Pith. sign in

REVIEW 2 major objections 4 minor 39 references

Improved Upper Bounds for Slicing the Hypercube

T0 review · 2 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read The paper proves that roughly 4n/5 hyperplanes suffice to slice all edges of the n-dimensional hypercube, improving the long-standing 5n/6 upper bound via a construction of eight planes slicing the 10-cube.

desk verdict A genuine fifty-year improvement on the hypercube-slicing upper bound, resting on a finite construction that is explicit and almost certainly right, but the S(10)≤8 verification should be fully pinned down before I'd call it settled. read the letter →

arxiv 2602.16807 v2 pith:ZCHT2OJH submitted 2026-02-06 cs.AI cs.DMmath.CO

Duncan Soiffer , Nathaniel Itty , Christopher D. Rosin , Blake Bruell , Mason DiCicco , Gábor N. Sárközy , Ryan Offstein , Daniel Reichman This is my paper
classification cs.AIcs.DMmath.CO MSC 52C35
keywords hypercubeslicinghyperplanearrangementupperboundS(n)subadditivityreducedlocalsearchAI-assisteddiscovery
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 tackles the minimum number S(n) of hyperplanes needed to cut every edge of the n-dimensional hypercube, a long-standing problem in discrete geometry with links to threshold circuits. Its central claim is that for every n, S(n) is at most roughly 4n/5, improving the previous roughly 5n/6 bound that had stood since 1971. The proof rests on an explicit set of eight hyperplanes that slice all 5,120 edges of the 10-dimensional cube, combined with the subadditivity of S(n). The paper also reports new lower bounds on the maximum number of edges that can be sliced with fewer than n hyperplanes. A reader should care because the construction was found by an AI-driven search with human refinement, illustrating a workflow that can break decades-old barriers in combinatorial design.

What carries the argument

The load-bearing object is the set of eight hyperplanes in Eq. (1): each plane fixes its first six coefficients to the same value, so the arrangement respects the composition [6,1,1,1,1] of n=10. Under that composition the 10-cube reduces to a 6×1×1×1×1 grid with only 320 edges (16-fold fewer than Q10), and slicing of the reduced grid is equivalent to slicing the original cube. The proof then lifts the finite fact S(10)≤8 to all n using the subadditivity identity S(k+ℓ) ≤ S(k)+S(ℓ). The construction was found by an adaptive edge-weighted hill-climbing search, with the composition constraint itself discovered by observing structure in outputs of an AI-driven search tool.

What would settle it

Run the verification algorithm in Section G (or an independent exhaustive check) on the eight planes of Eq. (1): enumerate all 5,120 edges of Q10 and test whether some plane changes sign in the interior. If any edge has no sign change, the central claim collapses; the check is finite and publicly described.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that S(10) ≤ 8 by presenting eight explicit hyperplanes, each with six equal leading coefficients, and asserting (with an auxiliary verification algorithm) that these planes slice every edge of Q10. From this single finite construction, subadditivity yields the global bound: S(n) ≤ ⌈4n/5⌉ for all n except n an odd multiple of 5, where S(n) ≤ 4n/5 + 1. In addition, the paper claims new lower bounds for S(n,k), the maximum number of edges sliceable by k hyperplanes, for n up to 11 and selected larger dimensions, again via explicit search-discovered plane sets.

Load-bearing premise

The listed eight hyperplanes in Eq. (1) actually slice all 5,120 edges of Q10; the paper defers this finite fact to a verification algorithm and a code link rather than providing a printed or machine-checked witness.

Editorial extensions

If this is right

  • For every n not an odd multiple of 5, S(n) ≤ ⌈4n/5⌉; for n an odd multiple of 5, S(n) ≤ 4n/5 + 1.
  • The explicit eight-plane slicing of Q10 is the seed: S(10) ≤ 8, which via subadditivity yields all the asymptotic cases.
  • New records for S(n,k) follow for n up to 11 and for selected larger pairs (12,9), (13,10), (14,11), and (15,12), where k = n−3.
  • The construction shows that AI-assisted search with human refinement can produce finite combinatorial objects that improve long-standing bounds.

