Pith. sign in

REVIEW 6 minor 117 references

Online Koml\'os converges to mean curvature flow

T0 review · 0 major / 6 minor · reviewed 2026-07-13 · grok-4.5

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

Pith's one-line read The large-T value of the online Komlós game is set by the extinction time of the unit cube under mean curvature flow.

desk verdict Clean continuum limit for the adaptive online Komlós game: value ~ √(T/(2τ)) with τ the extinction time under the natural curvature flow, plus matching √log m bounds for the cube. read the letter →

arxiv 2607.08943 v1 pith:BQ3X6QIG submitted 2026-07-09 math.CO cs.DMmath.AP

Nestor Guillen , Vladimir A. Kobzar This is my paper · ORCID
classification math.COcs.DMmath.AP MSC 05D4035K9349L2552A2091A05
keywords onlineKomlósgamevectorbalancingmeancurvatureflowviscositysolutionsBanaszczyktheoremdiscrepancytheoryextinctiontimelevel-setPDE
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 studies an online vector-balancing game that interpolates between the classic Komlós problem and fully online discrepancy. Paul chooses batches of n unit-ball vectors; Carol assigns signs; after T rounds the ℓ∞ norm of the cumulative sum is the payoff. For large T the leading term of the game value is exactly 1/√(2τ), where τ is the time at which the unit cube shrinks to a point under a curvature flow whose speed is the sum of its largest min(n,m-1) principal curvatures. When n≥m-1 the flow is ordinary mean curvature flow and the constant is of order √log m. The continuum limit is obtained by showing that a rescaled discrete value function converges, in the viscosity sense, to the level-set PDE of that curvature flow; the identification of the limiting Bellman operator uses Banaszczyk’s ℓ^{2} discrepancy theorem. The same asymptotic holds for any centrally symmetric convex body in place of the cube. The result turns a long-horizon combinatorial game into a geometric evolution problem and thereby supplies a new analytic window on vector balancing.

What carries the argument

The rescaled discrete value function w_δ(x,t)=δ v(x/δ,t/δ^{2}) of the online game, which converges locally uniformly to the unique viscosity solution of the level-set PDE for the relevant curvature flow; the limiting nonlinear operator is identified via Banaszczyk’s theorem.

What would settle it

Compute the exact extinction time of the unit cube under mean curvature flow for moderate dimension m and check whether K_T(m,m-1)/√T approaches 1/√(2τ) for large T; a persistent discrepancy would refute the claimed limit.

Watch

Extended reading notes

Core claim

For every centrally symmetric convex body E the asymptotic value of the online Komlós game satisfies lim T o∞ K_{n,T}(E)/√T = 1/√(2 τ_n(E)), where τ_n(E) is the extinction time of E under the flow whose normal velocity equals the sum of the largest min(n,m-1) principal curvatures. When n≥m-1 and E is the unit cube this constant is Θ(√log m).

Load-bearing premise

That the half-relaxed limits of the discrete value functions are viscosity sub- and supersolutions of the continuum curvature PDE, so that the comparison principle forces them to coincide.

Editorial extensions

If this is right

  • The dimensional growth of the large-T value is completely determined by the Gaussian width of the dual body and by the maximal dual basis vectors.
  • Any centrally symmetric norm, not merely the ℓ∞ norm, has an online Komlós asymptotic given by the same geometric extinction time.
  • When the batch size n is smaller than m-1 the continuum limit is flow by the sum of the largest n principal curvatures rather than mean curvature flow.
  • Upper and lower bounds on extinction times for convex bodies become upper and lower bounds on the asymptotic value of the corresponding online balancing game.

Reading between the lines

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

  • The continuum description may supply a new route to lower bounds for the classic (T=1) Komlós problem by examining the short-time expansion of the same PDE.
  • The same viscosity machinery should apply to online balancing against oblivious or stochastic adversaries once the corresponding Bellman operators are identified.
  • Explicit barriers built from the heat equation give the first analytic estimates of the extinction time of the cube under mean curvature flow that are sharp up to the constant √2.
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

0 major / 6 minor

Summary. The paper determines the large-T asymptotics of the online Komlós game: Paul presents batches of n unit-ball vectors in R^m and Carol chooses signs, with the value K_{n,T}(E) equal to the resulting ||sum||_E after T rounds. Theorem 2.1 states that lim_{T o∞} K_{n,T}(E)/√T = 1/√(2 τ_n(E)), where τ_n(E) is the extinction time of the centrally symmetric convex body E under the flow whose normal speed is the sum of the largest min(n,m−1) principal curvatures. When n≥m−1 this is mean curvature flow; for the cube one obtains the dimensional bounds (1−o(1))√log m ≤ lim K_T(m,n)/√T ≤ (√2+o(1))√log m (Theorem 1.2 / 2.2). The argument constructs a dynamic-programming value function, rescales it to a finite-difference scheme, identifies the continuum Bellman operator via Banaszczyk’s ℓ² discrepancy theorem, passes to half-relaxed limits that are viscosity sub- and supersolutions, and invokes comparison plus convexity preservation to obtain uniqueness; non-smooth bodies are recovered by sandwich approximation.

