a compact formulation of the feasible region is unknown or does not exist. We propose an iterative method based on a Frank-Wolfe type algorithm applied to a smoothed version of the piecewise linear objective function. Our approach bridges several previous efforts from the literature, attains the best known oracle complexity for the problem and performs better than state-of-the-art on high-dimensional problem instances, in particular for larger uncertainty sets.

Figures

Figures reproduced from arXiv: 2411.19848 by the authors.

Figure 1
Figure 1. Smoothing of the function f(x) = max −1≤c≤1 c ⊤x for different µ values. calculated in each iteration, while CGA only works with an optimal solution c ∗ ∈ U. Afterwards, both algorithm apply the LMO to compute the next vertex xt ∈ X . CGA only has to solve Problem (6) in the last iteration to obtain the final optimal solution. Finally, we highlight that SDA itself is equivalent to fully-corrective Frank-Wolfe (FCFW)… view at source ↗
Figure 2
Figure 2. Comparison of the constraint generation and FW-based algorithms on a robust spanning tree example. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Primal value against iteration for n = 300 for the ST problem for Γ ∈ {30, 60, 90}. For the spanning tree example, the LMO uses the Kruskal algorithm as implemented in Graphs.jl. We use the FrankWolfe.jl [Besanc¸on et al., 2022] implementation of the blended pairwise conditional gradient with lazification, meaning the algorithm only calls the LMO when it cannot perform sufficient progress over the current set of ver… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Comparison of the constraint generation and FW-based algorithms on a TSP example. [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

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

  1. First-Order Methods for Distributionally Robust Constrained Optimization

    math.OC 2026-07 conditional novelty 5.0 of 10

    Entropic smoothing plus momentum stochastic Frank–Wolfe yields a general first-order method for constrained Wasserstein DRO with convergence guarantees and better out-of-sample behavior than ERM on traffic assignment ...

Reference graph

Works this paper leans on

30 extracted references · 28 canonical work pages · cited by 1 Pith paper

  1. [1]

    A F rank- W olfe based algorithm for robust discrete optimization under uncertainty

    Chifaa Al Dahik, Zeina Al Masry, St \'e phane Chr \'e tien, Jean-Marc Nicod, and Landy Rabehasaina. A F rank- W olfe based algorithm for robust discrete optimization under uncertainty. In 2020 Prognostics and Health Management Conference (PHM-Besan c on) , pages 247--252. IEEE, 2020

    work page 2020

  2. [2]

    A note on the B ertsimas & S im algorithm for robust combinatorial optimization problems

    Eduardo \'A lvarez-Miranda, Ivana Ljubi \'c , and Paolo Toth. A note on the B ertsimas & S im algorithm for robust combinatorial optimization problems. 4OR, 11: 0 349--360, 2013

    work page 2013

  3. [3]

    Robust optimization

    Aharon Ben-Tal, Arkadi Nemirovski, and Laurent El Ghaoui. Robust optimization. 2009

    2009

  4. [4]

    Oracle-based robust optimization via online learning

    Aharon Ben-Tal, Elad Hazan, Tomer Koren, and Shie Mannor. Oracle-based robust optimization via online learning. Operations Research, 63 0 (3): 0 628--638, 2015

    work page 2015

  5. [5]

    Bertsimas and D

    D. Bertsimas and D. den Hertog. Robust and Adaptive Optimization. Dynamic Ideas LLC, 2022. ISBN 9781733788526

    work page 2022

  6. [6]

    Robust discrete optimization and network flows

    Dimitris Bertsimas and Melvyn Sim. Robust discrete optimization and network flows. Mathematical programming, 98 0 (1-3): 0 49--71, 2003

    work page 2003

  7. [7]

    FrankWolfe.jl : A high-performance and flexible toolbox for Frank-Wolfe algorithms and conditional gradients

    Mathieu Besan c on, Alejandro Carderera, and Sebastian Pokutta. FrankWolfe.jl : A high-performance and flexible toolbox for Frank-Wolfe algorithms and conditional gradients. INFORMS Journal on Computing, 2022

    work page 2022

  8. [8]

    An oracle-based framework for robust combinatorial optimization

    Enrico Bettiol, Christoph Buchheim, Marianna De Santis, and Francesco Rinaldi. An oracle-based framework for robust combinatorial optimization. Journal of Global Optimization, pages 1--25, 2023

    work page 2023

Show all 30 references
  1. [9]

    F rank- Wolfe and friends: a journey into projection-free first-order optimization methods

    Immanuel M Bomze, Francesco Rinaldi, and Damiano Zeffiro. F rank- Wolfe and friends: a journey into projection-free first-order optimization methods. 4OR, 19: 0 313--345, 2021

    2021

  2. [10]

    An oracle-based algorithm for robust planning of production routing problems in closed-loop supply chains of beverage glass bottles

    Ali Borumand, Ahmadreza Marandi, Ali S Nookabadi, and Z \"u mb \"u l Atan. An oracle-based algorithm for robust planning of production routing problems in closed-loop supply chains of beverage glass bottles. Omega, 122: 0 102939, 2024

    2024

  3. [11]

    Lazifying conditional gradient algorithms

    G \'a bor Braun, Sebastian Pokutta, and Daniel Zink. Lazifying conditional gradient algorithms. Journal of Machine Learning Research, 20 0 (71): 0 1--42, 2019