Reading between the lines

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

  • The near-complete slicings for n=11..15 (only a dozen or so edges short) suggest that the constant 4/5 may not be optimal, and that further construction-focused search could lower the upper bound again.
  • The reduced-hypercube representation, which groups equal coefficients, is a general search-space reduction that could be applied to other hyperplane-arrangement problems beyond edge slicing.
  • A machine-checked certificate (for example, a SAT-solver proof or a fully logged exhaustive enumeration) for the Q10 claim would make the theorem independently verifiable without trusting the accompanying code repository.
  • If the conjecture S(n)=Ω(n) is correct, then constructive constant-factor improvements of the type demonstrated here may be near the practical ceiling for upper bounds; the paper's methods do not address the lower-bound side.
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

2 major / 4 minor

Summary. The paper announces new upper bounds for the hypercube edge-slicing number S(n), improving Paterson's 1971 bound S(n) ≤ ceil(5n/6) to S(n) ≤ ceil(4n/5) for n not an odd multiple of 5, and to S(n) ≤ 4n/5 + 1 otherwise. The proof rests on subadditivity and one finite construction: eight explicit hyperplanes claimed to slice all 5120 edges of Q_10, listed in Eq. (1). The paper also reports lower bounds for S(n,k), the maximum number of sliced edges with k planes, in Tables 1 and 2, and describes an AI-assisted local-search methodology using CPro1, a human-modified LLM-driven discovery system. A verification algorithm is given in Section G, and code is referenced at a GitHub URL.

Significance. If the claimed finite construction is correct, this is a genuine improvement over a fifty-year-old upper bound and the first progress on the upper bound for S(n) since Paterson. The main construction is explicit and, in principle, checkable in seconds by a brute-force program. The subadditivity argument establishing the reduction to S(10) ≤ 8 is standard and internally consistent. A further strength is the paper's transparency about the AI-assisted discovery process: the theorem does not depend on the AI tool's correctness, only on the explicit post-hoc construction. The S(n,k) tables, if accompanied by witnesses, would also be useful computational data for a problem that has resisted theoretical progress.

major comments (2)
  1. [§G and Eq. (1)] The central claim of the paper, S(10) ≤ 8, is supported only by the listing of eight planes and the statement that they can be verified using the pseudocode in Section G. The manuscript contains no certificate of that verification: no output showing that all 5120 edges are sliced, no hash of the verification result, and no machine-checked proof. Since every upper bound in Theorem 1 reduces to this finite predicate, the paper should either include a printer-friendly certificate (e.g., the complete list of unsliced edges, which should be empty, produced by a deterministic verifier) or state explicitly that the computation was independently checked and how. Without this, the central claim is not reproducible from the manuscript alone.
  2. [Tables 1–2 and §4.1] The paper's secondary contribution, new lower bounds on S(n,k), is presented as numerical table entries but the corresponding hyperplane arrangements are not given in the manuscript. Unlike the S(10) ≤ 8 construction in Eq. (1), these table entries cannot be checked from the text. If these bounds are part of the paper's contribution, the authors should provide the actual k-plane collections for every bold entry, or at minimum an explicit, versioned digital artifact and a procedure that maps each table entry to its witness. As written, the only fully documented S(n,k) improvement is the S(15,12) construction in Appendix C.
minor comments (4)
  1. [§2, proof of Theorem 1] The case n ≥ 20 is dismissed with 'inequalities follow by repeated applications of subadditivity in a manner analogous to the cases for 10 ≤ n < 20.' This is easy to fill in by writing n = 10q + r and citing S(r) ≤ r for r ≤ 5 and S(6+j) ≤ 5 + j for j = 0,1,2,3, but it should be stated explicitly for a theorem whose proof is otherwise case-based.
  2. [§D, Theorem 3] The statement of Theorem 3 omits the bias term that appears in the proof. The notation aβ·vβ = a·v is only correct if the hyperplane equation is written as ⟨a,x⟩ = b with b included in both sides; please clarify this in the statement or the proof.
  3. [Tables 1–2] The table captions refer to bold and underlined entries, but the text formatting is lost in the arXiv rendering. Please ensure the final version either preserves the formatting or annotates each entry with a marker such as 'new' or 'full slicing' so that the claimed improvements over previous tables can be identified.
  4. [§G and References] The GitHub URL is given as 'github.com/DSoiffer/upper-bounds-for-hypercube-slicing' in the appendix but is only referred to as 'this GitHub URL' in the main text. Please include the explicit URL in the main text and, ideally, include a versioned commit hash or a DOI for the code to ensure reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the upper bound rests on an explicit finite construction, not on a self-referential derivation.