Significance. The work supplies a clean continuum limit that localizes the classic Komlós problem to a curvature-driven geometric evolution, thereby importing the full apparatus of viscosity solutions and comparison principles into online vector balancing. The identification of the discrete max-min with the sum of the largest positive projected eigenvalues (Lemmas 5.4–5.6) is a non-trivial extension of the Kohn–Serfaty framework beyond orthonormal frames, made possible by Banaszczyk. The resulting extinction-time bounds for the cube under mean curvature flow are of independent geometric interest and are obtained by elementary heat-equation barriers. The proofs are self-contained once standard viscosity machinery is granted, and the statements are parameter-free. Even if the hoped-for feedback into the offline Komlós conjecture remains unrealized, the localization perspective and the precise constant 1/√(2τ) constitute a substantial contribution at the interface of combinatorial discrepancy and geometric PDE.

minor comments (6)
  1. Notation for the game value oscillates between K_T(m,n) (Introduction) and K_{n,T}(E) (Section 2 onward). A single consistent symbol would help the reader.
  2. In the elementary strategy bounds (3.1) the constant c(m,n)=min(n,m−1)/(min(n,m−1)+1) is stated without a one-line derivation that |y_t|≤√(n*+1)∥y_t∥_∞; adding it would make the comparison with the continuum constant immediate.
  3. Lemma 5.4 asserts sharpness of both half-relaxed envelopes at p=0 by exhibiting sequences; a short explicit construction for the liminf sequence (beyond the sketch with p_k=√δ_k σ_0) would make the claim fully self-contained.
  4. Appendix B cites Giga–Goto–Ishii–Sato for the comparison principle under linear growth. A one-sentence verification that the structural hypotheses (degenerate ellipticity, F^*=F_* at the origin, local bound on |F|) hold for H_n would spare the reader a cross-check.
  5. Figures 1 and 2 are conceptually helpful but lack axis labels and a precise caption statement of the ambient dimension and batch size; adding these would improve accessibility.
  6. Typographical: “Koml´os” appears with inconsistent accent placement in a few places (e.g., page 2); standardize throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: continuum limit of discrete game value is derived independently of the geometric extinction time.

full rationale

The derivation is a standard viscosity-solution continuum limit (rescaled value function w_δ solves a finite-difference scheme whose consistency limit is the level-set PDE for curvature flow; half-relaxed limits are sub/supersolutions by Lemmas 5.4/5.6 + 6.5; uniqueness via external comparison of Giga–Goto–Ishii–Sato). The extinction time τ_n(E) is defined purely geometrically (2.1) and is related to the PDE solution only after the limit is established (via self-similarity Lemma B.3 and convexity preservation Theorem B.4, both external). Banaszczyk is used only to identify the Bellman operator with the sum of largest positive eigenvalues of the projected Hessian; no parameter is fitted to game values, no uniqueness theorem is imported from the authors’ prior work, and the heat-equation barriers for the dimensional bounds are independent comparison arguments. The argument is self-contained against external geometric and PDE benchmarks.

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

The paper rests entirely on classical theorems of discrepancy theory and geometric PDE; no free parameters are fitted and no new physical or combinatorial entities are postulated.

assumptions (4)
  • standard math Banaszczyk’s ℓ² discrepancy theorem (Theorem 5.1): for any ellipsoid D the signed sum of vectors inside D has Euclidean length at most the sum of squared semi-axes.
    Used to evaluate the max-min of the quadratic form that becomes the curvature operator H_n (Remark 5.2, Lemma 5.6).
  • standard math Comparison principle for viscosity solutions of geometric parabolic equations (Giga–Goto–Ishii–Sato).
    Invoked as Theorem B.2 to force the half-relaxed limits of w_δ to coincide.
  • standard math Preservation of convexity of sublevel sets under the curvature flow (Huisken / Evans–Spruck / Andrews).
    Guarantees that the positive-curvature operator H_n and the unrestricted operator H̲_n agree on the solution with terminal data ∥·∥_E (Theorem B.4).
  • standard math Existence of a unique viscosity solution to the terminal-value problem for the level-set curvature flow with continuous terminal data of linear growth.
    Standard consequence of the comparison principle; used to identify the continuum limit of the rescaled game values.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Online Koml\'os converges to mean curvature flow." pith.science (2026). https://pith.science/paper/BQ3X6QIG

@misc{pith2026260708943,
  author       = {Pith},
  title        = {Pith review of: Online Koml\'os converges to mean curvature flow},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BQ3X6QIG}},
  note         = {Machine review of arXiv:2607.08943}
}
abstract

We determine the asymptotics of a game inspired by classic vector balancing problems in combinatorial discrepancy theory. In this game, which we call the online Koml\'os game, two players, Paul and Carol, update the state vector $y$ in $\mathbb{R}^m$, initially placed at $0$. At each round, Paul chooses freely a set of $n$ vectors in the Euclidean unit ball, and Carol chooses, for each such vector, whether to leave it unchanged or reverse its sign. The resulting vectors are all added to $y$, and the game proceeds to a new round. After $T$ rounds, the game ends, and the $\ell_\infty$ norm of the state vector $y$ is determined. Paul's objective throughout the game is to maximize this norm, and Carol's objective is to minimize it. As $T$ gets large, we establish that the leading order term of the value of this game is $\sqrt{T/2\tau}$, where $\tau$ is the extinction time of the unit cube in $\mathbb{R}^m$ under a curvature-based flow characterized by the values of $m$ and $n$. When $n\geq m-1$, this flow is the mean curvature flow, and we show that $1/\sqrt{2\tau} =\Theta(\sqrt{\log m})$. Our results build upon the work of Kohn and Serfaty on deterministic games and mean curvature flow, combined with Banaszczyk's $\ell^2$ analogue of the Beck-Fiala theorem. As the large $T$ limit of the online Koml\'os game amounts to a localization of the classic Koml\'os problem, we hope this work can shed light on this and other vector balancing problems. Our results generalize to the version of the online Koml\'os game with the final value given by an arbitrary norm in $\mathbb{R}^m$.

Figures

Figures reproduced from arXiv: 2607.08943 by the authors.

