Point-transitive and 1-rotational unitals of order 5

Alex Ravsky; Ivan Hetman; Taras Banakh

REVIEW 3 major objections 5 minor 8 references

Point-transitive and 1-rotational unitals of order 5

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims a complete census of point-transitive and one-rotational unitals of order 5, with explicit difference families for all 1,311 designs.

desk verdict Useful cache of explicit unitals of order 5 with prescribed automorphism groups, but the census counts rest on undocumented computation and need code or a detailed algorithm before the completeness claim can be trusted. read the letter →

arxiv 2504.18390 v1 pith:HWG762LT submitted 2025-04-24 math.CO

Ivan Hetman , Taras Banakh , Alex Ravsky This is my paper · ORCID

classification math.CO MSC 05B0505B3020B2551E20

keywords unitalSteinersystemsdifferencefamiliespoint-transitivedesigns1-rotationalorder5finitegeometryhyperbolicfrequencyfingerprint

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 sets out to enumerate all unitals of order 5, meaning Steiner systems $S(2,6,126)$, whose automorphism groups either act transitively and effectively on the 126 points or fix one point and act transitively on the other 125. For every one of the 16 groups of order 126 and the 5 groups of order 125, it lists explicit difference families and reports how many non-isomorphic unitals each group produces. Summing the tables gives 1,214 point-transitive designs and 97 one-rotational designs, for a total of 1,311 unitals, including the classical and resolvable unitals previously known. If the enumeration is accepted, this is the first systematic census of these symmetry classes and by far the largest explicit algebraic collection of order-5 unitals.

What carries the argument

The load-bearing object is a difference family in a finite group: a collection of six-element base blocks such that every nonzero group element appears exactly once as a difference of two points inside some base block. In a group of order 126 acting regularly on the 126 points, or of order 125 acting with one singleton orbit $\{\infty\}$ and one regular orbit, such a family develops, through the left action and a fixed 0-based Cayley table, into the full block set of a Steiner system $S(2,6,126)$. The paper pairs each family with a hyperbolic frequency fingerprint, a count distribution over certain quadruples used as a quick distinguishing statistic, while recording cases where two non-isomorphic designs share the same fingerprint.

What would settle it

Re-run the difference-family search independently for one group, such as SmallGroup(126,10): enumerate all candidate six-element base blocks, develop each family into an $S(2,6,126)$, and compare the isomorphism classes with the 900 listed; finding one additional class, or identifying two listed classes as isomorphic, would refute the count.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is a complete difference-family census: up to isomorphism, every point-transitive unital of order 5 arises from one of the listed base-block collections in a group of order 126, and every one-rotational unital from a listed family in a group of order 125 with one point fixed. The counts allocate the designs across the 21 groups, with $C_6\times(C_7:C_3)$ (SmallGroup(126,10)) alone giving 900, $S_3\times(C_7:C_3)$ giving 129, and $\mathbb{Z}_{126}$ giving 64; seven groups of order 126 yield no designs. Two prior designs from the literature are reproduced inside the lists, and the paper records for each design a hyperbolic frequency fingerprint, which it uses as a distinguishing statistic while noting that the fingerprint is not by itself a complete invariant.

Load-bearing premise

The census counts are only as reliable as the unstated computer search and isomorphism check that found no missing base block and no false merges among the listed designs.

Editorial extensions

If this is right

If the census is exhaustive, the number of non-isomorphic unitals of order 5 in these two symmetry classes is exactly 1,214 + 97 = 1,311, recovering the previously known explicit examples and the earlier cyclic families.
Each tabulated difference family is a complete reconstruction recipe: applying the left action of the group to the listed base blocks yields the full block set of a Steiner system $S(2,6,126)$ without further search.
Seven groups of order 126 admit no such design, while $C_6\times(C_7:C_3)$ admits 900, so the table gives a sharp numerical profile of which group structures can carry an order-5 unital.
The overlap between the algebraic designs and the earlier projective-plane subdesign census is limited to at most three unitals from the Desarguesian plane of order 25, so almost all 1,311 designs are new relative to that census.