full rationale

The paper's central claim, Theorem 1, is an upper bound S(n) ≤ ceil(4n/5) (with the odd-multiple-of-5 exception) derived from subadditivity and the explicit claim that the eight hyperplanes in Eq. (1) slice all 5120 edges of Q10. The subadditivity argument and the case analysis are independent of how the construction was discovered. The finite construction is checkable by the algorithm in Section G and the linked code; the absence of a printed certificate or machine-checked proof is a verification gap, not circularity. The self-citations to CPro1 ([29]–[31]) describe the discovery process and are not load-bearing: no theorem step assumes the correctness of CPro1 or of any fitted parameter. There is no self-definitional reduction, no fitted input renamed as a prediction, and no imported uniqueness theorem forcing the result. The proof would stand or fall on the explicit hyperplane list alone. Accordingly, no circular step is identifiable by the standards requiring a quoted reduction or definitional equivalence.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

No new physical or mathematical entities are posited. The reduced hypercube is a derived coordinate projection, not an independent object: its properties are proven in Appendix D. Search hyperparameters are discovery tools, not axioms of the theorem.

free parameters (3)
  • fixed-column constant for [6,1,1,1,1] composition = -2 (also -9 in appendix constructions)
    Human-in-the-loop search restricted the first six coefficients of every hyperplane to one shared value and varied it; the listed proof uses c=-2. This is a search knob, not a fitted parameter in the theorem, but it is the main hand-chosen choice that shaped the discovered solution.
  • bias offset +0.5 = 0.5
    All hyperplanes use bias 0.5, described as an arbitrary offset to avoid passing through vertices. The proof depends on this non-integer value.
  • local search hyperparameters (max_iterations, weight_period, weight_limit, neighborhood d, coefficient range [-40,40]) = not specified in the paper
    Algorithm 1's hyperparameters are chosen by hand or by automated tuning; they affect discovery but not the final verification of the construction.
assumptions (3)
  • standard math S(n) is subadditive: S(k+ℓ) ≤ S(k)+S(ℓ).
    Invoked in the proof of Theorem 1 and attributed to Saks [32].
  • domain assumption Exact small-cube values S(n)=n for n≤5 and S(6)=5.
    Used in the case analysis of Theorem 1; cited to [12]/[13] rather than re-proven.
  • domain assumption The verification algorithm in Section G correctly decides whether a given plane set slices all edges of Q_n.
    The central finite check is delegated to this computational procedure; no certificate is printed in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improved Upper Bounds for Slicing the Hypercube." pith.science (2026). https://pith.science/paper/ZCHT2OJH

@misc{pith2026260216807,
  author       = {Pith},
  title        = {Pith review of: Improved Upper Bounds for Slicing the Hypercube},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZCHT2OJH}},
  note         = {Machine review of arXiv:2602.16807}
}
abstract

A collection of hyperplanes $\mathcal{H}$ slices all edges of the $n$-dimensional hypercube $Q_n$ with vertex set $\{-1,1\}^n$ if, for every edge $e$ in the hypercube, there exists a hyperplane in $\mathcal{H}$ intersecting $e$ in its interior. Let $S(n)$ be the minimum number of hyperplanes needed to slice $Q_n$. We prove that $S(n) \leq \lceil \frac{4n}{5} \rceil$, except when $n$ is an odd multiple of $5$, in which case $S(n) \leq \frac{4n}{5} +1$. This improves upon the previously known upper bound of $S(n) \leq \lceil\frac{5n}{6} \rceil$ due to Paterson reported in 1971. We also obtain new lower bounds on the maximum number of edges in $Q_n$ that can be sliced using $k<n$ hyperplanes. We prove the improved upper bound on $S(n)$ by constructing $8$ hyperplanes slicing $Q_{10}$ aided by the recently introduced CPro1: an automatic tool that uses reasoning LLMs coupled with automated hyperparameter tuning to create search algorithms for the discovery of mathematical constructions.

Figures

Figures reproduced from arXiv: 2602.16807 by the authors.

Figure 1
Figure 1. A 6-dimensional hypercube under the [3, 2, 1] composition reduces to 46 edges, and is transformed into a 3 × 2 × 1 lattice in three dimensions. Definition 2. A hyperplane that a satisfies a composition β in ℓ dimensions has a reduced hyperplane representation, a β = (a β 1 , . . . , a β ℓ ) ∈ R ℓ where ai = a β j for all i ∈ Bj . Definition 3. Any vertex v = (v1, v2, · · · , vn) has a reduced vertex representation v… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 9 linked inside Pith

  1. [1]

    An identity in combinatorial extremal theory.Advances in Mathematics, 80(2):137–151, 1990