Figure 1
Figure 1. We have a convex body E and a frame with origin placed at the body’s center. If A is the matrix whose columns represent the vectors in the frame, then as ¯ε ranges over {−1, 1} 3 we obtain 8 vectors Aε¯ whose locations are marked above by white circles. In the classic Koml´os problem Paul aims to choose the n (in this figure, n = 3) column vectors of A so that the 2n vertices {Aε}ε of the frame it defines all lie ou… view at source ↗
Figure 2
Figure 2. For the online problem (m = 3, n = 2 in this figure) we expect after a large number of steps t that yt will lie on λ∂E for a large λ. Moreover, since the column vectors of A have length at most 1, the game is effectively zooming in to a small portion of ∂E. If the body E is smooth, the area where Paul and Carol play is well approximated by paraboloid tangent to ∂E at ˆyt := ∥yt∥ −1 E yt. Therefore, Paul’s choice for… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

117 extracted references · 1 linked inside Pith

  1. [1]

    Smoothed analysis of the Koml´ os conjecture: Rademacher noise

    Elad Aigner-Horev, Dan Hefetz, and Michael Trushkin. Smoothed analysis of the Koml´ os conjecture: Rademacher noise. The Electronic Journal of Combinatorics, 32(1):P1.52, Mar. 2025

    2025

  2. [2]

    On the structure of bad science matrices, 2025

    Alex Albors, Hisham Bhatti, Lukshya Ganjoo, Raymond Guo, Dmitriy Kunisky, Rohan Mukherjee, Alicia Stepin, and Tony Zeng. On the structure of bad science matrices, 2025

    2025

  3. [3]

    Spencer.The Probabilistic Method

    Noga Alon and Joel H. Spencer.The Probabilistic Method. John Wiley & Sons, Inc., 2000

    2000

  4. [4]

    Altschuler and Jonathan Niles-Weed

    Dylan J. Altschuler and Jonathan Niles-Weed. The discrepancy of random rectangular matrices.Random Structures & Algorithms, 60(4):551–593, 2022

    2022

  5. [5]

    Altschuler and Konstantin Tikhomirov

    Dylan J. Altschuler and Konstantin Tikhomirov. A threshold for online balancing of sparse i.i.d. vectors, 2025

    2025

  6. [6]

    Liu, and Mehtaab Sawhney

    Ryan Alweiss, Yang P. Liu, and Mehtaab Sawhney. Discrepancy minimization via a self-balancing walk. InProceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing, STOC 2021, page 14–20, New York, NY, USA,

    2021

  7. [7]

    Association for Computing Machinery

  8. [8]

    Contraction of convex hypersurfaces in Euclidean space.Calculus of Variations and Partial Differential Equations, 2(2):151–171, 1994

    Ben Andrews. Contraction of convex hypersurfaces in Euclidean space.Calculus of Variations and Partial Differential Equations, 2(2):151–171, 1994

    1994

Show all 117 references
  1. [9]

    Storage capacity in symmetric binary perceptrons.Journal of Physics A: Mathematical and Theoretical, 52(29):294003, June 2019

    Benjamin Aubin, Will Perkins, and Lenka Zdeborov´ a. Storage capacity in symmetric binary perceptrons.Journal of Physics A: Mathematical and Theoretical, 52(29):294003, June 2019

    2019

  2. [10]

    A Beck—Fiala-type theorem for Euclidean norms.European Journal of Combinatorics, 11(6):497– 500, 1990

    Wojciech Banaszczyk. A Beck—Fiala-type theorem for Euclidean norms.European Journal of Combinatorics, 11(6):497– 500, 1990

    1990

  3. [11]

    Balancing vectors and convex bodies.Studia Mathematica, 106(1):93–100, 1993

    Wojciech Banaszczyk. Balancing vectors and convex bodies.Studia Mathematica, 106(1):93–100, 1993

    1993

  4. [12]

    Balancing vectors and Gaussian measures of n-dimensional convex bodies.Random Structures & Algorithms, 12(4):351–360, 1998

    Wojciech Banaszczyk. Balancing vectors and Gaussian measures of n-dimensional convex bodies.Random Structures & Algorithms, 12(4):351–360, 1998

    1998

  5. [13]

    Bandeira and Helmut B¨ olcskei

    Afonso S. Bandeira and Helmut B¨ olcskei. Matrix discrepancy for representations of finite groups, 2026

    2026

  6. [14]

    Bandeira, Dmitriy Kunisky, Dustin G

    Afonso S. Bandeira, Dmitriy Kunisky, Dustin G. Mixon, and Xinmeng Zeng. On the concentration of Gaussian Cayley matrices.Applied and Computational Harmonic Analysis, 73:101694, 2024

    2024

  7. [15]

    A remark on Kashin’s discrepancy argument and partial coloring in the Koml´ os conjecture.Portugaliae Mathematica, 79(3):311–316, 2022

    Afonso S Bandeira, Antoine Maillard, and Nikita Zhivotovskiy. A remark on Kashin’s discrepancy argument and partial coloring in the Koml´ os conjecture.Portugaliae Mathematica, 79(3):311–316, 2022

    2022

  8. [16]

    Constructive algorithms for discrepancy minimization

    Nikhil Bansal. Constructive algorithms for discrepancy minimization. In2010 IEEE 51st Annual Symposium on Foun- dations of Computer Science, pages 3–10, 2010

    2010

  9. [17]

    Discrepancy theory and related algorithms

    Nikhil Bansal. Discrepancy theory and related algorithms. InProc. Int. Cong. Math, volume 7, pages 5178–5210. Inter- national Mathematical Union, 2022

    2022

  10. [18]