    2019

  4. [12]

    Conditional gradient methods

    G \'a bor Braun, Alejandro Carderera, Cyrille W Combettes, Hamed Hassani, Amin Karbasi, Aryan Mokhtari, and Sebastian Pokutta. Conditional gradient methods. arXiv preprint arXiv:2211.14103, 2022

    2022 arXiv

  5. [13]

    A note on the nonexistence of oracle-polynomial algorithms for robust combinatorial optimization

    Christoph Buchheim. A note on the nonexistence of oracle-polynomial algorithms for robust combinatorial optimization. Discrete Applied Mathematics, 285: 0 591--593, 2020

    2020

  6. [14]

    Min--max--min robust combinatorial optimization

    Christoph Buchheim and Jannis Kurtz. Min--max--min robust combinatorial optimization. Mathematical Programming, 163: 0 1--23, 2017

    2017

  7. [15]

    Robust combinatorial optimization under convex and discrete cost uncertainty

    Christoph Buchheim and Jannis Kurtz. Robust combinatorial optimization under convex and discrete cost uncertainty. EURO Journal on Computational Optimization, 6 0 (3): 0 211--238, 2018

    2018

  8. [16]

    A F rank-- Wolfe based branch-and-bound algorithm for mean-risk optimization

    Christoph Buchheim, Marianna De Santis, Francesco Rinaldi, and Long Trieu. A F rank-- Wolfe based branch-and-bound algorithm for mean-risk optimization. Journal of Global Optimization, 70: 0 625--644, 2018

    2018

  9. [17]

    An introduction to robust combinatorial optimization

    Marc Goerigk and Michael Hartisch. An introduction to robust combinatorial optimization. International Series in Operations Research and Management Science, 2024

    2024

  10. [18]

    Revisiting frank-wolfe: Projection-free sparse convex optimization

    Martin Jaggi. Revisiting frank-wolfe: Projection-free sparse convex optimization. In International conference on machine learning, pages 427--435. PMLR, 2013

    2013

  11. [19]

    Oracle-based algorithms for binary two-stage robust optimization

    Nicolas K \"a mmerling and Jannis Kurtz. Oracle-based algorithms for binary two-stage robust optimization. Computational Optimization and Applications, 77: 0 539--569, 2020

    2020

  12. [20]

    Robust discrete optimization and its applications, volume 14

    Panos Kouvelis and Gang Yu. Robust discrete optimization and its applications, volume 14. Springer Science & Business Media, 2013

    2013

  13. [21]

    A short note on the robust combinatorial optimization problems with cardinality constrained uncertainty

    Taehan Lee and Changhyun Kwon. A short note on the robust combinatorial optimization problems with cardinality constrained uncertainty. 4OR, 12: 0 373--378, 2014

    2014

  14. [22]

    JuMP 1.0: Recent improvements to a modeling language for mathematical optimization

    Miles Lubin, Oscar Dowson, Joaquim Dias Garcia, Joey Huchette, Beno \^ t Legat, and Juan Pablo Vielma. JuMP 1.0: Recent improvements to a modeling language for mathematical optimization. Mathematical Programming Computation, 15 0 (3): 0 581--589, 2023

    2023

  15. [23]

    GLPK ( GNU linear programming kit), 2008

    Andrew Makhorin. GLPK ( GNU linear programming kit), 2008. http://www.gnu.org/s/glpk/glpk.html

    2008

  16. [24]

    Smooth minimization of non-smooth functions

    Yurii Nesterov. Smooth minimization of non-smooth functions. Mathematical programming, 103: 0 127--152, 2005

    2005

  17. [25]

    Complexity bounds for primal-dual methods minimizing the model of objective function

    Yurii Nesterov. Complexity bounds for primal-dual methods minimizing the model of objective function. Mathematical Programming, 171 0 (1-2): 0 311--330, 2018

    2018

  18. [26]

    Algorithms for the continuous nonlinear resource allocation problem—new implementations and numerical studies

    Michael Patriksson and Christoffer Str \"o mberg. Algorithms for the continuous nonlinear resource allocation problem—new implementations and numerical studies. European Journal of Operational Research, 243 0 (3): 0 703--722, 2015

    2015

  19. [27]

    Robust combinatorial optimization with knapsack uncertainty

    Michael Poss. Robust combinatorial optimization with knapsack uncertainty. Discrete Optimization, 27: 0 88--102, 2018

    2018

  20. [28]

    Pairwise conditional gradients without swap steps and sparser kernel herding

    Kazuma K Tsuji, Ken'ichiro Tanaka, and Sebastian Pokutta. Pairwise conditional gradients without swap steps and sparser kernel herding. In International Conference on Machine Learning, pages 21864--21883. PMLR, 2022

    2022

  21. [29]

    A conditional gradient framework for composite convex minimization with applications to semidefinite programming

    Alp Yurtsever, Olivier Fercoq, Francesco Locatello, and Volkan Cevher. A conditional gradient framework for composite convex minimization with applications to semidefinite programming. In International Conference on Machine Learning, pages 5727--5736. PMLR, 2018

    2018

  22. [30]

    Limited memory Kelley 's method converges for composite convex and submodular objectives

    Song Zhou, Swati Gupta, and Madeleine Udell. Limited memory Kelley 's method converges for composite convex and submodular objectives. Advances in Neural Information Processing Systems, 31, 2018

    2018

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.