    Rudolf Ahlswede and Zhen Zhang. An identity in combinatorial extremal theory.Advances in Mathematics, 80(2):137–151, 1990

    1990

  2. [2]

    Balancing sets of vectors.IEEE Transactions on Information Theory, 34(1):128–130, 2002

    Noga Alon, Ernest E Bergmann, Don Coppersmith, and Andrew M Odlyzko. Balancing sets of vectors.IEEE Transactions on Information Theory, 34(1):128–130, 2002

    2002

  3. [3]

    Codeevolve: An open source evolutionary coding agent for algorithm discovery and optimization.arXiv preprint arXiv:2510.14150, 2025

    Henrique Assump¸ c˜ ao, Diego Ferreira, Leandro Campos, and Fabricio Mu- rai. Codeevolve: An open source evolutionary coding agent for algorithm discovery and optimization.arXiv preprint arXiv:2510.14150, 2025

    arXiv 2025

  4. [4]

    Flow-based extremal mathematical structure discovery.arXiv preprint arXiv:2601.18005, 2026

    Gergely B´ erczi, Baran Hashemi, and Jonas Kl¨ uver. Flow-based extremal mathematical structure discovery.arXiv preprint arXiv:2601.18005, 2026. 12

    arXiv 2026

  5. [5]

    Global optimization for combinatorial geometry problems revisited in the era of llms.arXiv e-prints, pages arXiv–2601, 2026

    Timo Berthold, Dominik Kamp, Gioni Mexi, Sebastian Pokutta, and Imre P´ olik. Global optimization for combinatorial geometry problems revisited in the era of llms.arXiv e-prints, pages arXiv–2601, 2026

    2026

  6. [6]

    Early science acceleration experiments with gpt-5.arXiv preprint arXiv:2511.16072, 2025

    S´ ebastien Bubeck, Christian Coester, Ronen Eldan, Timothy Gowers, Yin Tat Lee, Alexandru Lupsasca, Mehtaab Sawhney, Robert Scherrer, Mark Sellke, Brian K Spears, et al. Early science acceleration experiments with gpt-5.arXiv preprint arXiv:2511.16072, 2025

    arXiv 2025

  7. [7]

    NuMVC: An efficient local search algorithm for minimum vertex cover.Journal of Artificial Intelligence Research, 46:687–716, 2013

    Shaowei Cai, Kaile Su, Chuan Luo, and Abdul Sattar. NuMVC: An efficient local search algorithm for minimum vertex cover.Journal of Artificial Intelligence Research, 46:687–716, 2013

    2013

  8. [8]

    Patternboost: Constructions in mathematics with a little help from ai.arXiv preprint arXiv:2411.00566, 2024

    Fran¸ cois Charton, Jordan S Ellenberg, Adam Zsolt Wagner, and Geordie Williamson. Patternboost: Constructions in mathematics with a little help from ai.arXiv preprint arXiv:2411.00566, 2024

    arXiv 2024

Show all 39 references
  1. [9]

    Barbarians at the gate: How AI is upending systems research.arXiv preprint arXiv:2510.06189, 2025

    Audrey Cheng, Shu Liu, Melissa Pan, Zhifei Li, Bowen Wang, Alex Krentsel, Tian Xia, Mert Cemri, Jongseok Park, Shuo Yang, et al. Barbarians at the gate: How AI is upending systems research.arXiv preprint arXiv:2510.06189, 2025

    2025

  2. [10]

    On the cuts and cut number of the 4-cube

    M Reza Emamy-Khansary. On the cuts and cut number of the 4-cube. Journal of Combinatorial Theory, Series A, 41(2):221–227, 1986

    1986

  3. [11]

    On the cut number problem for the 4, and 5-cubes.Discrete Applied Mathematics, 303:66–75, 2021

    M Reza Emamy-Khansary and R Arce-Nazario. On the cut number problem for the 4, and 5-cubes.Discrete Applied Mathematics, 303:66–75, 2021

    2021

  4. [12]

    New bounds for hypercube slicing numbers.Discrete Mathematics & Theoretical Computer Science, AA:155–164, 2001

    M Reza Emamy-Khansary and Martin Ziegler. New bounds for hypercube slicing numbers.Discrete Mathematics & Theoretical Computer Science, AA:155–164, 2001. Special Issue: DMTCS Proceedings vol. AA, Discrete Models: Combinatorics, Computation, and Geometry (DM-CCG 2001)

    2001

  5. [13]

    On the coverings of the d-cube ford≤6.Discrete Applied Mathematics, 156(17):3156–3165, 2008