    An algorithm for Koml´ os conjecture matching Banaszczyk’s bound

    Nikhil Bansal, Daniel Dadush, and Shashwat Garg. An algorithm for Koml´ os conjecture matching Banaszczyk’s bound. SIAM Journal on Computing, 48(2):534–553, 2019

    2019

  11. [19]

    The Gram–Schmidt walk: A cure for the Banaszczyk blues.Theory of Computing, 15(21):1–27, 2019

    Nikhil Bansal, Daniel Dadush, Shashwat Garg, and Shachar Lovett. The Gram–Schmidt walk: A cure for the Banaszczyk blues.Theory of Computing, 15(21):1–27, 2019

    2019

  12. [20]

    Decoupling via affine spectral-independence: Beck-Fiala and Koml´ os bounds beyond Banaszczyk

    Nikhil Bansal and Haotian Jiang. Decoupling via affine spectral-independence: Beck-Fiala and Koml´ os bounds beyond Banaszczyk. InProceedings of the 58th Annual ACM SIGACT Symposium on Theory of Computing, to appear, STOC 2026, New York, NY, USA, 2026. Association for Computin...

    2026

  13. [21]

    Online discrepancy minimization for stochastic arrivals

    Nikhil Bansal, Haotian Jiang, Raghu Meka, Sahil Singla, and Makrand Sinha. Online discrepancy minimization for stochastic arrivals. InProceedings of the 2021 ACM-SIAM Symposium on Discrete Algorithms (SODA), pages 2842– 2861

    2021

  14. [22]

    Prefix Discrepancy, Smoothed Analysis, and Combinatorial Vector Balancing

    Nikhil Bansal, Haotian Jiang, Raghu Meka, Sahil Singla, and Makrand Sinha. Prefix Discrepancy, Smoothed Analysis, and Combinatorial Vector Balancing. In Mark Braverman, editor,13th Innovations in Theoretical Computer Science Conference (ITCS 2022), volume 215 ofLeibniz Interna...

    2022

  15. [23]

    Smoothed Analysis of the Koml´ os Con- jecture

    Nikhil Bansal, Haotian Jiang, Raghu Meka, Sahil Singla, and Makrand Sinha. Smoothed Analysis of the Koml´ os Con- jecture. In Miko laj Boja´ nczyk, Emanuela Merelli, and David P. Woodruff, editors,49th International Colloquium on Automata, Languages, and Programming (ICALP 202...

    2022

  16. [24]

    Online vector balancing and geometric discrepancy

    Nikhil Bansal, Haotian Jiang, Sahil Singla, and Makrand Sinha. Online vector balancing and geometric discrepancy. In Proceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing, STOC 2020, page 1139–1152, New York, NY, USA, 2020. Association for Computing Machinery

    2020

  17. [25]

    A Unified Approach to Discrepancy Minimization

    Nikhil Bansal, Aditi Laddha, and Santosh Vempala. A Unified Approach to Discrepancy Minimization. In Amit Chakrabarti and Chaitanya Swamy, editors,Approximation, Randomization, and Combinatorial Optimization. Algo- rithms and Techniques (APPROX/RANDOM 2022), volume 245 ofLeibn...

    2022

  18. [26]

    Nikhil Bansal and Joel H. Spencer. On-line balancing of random inputs.Random Structures & Algorithms, 57(4):879–891, 2020

    2020

  19. [27]

    Springer, 1997

    Martino Bardi, Italo Capuzzo Dolcetta, et al.Optimal control and viscosity solutions of Hamilton-Jacobi-Bellman equa- tions, volume 12. Springer, 1997

    1997

  20. [28]

    Barles and P

    G. Barles and P. E. Souganidis. Convergence of approximation schemes for fully nonlinear second order equations.As- ymptotic Anal., 4(3):271–283, 1991

    1991

  21. [29]

    Integer-making

    J´ ozsef Beck and Tibor Fiala. “Integer-making” theorems.Discrete Applied Mathematics, 3(1):1–8, 1981

    1981

  22. [30]

    Some remarks on the Gram-Schmidt walk algorithm and consequences for the Koml´ os conjecture

    Witold Bednorz and Piotr Godlewski. Some remarks on the Gram-Schmidt walk algorithm and consequences for the Koml´ os conjecture. In Rados law Adamczak, Nathael Gozlan, Karim Lounici, Mokshay Madiman, Florence Merlev` ede, and Elisabeth Werner, editors,High Dimensional Probabi...

    2026

  23. [31]

    Princeton Landmarks in Mathematics

    Richard Bellman.Dynamic programming. Princeton Landmarks in Mathematics. Princeton University Press, Princeton, NJ, 2010. Reprint of the 1957 edition, With a new introduction by Stuart Dreyfus

    2010

  24. [32]

    Brakke.The motion of a surface by its mean curvature

    Kenneth A. Brakke.The motion of a surface by its mean curvature. Mathematical Notes. Princeton Univ. Press, Princeton, NJ, 1978

    1978

  25. [33]

    A representation formula for the mean curvature motion

    Rainer Buckdahn, Pierre Cardaliaguet, and Marc Quincampoix. A representation formula for the mean curvature motion. SIAM Journal on Mathematical Analysis, 33(4):827–846, 2001

    2001

  26. [34]

    Lecture notes on viscosity solutions, 2024

    Jeff Calder. Lecture notes on viscosity solutions, 2024. Available athttps://www-users.cse.umn.edu/ ~jwcalder/ viscosity_solutions.pdf

    2024

  27. [35]

    Karthekeyan Chandrasekaran and Santosh S. Vempala. Integer feasibility of random polytopes: random integer programs. InProceedings of the 5th Conference on Innovations in Theoretical Computer Science, ITCS ’14, page 449–458, New York, NY, USA, 2014. Association for Computing Machinery