Reading between the lines

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

Because the paper's own Section 3 reports two non-isomorphic designs with identical fingerprints, the hyperbolic frequency fingerprint is not a complete invariant, and future automated extensions should combine it with a canonical-label or graph-isomorphism check.
The same difference-family recipe should transfer to other unital orders $k$, using groups of order $k^3+1$ for transitive designs and $k^3$ for one-rotational designs, with the bottleneck being a certified enumeration rather than block generation.
Most of the 1,311 designs probably do not embed in the Desarguesian plane of order 25; checking embeddability of the listed families would link this algebraic census to the earlier subdesign census.
The seven zero groups of order 126 invite a structural characterization of when a group admits a unital difference family, for example by comparing normal Sylow structure or abelianization across the zero and nonzero groups in the table.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The paper gives the first systematic sweep for unitals of order 5 over all groups of orders 125 and 126, together with explicit difference families and per-group counts. That is a real contribution: the construction is standard difference-family stuff, the left-action convention for non-commutative groups is sensible, and the listed families are concrete enough for a reader to verify that each produces an S(2,6,126). If the counts are right, this is a big jump from the 938 subdesign unitals of Stoichev–Gezek to 1311 additional designs, and the authors are appropriately careful about the small overlap with earlier work.

The soft spot is exactly where the stress-test note lands. The paper's central claim is not just existence of these unitals but completeness: for each group, exactly N designs up to isomorphism. The search over candidate difference families is never described, and the isomorphism reduction is never described. The fingerprint from the authors' earlier paper is explicitly admitted to collide, yet the only evidence for separating the two designs that share a fingerprint is a remark that "computer calculations" show non-isomorphism. No code, no algorithm, no canonical labelling procedure, no data file. A bug in that unstated computation, or an incomplete enumeration of base blocks, would change the reported counts without being detectable from the paper. The explicit families themselves are checkable, but the "exactly N" claim is not.

I do not think this is a case of a load-bearing flaw in the math; the construction itself is sound and the parameter checks pass. The flaw is in the evidence for exhaustiveness. That is fixable by releasing code or writing a precise description of the search and the isomorphism test, and the authors should be asked to do that before acceptance. The paper would also benefit from a cleaner write-up; the current text is rough, with typos and asides that do not help.

For a reader in finite geometry or design theory, the explicit difference families are useful regardless of the census counts, and the counts are plausible. But I would not rely on the exact numbers until the computation is reproducible. A serious referee should see this, and the right recommendation is to send it to review with a request for code or a full algorithm description, not to desk reject it.

Referee Report

3 major / 5 minor

Summary. The paper claims to give a complete enumeration, up to isomorphism, of unitals of order 5 (Steiner systems S(2,6,126)) whose automorphism group acts point-transitively, together with 1-rotational unitals admitting one fixed point and an orbit of size 125. For each of the 16 groups of order 126 and the 5 groups of order 125, it lists explicit difference families in GAP notation and tabulates the resulting numbers of non-isomorphic unitals, including counts such as 64 for Z126, 129 for SmallGroup(126,8), and 900 for SmallGroup(126,10). The paper also introduces a 'hyperbolic frequency fingerprint' invariant used to separate the designs, while noting that the fingerprint is not injective. The central assertion is that the displayed families produce, up to isomorphism, exactly the tabulated numbers of unitals, thus adding many new examples to the previously known ones.

Significance. If the census is correct, this is a valuable systematic enumeration: it would be the first complete classification of point-transitive and 1-rotational unitals of order 5, with explicit, independently checkable difference families and a nontrivial separating invariant. The listed families and fingerprints are concrete data that can be verified by any reader with a short script, and the paper honestly flags the known collision of its fingerprint. The strength of the contribution is, however, contingent on the completeness and exactness of the counts, which rest entirely on undocumented computation.

major comments (3)