    M Reza Emamy-Khansary and Martin Ziegler. On the coverings of the d-cube ford≤6.Discrete Applied Mathematics, 156(17):3156–3165, 2008

    2008

  6. [14]

    Mathematical exploration and discovery at scale.arXiv preprint arXiv:2511.02864, 2025

    Bogdan Georgiev, Javier G´ omez-Serrano, Terence Tao, and Adam Zsolt Wagner. Mathematical exploration and discovery at scale.arXiv preprint arXiv:2511.02864, 2025

    2025 arXiv

  7. [15]

    Random baselines for simple code problems are competitive with code evolution

    Yonatan Gideoni, Yujin Tang, Sebastian Risi, and Yarin Gal. Random baselines for simple code problems are competitive with code evolution. NeurIPS Fourth Workshop on Deep Learning for Code, 2025

    2025

  8. [16]

    The bunkbed conjecture is false.Proceedings of the National Academy of Sciences, 122(24):e2420725122, 2025

    Nikita Gladkov, Igor Pak, and Aleksandr Zimin. The bunkbed conjecture is false.Proceedings of the National Academy of Sciences, 122(24):e2420725122, 2025

    2025

  9. [17]

    Spectral properties of threshold functions

    Craig Gotsman and Nathan Linial. Spectral properties of threshold functions. Combinatorica, 14(1):35–50, 1994. 13

    1994

  10. [18]

    How to cut all edges of a polytope?The American Mathematical Monthly, 79(8):890–895, 1972

    Branko Gr¨ unbaum. How to cut all edges of a polytope?The American Mathematical Monthly, 79(8):890–895, 1972

    1972

  11. [19]

    Slicing all edges of an n-cube requires n 2/3 hyperplanes

    Ohad Klein. Slicing all edges of an n-cube requires n 2/3 hyperplanes. In 2023 IEEE 64th Annual Symposium on Foundations of Computer Science (FOCS), pages 1931–1936. IEEE, 2023

    2023

  12. [20]

    Can language models discover scaling laws?arXiv preprint arXiv:2507.21184, 2025

    Haowei Lin, Haotian Ye, Wenzheng Feng, Quzhe Huang, Yujun Li, Hubert Lim, Zhengrui Li, Xiangyu Wang, Jianzhu Ma, James Zou, et al. Can language models discover scaling laws?arXiv preprint arXiv:2507.21184, 2025

    2025

  13. [21]

    Finding increasingly large extremal graphs with alphazero and tabu search.arXiv preprint arXiv:2311.03583, 2023

    Abbas Mehrabian, Ankit Anand, Hyunjik Kim, Nicolas Sonnerat, Matej Balog, Gheorghe Comanici, Tudor Berariu, Andrew Lee, Anian Ruoss, Anna Bulanova, et al. Finding increasingly large extremal graphs with alphazero and tabu search.arXiv preprint arXiv:2311.03583, 2023

    2023 arXiv

  14. [22]

    Alphaevolve: A coding agent for scientific and algorithmic discovery.arXiv preprint arXiv:2506.13131, 2025

    Alexander Novikov, Ngˆ an V˜ u, Marvin Eisenberger, Emilien Dupont, Po- Sen Huang, Adam Zsolt Wagner, Sergey Shirobokov, Borislav Kozlovskii, Francisco JR Ruiz, Abbas Mehrabian, et al. Alphaevolve: A coding agent for scientific and algorithmic discovery.arXiv preprint arXiv:25...

    2025 arXiv

  15. [23]

    Slicing the n-cube

    Manuel Nowack. Slicing the n-cube. Master’s thesis, ETH Z¨ urich, Z¨ urich, Switzerland, January 2022. Available:https://gitlab.ethz.ch/ mnowack/slicing-n-cube

    2022

  16. [24]

    Hyperplane cuts of an n-cube.Discrete Mathematics, 1(2):193–195, 1971

    Patrick E O’Neil. Hyperplane cuts of an n-cube.Discrete Mathematics, 1(2):193–195, 1971

    1971

  17. [25]

    New ramsey multiplicity bounds and search heuristics.Foundations of Computational Mathematics, 25(5):1777–1814, 2025

    Olaf Parczyk, Sebastian Pokutta, Christoph Spiegel, and Tibor Szab´ o. New ramsey multiplicity bounds and search heuristics.Foundations of Computational Mathematics, 25(5):1777–1814, 2025

    2025

  18. [26]