    2014

  28. [36]

    Tight hardness results for minimizing discrepancy

    Moses Charikar, Alantha Newman, and Aleksandar Nikolov. Tight hardness results for minimizing discrepancy. InPro- ceedings of the Twenty-Second Annual ACM-SIAM Symposium on Discrete Algorithms, SODA ’11, page 1607–1614, USA, 2011. Society for Industrial and Applied Mathematics

    2011

  29. [37]

    A mixed problem for the infinity laplacian via tug-of-war games.Calculus of Variations and Partial Differential Equations, 34(3):307–320, 2009

    Fernando Charro, Jesus Garc´ ıa Azorero, and Julio D Rossi. A mixed problem for the infinity laplacian via tug-of-war games.Calculus of Variations and Partial Differential Equations, 34(3):307–320, 2009

    2009

  30. [38]

    A note on norms of signed sums of vectors.Advances in Geometry, 21(1):5–14, 2021

    Giorgos Chasapis and Nikos Skarmogiannis. A note on norms of signed sums of vectors.Advances in Geometry, 21(1):5–14, 2021

    2021

  31. [39]

    Cambridge University Press, 2000

    Bernard Chazelle.The Discrepancy Method: Randomness and Complexity. Cambridge University Press, 2000

    2000

  32. [40]

    J. Cheeger. A lower bound for the smallest eigenvalue of the Laplacian. Probl. Analysis, Sympos. in Honor of Salomon Bochner, Princeton Univ. 1969, 195-199 (1970)., 1970

    1969

  33. [41]

    Springer Cham, 2014

    William Chen, Anand Srivastav, and Giancarlo Travaglini, editors.A Panorama of Discrepancy Theory. Springer Cham, 2014

    2014

  34. [42]

    Uniqueness and existence of viscosity solutions of generalized mean curvature flow equations.Proc

    Yun Gang Chen, Yoshikazu Giga, and Shun’ichi Goto. Uniqueness and existence of viscosity solutions of generalized mean curvature flow equations.Proc. Japan Acad. Ser. A Math. Sci., 65(7):207–210, 1989

    1989

  35. [43]

    Gaussian discrepancy: A probabilistic relaxation of vector balancing.Discrete Applied Mathematics, 322:123–141, 2022

    Sinho Chewi, Patrik Gerber, Philippe Rigollet, and Paxton Turner. Gaussian discrepancy: A probabilistic relaxation of vector balancing.Discrete Applied Mathematics, 322:123–141, 2022

    2022

  36. [44]

    Mean curvature flow.Bulletin of the American Mathematical Society, 52(2):297–333, 2015

    Tobias Colding, William Minicozzi, Erik Pedersen, et al. Mean curvature flow.Bulletin of the American Mathematical Society, 52(2):297–333, 2015

    2015

  37. [45]

    Costello

    Kevin P. Costello. Balancing Gaussian vectors.Israel Journal of Mathematics, 172(1):145–156, 2009

    2009

  38. [46]

    Sur la forme int´ egro-diff´ erentielle des op´ erateurs dec∞ k danscsatisfaisant au principe du maximum

    Philippe Courrege. Sur la forme int´ egro-diff´ erentielle des op´ erateurs dec∞ k danscsatisfaisant au principe du maximum. S´ eminaire de th´ eorie du potentiel, 10(1):1–38, 1965

    1965

  39. [47]

    User’s guide to viscosity solutions of second order partial differential equations.Bulletin of the American mathematical society, 27(1):1–67, 1992

    Michael G Crandall, Hitoshi Ishii, and Pierre-Louis Lions. User’s guide to viscosity solutions of second order partial differential equations.Bulletin of the American mathematical society, 27(1):1–67, 1992

    1992

  40. [48]

    Anirban DasGupta, S. N. Lahiri, and Jordan Stoyanov. Sharp fixednbounds and asymptotic expansions for the mean and the median of a Gaussian sample maximum, and applications to the Donoho–Jin model.Statistical Methodology, 20:40–62, 2014

    2014

  41. [49]

    Efficient algorithms for discrepancy minimization in convex sets.Random Structures & Algorithms, 53(2):289–307, 2018

    Ronen Eldan and Mohit Singh. Efficient algorithms for discrepancy minimization in convex sets.Random Structures & Algorithms, 53(2):289–307, 2018

    2018

  42. [50]

    Motion of level sets by mean curvature

    Lawrence C Evans and Joel Spruck. Motion of level sets by mean curvature. I. InFundamental Contributions to the Continuum Theory of Evolving Phase Interfaces in Solids: A Collection of Reprints of 14 Seminal Papers, pages 328–

  43. [51]

    On the Beck-Fiala conjecture for random set systems.Random Structures & Algorithms, 54(4):665–675, 2019

    Esther Ezra and Shachar Lovett. On the Beck-Fiala conjecture for random set systems.Random Structures & Algorithms, 54(4):665–675, 2019

    2019

  44. [52]

    The mean-field limit of online stochastic vector balancing, 2026

    Christian Fiedler, Joe Jackson, Daniel Lacker, and Jonathan Niles-Weed. The mean-field limit of online stochastic vector balancing, 2026

    2026

  45. [53]

    Springer, 2006

    Wendell H Fleming and H Mete Soner.Controlled Markov processes and viscosity solutions. Springer, 2006. ONLINE KOML ´OS CONVERGES TO MEAN CURVATURE FLOW 31

    2006

  46. [54]

    On the discrepancy of random matrices with many columns.Random Structures & Algorithms, 57(1):64–96, 2020

    Cole Franks and Michael Saks. On the discrepancy of random matrices with many columns.Random Structures & Algorithms, 57(1):64–96, 2020

    2020

  47. [55]