[Section 1, Table 1; Sections 2 and 3] The central claim that the displayed difference families produce, up to isomorphism, exactly the tabulated numbers of unitals (e.g., 900 for SmallGroup(126,10), 129 for SmallGroup(126,8), 32 for Z5 x Z25) is not supported by any description of the enumeration algorithm, the search for base blocks, or the isomorphism reduction. The only hint in Section 3 says 'computer calculations show' that two designs with identical fingerprints are non-isomorphic, but no method, code, or certificate is supplied. Since the hyperbolic frequency fingerprint is explicitly admitted to collide, the 'exactly N' claims cannot be checked from the manuscript. Please provide a full algorithmic description of the exhaustive search, the invariant-based filtering, and the exact isomorphism check (or an independently runnable program), or else weaken the claims to 'at least N' and state that completeness is conjectural.
[Sections 2 and 3 (difference families)] For each listed difference family, the paper does not state how the S(2,6,126) property was verified, nor how the development of the family under a nonabelian group is defined beyond a brief remark about using left multiplication and the Cayley table. A difference family in a nonabelian group does not automatically give a Steiner system under the standard orbit development; one must prove that the union of the group-orbits of the base blocks has exactly 525 blocks and that every pair of points occurs in exactly one block. If any listed family fails this check, the corresponding count is wrong. Please include the verification method (e.g., a computer check with available code, or a short argument that the orbit sizes sum correctly) for every family in the tables.
[Section 1, Introduction] The paper does not compare the newly constructed unitals with the 938 unitals of order 5 found by Stoichev and Gezek as subdesigns of projective planes of order 25, even though the introduction explicitly notes that three unitals from the Desarguesian plane could appear in the lists. For a claim of an up-to-isomorphism census, such a comparison is necessary; otherwise the statement that the paper 'add[s] additional unitals' may double-count known designs. Please either perform the comparison and report which of the new families are isomorphic to the known subdesign unitals, or explicitly restrict the classification claim to unitals with the specified automorphism groups and state that the intersection with previously known unitals has not been determined.

minor comments (5)

[Throughout] The manuscript contains many typos and misspellings (e.g., 'authomorphism', 'nodoby', 'se ems', '1-rotat ional') and should be carefully proofread before submission.
[Section 1, fingerprint definition] The hyperbolic frequency fingerprint is defined only informally, and the example shows frequencies for counts 1 to 4, but many later fingerprints include a 0-frequency component (e.g., {0=1250, 1=42000, ...}). Please clarify the exact parameter range and how the value 0 is treated in the definition.
[References] The references are not visible in the manuscript text provided; in particular, the source for the classification of groups of order 126 and 125 (cited as [7]) and the exact contents of [1], [2], [3], and [6] should be fully specified.
[Section 1, Table 1] The labels 'Classical unital' for SmallGroup(125,3) and 'resolvable unital' for SmallGroup(125,5) are asserted without explanation or citation; please provide a justification or reference for these identifications.
[Section 3] For SmallGroup(126,6) and SmallGroup(126,16) the paper says 'we don't include them and just point to [1] and [2]', but the table gives counts for these groups. For a self-contained census, either reproduce the corresponding difference families or give precise statements of the cited results.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the difference-family construction and fingerprint invariant are independent of the census counts; the undocumented computer search is a reproducibility gap, not circular reasoning.

full rationale

The paper's core construction is standard: for each listed group it gives explicit difference families, and each family generates a Steiner system S(2,6,126); the hyperbolic frequency fingerprint is only a computed invariant used to label designs. No fitted parameter is used to produce the claimed counts, and no claim is defined in terms of another claim. The table's 'Number of unitals' asserts results of computer calculations that are not described; that is an underdocumentation and verifiability problem, not circularity, because the search is not shown to be equivalent to its own output. The self-citations are routine: [1] and [2] provide previously known designs for Z126 and Z2 x Z3 x Z3 x Z7 ('we don't include them and just point to [1] and [2]'), and [2] introduced the fingerprint, which the paper uses as a labelling device. None of these citations supplies the new enumeration or forbids alternatives. No equation, definition, or statistical reduction shows that a 'prediction' reduces to its own input, so no circular step can be exhibited.

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

The claim uses no free parameters and invents no new mathematical objects. It rests on standard difference-family theory, the GAP group lists, the fingerprint invariant from [2], and a critical unstated computational completeness assumption.

assumptions (5)

domain assumption The GAP SmallGroup IDs correctly enumerate all groups of order 125 and 126.
The paper relies on [7] for this classification; no group presentations are reproduced in the paper.
standard math A difference family over a group G that satisfies the pairwise balance condition generates an S(2,6,126) unital.
This is the textbook criterion for orbit designs; the paper invokes it implicitly when presenting difference families.
domain assumption The hyperbolic frequency fingerprint is invariant under isomorphism.
Introduced in [2]; the paper uses it to label designs, but notes in Section 3 that the fingerprint is not complete, so it is a heuristic.
ad hoc to paper The computer search for difference families was exhaustive and the isomorphism checks were correct.
No algorithm or code is provided; the enumeration counts in Sections 2 and 3 depend entirely on this unstated computational claim.
domain assumption At most three of the constructed unitals can coincide with the 938 unitals of Stoichev and Gezek.
The paper argues this from automorphism group orders (multiples of 125 or 126) rather than from an explicit isomorphism test against [6].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Point-transitive and 1-rotational unitals of order 5." pith.science (2026). https://pith.science/paper/HWG762LT