    On threshold circuits for parity

    Ramamohan Paturi and Michael E Saks. On threshold circuits for parity. InProceedings [1990] 31st Annual Symposium on Foundations of Computer Science, pages 397–404. IEEE, 1990

    1990

  19. [27]

    Using large-scale experiments and machine learning to discover theories of human decision-making.Science, 372(6547):1209– 1214, 2021

    Joshua C Peterson, David D Bourgin, Mayank Agrawal, Daniel Reichman, and Thomas L Griffiths. Using large-scale experiments and machine learning to discover theories of human decision-making.Science, 372(6547):1209– 1214, 2021

    2021

  20. [28]

    Mathemati- cal discoveries from program search with large language models.Nature, 625(7995):468–475, 2024

    Bernardino Romera-Paredes, Mohammadamin Barekatain, Alexander Novikov, Matej Balog, M Pawan Kumar, Emilien Dupont, Francisco JR Ruiz, Jordan S Ellenberg, Pengming Wang, Omar Fawzi, et al. Mathemati- cal discoveries from program search with large language models.Nature, 625(799...

    2024

  21. [29]

    LLM-generated search heuristics can solve open in- stances of combinatorial design problems.MATH-AI Workshop at NeurIPS, 2025

    Christopher D Rosin. LLM-generated search heuristics can solve open in- stances of combinatorial design problems.MATH-AI Workshop at NeurIPS, 2025

    2025

  22. [30]

    Using code generation to solve open instances of combinatorial design problems.arXiv preprint arXiv:2501.17725, 2025

    Christopher D Rosin. Using code generation to solve open instances of combinatorial design problems.arXiv preprint arXiv:2501.17725, 2025

    2025 arXiv

  23. [31]

    Using reasoning models to generate search heuristics that solve open instances of combinatorial design problems.arXiv preprint arXiv:2505.23881, 2025

    Christopher D Rosin. Using reasoning models to generate search heuristics that solve open instances of combinatorial design problems.arXiv preprint arXiv:2505.23881, 2025

    2025 arXiv

  24. [32]

    Slicing the hypercube.Surveys in combinatorics, 1993:211– 255, 1993

    Michael Saks. Slicing the hypercube.Surveys in combinatorics, 1993:211– 255, 1993

    1993

  25. [33]

    Improved lower bound for hypercube edge slicing.arXiv preprint arXiv:2510.16592, 2025

    Lisa Sauermann and Zixuan Xu. Improved lower bound for hypercube edge slicing.arXiv preprint arXiv:2510.16592, 2025

    2025

  26. [34]

    Nondegenerate hyperplane covers of the hypercube.arXiv preprint arXiv:2507.00773, 2025

    Lisa Sauermann and Zixuan Xu. Nondegenerate hyperplane covers of the hypercube.arXiv preprint arXiv:2507.00773, 2025

    2025

  27. [35]

    Openevolve: an open-source evolutionary coding agent, 2025

    Asankhaya Sharma. Openevolve: an open-source evolutionary coding agent, 2025

    2025

  28. [36]

    Advancing geometry with ai: Multi-agent generation of polytopes.arXiv preprint arXiv:2502.05199, 2025

    Grzegorz Swirszcz, Adam Zsolt Wagner, Geordie Williamson, Sam Blackwell, Bogdan Georgiev, Alex Davies, Ali Eslami, Sebastien Racaniere, Theophane Weber, and Pushmeet Kohli. Advancing geometry with ai: Multi-agent generation of polytopes.arXiv preprint arXiv:2502.05199, 2025

    2025 arXiv

  29. [37]

    [Mastodon Post] - There are now sufficiently many dif- ferent examples of Erdos problems

    Terence Tao. [Mastodon Post] - There are now sufficiently many dif- ferent examples of Erdos problems.. https://mathstodon.xyz/@tao/ 115891256726420022, 2025. Mathstodon. Accessed: 2026-01-29

    2025

  30. [38]

    Slicing the hypercube is not easy.arXiv preprint arXiv:2102.05536, 2021

    Gal Yehuda and Amir Yehudayoff. Slicing the hypercube is not easy.arXiv preprint arXiv:2102.05536, 2021

    2021 arXiv

  31. [39]

    Computing cut numbers

    Martin Ziegler and Christian Sohler. Computing cut numbers. InProceedings of the 12th Canadian Conference on Computational Geometry (CCCG’00), 2000. 15 A Paterson construction for slicingQ 6 Solutions are listed such that each row corresponds to the vector of coefficients of a...

    2000

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.