    Galambos.Asymptotic Theory of Extreme Order Statistics

    J. Galambos.Asymptotic Theory of Extreme Order Statistics. Wiley, New York, 1987

    1987

  48. [56]

    On some vector balancing problems.Studia Mathematica, 122(3):225–234, 1997

    Apostolos Giannopoulos. On some vector balancing problems.Studia Mathematica, 122(3):225–234, 1997

    1997

  49. [57]

    Y. Giga, S. Goto, H. Ishii, and M.-H. Sato. Comparison principle and convexity preserving properties for singular degen- erate parabolic equations on unbounded domains.Indiana Univ. Math. J., 40(2):443–470, 1991

    1991

  50. [58]

    Springer, 2006

    Yoshikazu Giga.Surface evolution equations: A level set approach. Springer, 2006

    2006

  51. [59]

    Monographs in Mathematics

    Yoshikazu Giga and Qing Liu.Surface Evolution Equations: A Level Set Approach. Monographs in Mathematics. Birkh¨ auser Cham, 2 edition, 2026

    2026

  52. [60]

    On a lower bound for the extinction time of surfaces moved by mean curvature

    Yoshikazu Giga and Kazuyuki Yama-uchi. On a lower bound for the extinction time of surfaces moved by mean curvature. Calculus of Variations and Partial Differential Equations, 1(4):417–428, 1993

    1993

  53. [61]

    Extremal properties of orthogonal parallelepipeds and their applications to the geometry of Banach spaces.Mathematics of the USSR-Sbornik, 64(1):85–96, 1989

    Efim Davydovich Gluskin. Extremal properties of orthogonal parallelepipeds and their applications to the geometry of Banach spaces.Mathematics of the USSR-Sbornik, 64(1):85–96, 1989

    1989

  54. [62]

    Rossi, and Jorge Ruiz-Cases

    Irene Gonzalvez, Alfredo Miranda, Julio D. Rossi, and Jorge Ruiz-Cases. A two-player zero-sum probabilistic game that approximates the mean curvature flow.Communications in Mathematics, Volume 34 (2026), Issue 2 (Special issue: Latin American mathematics), Jul 2025

    2026

  55. [63]

    Nestor Guillen and Russell W. Schwab. Min-max formulas for nonlocal elliptic operators.Calc. Var. Partial Differential Equations, 58(6):Paper No. 209, 79, 2019

    2019

  56. [64]

    Nestor Guillen and Russell W. Schwab. Min-max formulas for nonlocal elliptic operators on Euclidean space.Nonlinear Anal., 193:111468, 51, 2020

    2020

  57. [65]

    Sinan G¨ unt¨ urk

    C. Sinan G¨ unt¨ urk. Mathematics of analog-to-digital conversion.Communications on Pure and Applied Mathematics, 65(12):1671–1696, 2012

    2012

  58. [66]

    Smoothed analysis with adaptive adversaries.J

    Nika Haghtalab, Tim Roughgarden, and Abhishek Shetty. Smoothed analysis with adaptive adversaries.J. ACM, 71(3), June 2024

    2024

  59. [67]

    On a conjecture of Komlos about signed sums of vectors inside the sphere.European Journal of Combinatorics, 9(1):33–37, 1988

    D Hajela. On a conjecture of Komlos about signed sums of vectors inside the sphere.European Journal of Combinatorics, 9(1):33–37, 1988

    1988

  60. [68]

    Spielman, and Peng Zhang

    Christopher Harshaw, Fredrik S¨ avje, Daniel A. Spielman, and Peng Zhang. Balancing covariates in randomized experi- ments with the Gram–Schmidt walk design.Journal of the American Statistical Association, 119(548):2934–2946, 2024

    2024

  61. [69]

    A Fourier-analytic approach for the discrepancy of random set systems

    Rebecca Hoberg and Thomas Rothvoss. A Fourier-analytic approach for the discrepancy of random set systems. In Proceedings of the Thirtieth Annual ACM-SIAM Symposium on Discrete Algorithms, SODA ’19, page 2547–2556, USA,

  62. [70]

    Society for Industrial and Applied Mathematics

  63. [71]

    Flow by mean curvature of convex surfaces into spheres.Journal of Differential Geometry, 20(1):237– 266, 1984

    Gerhard Huisken. Flow by mean curvature of convex surfaces into spheres.Journal of Differential Geometry, 20(1):237– 266, 1984

    1984

  64. [72]

    Repeated games for non-linear parabolic integro-differential equations and integral cur- vature flows.Discrete and Continuous Dynamical Systems, 29(4):1517–1552, 2011

    Cyril Imbert and Sylvia Serfaty. Repeated games for non-linear parabolic integro-differential equations and integral cur- vature flows.Discrete and Continuous Dynamical Systems, 29(4):1517–1552, 2011

    2011

  65. [73]

    Online geometric discrepancy for stochastic arrivals with applica- tions to envy minimization, 2019

    Haotian Jiang, Janardhan Kulkarni, and Sahil Singla. Online geometric discrepancy for stochastic arrivals with applica- tions to envy minimization, 2019

    2019

  66. [74]

    Bounds on the expectation of the maximum of samples from a Gaussian, 1998

    Gautam Kamath. Bounds on the expectation of the maximum of samples from a Gaussian, 1998

    1998

  67. [75]

    B. S. Kashin. On an isometric operator inL 2(0,1).C. R. Acad. Bulg. Sci., 38:1613–1615, 1985

    1985

  68. [76]

    Vladimir A. Kobzar. The symmetric two-armed bandit: the general case, 2026

    2026

  69. [77]

    Kobzar and Robert V

    Vladimir A. Kobzar and Robert V. Kohn. A PDE-based analysis of the symmetric two-armed Bernoulli bandit, 2022

    2022

  70. [78]

    Kobzar, Robert V

    Vladimir A. Kobzar, Robert V. Kohn, and Zhilei Wang. New potential-based bounds for prediction with expert advice. In Jacob Abernethy and Shivani Agarwal, editors,Proceedings of the 33rd Annual Conference on Learning Theory (COLT), volume 125 ofProceedings of Machine Learning ...