@misc{pith2026250418390,
  author       = {Pith},
  title        = {Pith review of: Point-transitive and 1-rotational unitals of order 5},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HWG762LT}},
  note         = {Machine review of arXiv:2504.18390}
}

abstract

In this paper we introduce enumeration of unitals of order $5$, which are also Steiner systems $S(2,6,126)$, where automorphism group acts transitively and effectively on points or fixes one point.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

8 extracted references · 1 canonical work pages

[1]

Hetman Steiner systems S(2,6,121/126), S(2,7,169) based on diﬀer ence families , 2024-2025

I. Hetman Steiner systems S(2,6,121/126), S(2,7,169) based on diﬀer ence families , 2024-2025. https://doi.org/10.48550/arXiv.2401.08274

arXiv doi:10.48550/arxiv.2401.08274 2024
[2]

Hetman Point-transitive Steiner systems S(2,6,111/121/126), S( 2,7,169/175), 2025

I. Hetman Point-transitive Steiner systems S(2,6,111/121/126), S( 2,7,169/175), 2025. https://doi.org/10.48550/arXiv.2504.14931

arXiv doi:10.48550/arxiv.2504.14931 2025
[3]

C. J. Coulborn, J. H. Dinitz Handbook of Combinatorial Designs, Second Edition , 2007

2007
[4]

Krcadinac personal website https://web.math.pmf.u nizg.hr/ krcko/results/steiner.html

V. Krcadinac personal website https://web.math.pmf.u nizg.hr/ krcko/results/steiner.html
[5]

D´ avid Mez˝ oﬁ, G´ abor Nagy list of paramodiﬁed unitals https://davidmezoﬁ.github.io/unitals/
[6]

Unitals in Projective Planes o f Order 25

Stoichev, S.D., Gezek, M. Unitals in Projective Planes o f Order 25. Math.Comput.Sci. 17, 5 (2023). https://doi.org/10.1007/s11786-023-00556-9

work page doi:10.1007/s11786-023-00556-9 2023
[7]

The GAP Group, GAP – Groups, Algorithms, and Programming, Version 4.14.0 ; 2024, https://www.gap-system.org

2024
[8]

Nagy, G. P. and Vojtˇ echovsk´ y, P., loops, Computing with quasigroups and loops in GAP, Version 3.4.4 (2024) (GAP package), https://gap-packages.github.io/loops/

2024

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Hetman Steiner systems S(2,6,121/126), S(2,7,169) based on diﬀer ence families , 2024-2025

I. Hetman Steiner systems S(2,6,121/126), S(2,7,169) based on diﬀer ence families , 2024-2025. https://doi.org/10.48550/arXiv.2401.08274

arXiv doi:10.48550/arxiv.2401.08274 2024

[2] [2]

Hetman Point-transitive Steiner systems S(2,6,111/121/126), S( 2,7,169/175), 2025

I. Hetman Point-transitive Steiner systems S(2,6,111/121/126), S( 2,7,169/175), 2025. https://doi.org/10.48550/arXiv.2504.14931

arXiv doi:10.48550/arxiv.2504.14931 2025

[3] [3]

C. J. Coulborn, J. H. Dinitz Handbook of Combinatorial Designs, Second Edition , 2007

2007

[4] [4]

Krcadinac personal website https://web.math.pmf.u nizg.hr/ krcko/results/steiner.html

V. Krcadinac personal website https://web.math.pmf.u nizg.hr/ krcko/results/steiner.html

[5] [5]

D´ avid Mez˝ oﬁ, G´ abor Nagy list of paramodiﬁed unitals https://davidmezoﬁ.github.io/unitals/

[6] [6]

Unitals in Projective Planes o f Order 25

Stoichev, S.D., Gezek, M. Unitals in Projective Planes o f Order 25. Math.Comput.Sci. 17, 5 (2023). https://doi.org/10.1007/s11786-023-00556-9

work page doi:10.1007/s11786-023-00556-9 2023

[7] [7]

The GAP Group, GAP – Groups, Algorithms, and Programming, Version 4.14.0 ; 2024, https://www.gap-system.org

2024

[8] [8]

Nagy, G. P. and Vojtˇ echovsk´ y, P., loops, Computing with quasigroups and loops in GAP, Version 3.4.4 (2024) (GAP package), https://gap-packages.github.io/loops/

2024