    2020

  71. [79]

    Kobzar, Robert V

    Vladimir A. Kobzar, Robert V. Kohn, and Zhilei Wang. New potential-based bounds for the geometric-stopping version of prediction with expert advice. In Jianfeng Lu and Rachel Ward, editors,Proceedings of the 1st Annual Conference on Mathematical and Scientific Machine Learning...

    2020

  72. [80]

    Robert Kohn and Sylvia Serfaty. A deterministic-control-based approach to motion by curvature.Communications on Pure and Applied Mathematics: A Journal Issued by the Courant Institute of Mathematical Sciences, 59(3):344–407, 2006

    2006

  73. [81]

    Kohn and Sylvia Serfaty

    Robert V. Kohn and Sylvia Serfaty. Second-order PDE’s and deterministic games. InICIAM 07 : 6th International Conference on Industrial and Applied Mathematics, Zurich, Switzerland, July 2007

    2007

  74. [82]

    Kohn and Sylvia Serfaty

    Robert V. Kohn and Sylvia Serfaty. A deterministic-control-based approach to fully nonlinear parabolic and elliptic equations.Communications on Pure and Applied Mathematics, 63:1298–1350, 2010

    2010

  75. [83]

    Optimal online discrepancy minimization

    Janardhan Kulkarni, Victor Reis, and Thomas Rothvoss. Optimal online discrepancy minimization. InProceedings of the 56th Annual ACM Symposium on Theory of Computing, STOC 2024, page 1832–1840, New York, NY, USA, 2024. Association for Computing Machinery

    2024

  76. [84]

    The discrepancy of unsatisfiable matrices and a lower bound for the Koml´ os conjecture constant.SIAM Journal on Discrete Mathematics, 37(2):586–603, 2023

    Dmitriy Kunisky. The discrepancy of unsatisfiable matrices and a lower bound for the Koml´ os conjecture constant.SIAM Journal on Discrete Mathematics, 37(2):586–603, 2023

    2023

  77. [85]

    Asymptotic bounds and online algorithms for average-case matrix discrepancy, 2025

    Dmitriy Kunisky, Timm Oertel, Nicola Wengiel, and Peiyuan Zhang. Asymptotic bounds and online algorithms for average-case matrix discrepancy, 2025

    2025

  78. [86]

    Minimum curvature flow and martingale exit times.Electronic Journal of Probability, 29:1–32, 2024

    Martin Larsson and Johannes Ruf. Minimum curvature flow and martingale exit times.Electronic Journal of Probability, 29:1–32, 2024. 32 NESTOR GUILLEN AND VLADIMIR A. KOBZAR

    2024

  79. [87]

    Deterministic discrepancy minimization via the multiplicative weight update method

    Avi Levy, Harishchandra Ramadas, and Thomas Rothvoss. Deterministic discrepancy minimization via the multiplicative weight update method. In Friedrich Eisenbrand and Jochen Koenemann, editors,Integer Programming and Combinatorial Optimization, pages 380–391, Cham, 2017. Spring...

    2017

  80. [88]

    A game-theoretic proof of convexity-preserving properties for motion by curvature.Indiana Univ

    Qing Liu, Armin Schikorra, and Xiaodan Zhou. A game-theoretic proof of convexity-preserving properties for motion by curvature.Indiana Univ. Math. J., 65(1):171–197, 2016

    2016

  81. [89]

    Y. Lonke. Combinatorial problems in finite dimensional normed spaces (PhD dissertation, Hebrew university), 1998

    1998

  82. [90]

    Constructive discrepancy minimization by walking on the edges.SIAM Journal on Computing, 44(5):1573–1582, 2015

    Shachar Lovett and Raghu Meka. Constructive discrepancy minimization by walking on the edges.SIAM Journal on Computing, 44(5):1573–1582, 2015

    2015

  83. [91]

    Springer Berlin, Heidelberg, 1999

    Jiˇ r´ ı Matouˇ sek.Geometric Discrepancy. Springer Berlin, Heidelberg, 1999

    1999

  84. [92]

    Factorization norms and hereditary discrepancy.International Mathematics Research Notices, 2020(3):751–780, 02 2020

    Jiˇ r´ ı Matouˇ sek, Aleksandar Nikolov, and Kunal Talwar. Factorization norms and hereditary discrepancy.International Mathematics Research Notices, 2020(3):751–780, 02 2020

    2020

  85. [93]

    On the role of convexity in isoperimetry, spectral gap and concentration.Inventiones mathematicae, 177:1 – 43, 2009

    Emanuel Milman. On the role of convexity in isoperimetry, spectral gap and concentration.Inventiones mathematicae, 177:1 – 43, 2009

    2009

  86. [94]

    The geometry of differential privacy: the sparse and approximate cases

    Aleksandar Nikolov, Kunal Talwar, and Li Zhang. The geometry of differential privacy: the sparse and approximate cases. InProceedings of the Forty-Fifth Annual ACM Symposium on Theory of Computing, STOC ’13, page 351–360, New York, NY, USA, 2013. Association for Computing Machinery

    2013

  87. [95]

    Optimal non-asymptotic lower bound on the minimax regret of learning with expert advice, 2015

    Francesco Orabona and D´ avid P´ al. Optimal non-asymptotic lower bound on the minimax regret of learning with expert advice, 2015. Available athttps://arxiv.org/abs/1511.02176

    2015 arXiv

  88. [96]

    Applied Mathematical Sciences

    Stanley Osher and Ronald Fedkiw.Level Set Methods and Dynamic Implicit Surfaces. Applied Mathematical Sciences. Springer New York, NY, 2002

    2002

  89. [97]

    Tug-of-war and the infinity Laplacian.Journal of the American Mathematical Society, 22(1):167–210, 2009

    Yuval Peres, Oded Schramm, Scott Sheffield, and David Wilson. Tug-of-war and the infinity Laplacian.Journal of the American Mathematical Society, 22(1):167–210, 2009

    2009

  90. [98]

    Discrepancy minimization via regularization

    Lucas Pesenti and Adrian Vladu. Discrepancy minimization via regularization. InProceedings of the 2023 Annual ACM- SIAM Symposium on Discrete Algorithms (SODA), pages 1734–1758. SIAM, 2023

    2023

  91. [99]

    L. S. Pontryagin. The mathematical theory of optimal processes and differential games. volume 169, pages 119–158, 254–255. 1985. Topology, ordinary differential equations, dynamical systems

    1985

  92. [100]

    Constructive discrepancy minimization for convex sets.SIAM Journal on Computing, 46(1):224–234, 2017

    Thomas Rothvoss. Constructive discrepancy minimization for convex sets.SIAM Journal on Computing, 46(1):224–234, 2017

    2017

  93. [101]

    Viscosity solutions of elliptic equations, 2015

    Luis Silvestre. Viscosity solutions of elliptic equations, 2015. Available athttps://www.math.uchicago.edu/ ~luis/ preprints/viscosity-solutions.pdf

    2015

  94. [102]

    The structure of extremal bad science matrices, 2025

    Shridhar Sinha. The structure of extremal bad science matrices, 2025

    2025

  95. [103]

    Discrepancy and Fisher information, 2026

    Gleb Smirnov and Roman Vershynin. Discrepancy and Fisher information, 2026

    2026

  96. [104]

    Dynamic programming for stochastic target problems and geometric flows.Journal of the European Mathematical Society, 4(3):201–236, 2002

    H Mete Soner and Nizar Touzi. Dynamic programming for stochastic target problems and geometric flows.Journal of the European Mathematical Society, 4(3):201–236, 2002

    2002

  97. [105]

    A stochastic representation for mean curvature type geometric flows.The Annals of probability, 31(3):1145–1165, 2003

    H Mete Soner and Nizar Touzi. A stochastic representation for mean curvature type geometric flows.The Annals of probability, 31(3):1145–1165, 2003

    2003

  98. [106]

    Balancing games.Journal of Combinatorial Theory, Series B, 23(1):68–74, 1977

    Joel Spencer. Balancing games.Journal of Combinatorial Theory, Series B, 23(1):68–74, 1977

    1977

  99. [107]

    Six standard deviations suffice.Transactions of the American Mathematical Society, 289(2):679–706, 1985

    Joel Spencer. Six standard deviations suffice.Transactions of the American Mathematical Society, 289(2):679–706, 1985

    1985

  100. [108]

    Balancing vectors in the max norm.Combinatorica, 6(1):55–65, 1986

    Joel Spencer. Balancing vectors in the max norm.Combinatorica, 6(1):55–65, 1986

    1986

  101. [109]

    Society for Industrial and Applied Mathematics, 2nd edition edition, 1994

    Joel Spencer.Ten Lectures on the Probabilistic Method. Society for Industrial and Applied Mathematics, 2nd edition edition, 1994

    1994

  102. [110]

    Spielman and Shang-Hua Teng

    Daniel A. Spielman and Shang-Hua Teng. Smoothed analysis of algorithms: Why the simplex algorithm usually takes polynomial time.J. ACM, 51(3):385–463, May 2004

    2004

  103. [111]

    Bad science matrices, 2024

    Stefan Steinerberger. Bad science matrices, 2024

    2024

  104. [112]

    Springer, 2019

    Yoshihiro Tonegawa.Brakke’s Mean Curvature Flow: An Introduction. Springer, 2019

    2019

  105. [113]

    London Mathematical Society Stu- dent Texts

    Giancarlo Travaglini.Number Theory, Fourier Analysis and Geometric Discrepancy. London Mathematical Society Stu- dent Texts. Cambridge University Press, 2014

    2014

  106. [114]

    Global Comparison Property

    Paxton Turner, Raghu Meka, and Philippe Rigollet. Balancing Gaussian vectors in high dimension. In Jacob Abernethy and Shivani Agarwal, editors,Proceedings of Thirty Third Conference on Learning Theory, volume 125 ofProceedings of Machine Learning Research, pages 3455–3486. PM...

    2020

  107. [115]

    There is aK >0such that for allx∈R m, t∈[−T,0]we have u(x, t)≤K(|x|+ 1), v(x, t)≥ −K(|x|+ 1)

  108. [116]

    The functionsu(x, t), v(x, t)are continuous ast→0, that is, we haveu ∗(x,0) =u ∗(x,0)andv ∗(x,0) = v∗(x,0)for everyx∈R m

  109. [117]

    positive

    For some modulus of continuityρwe have u(x,0)−v(y,0)≤ρ(|x−y|). Then, we have u(x, t)≤v(x, t)∀x∈R m, t∈[−T,0]. Proof.This is a special case of [56, Theorem 3.2], a result that deals with general unbounded domainU (here we only need it forU=R m) and a more general PDEF(D 2u,∇u) ...

Pith tools

Reviewed July 13, 2026 · model on record in the stance chip above.