Graphs related to $2$-dimensional simplex codes

Mariusz Kwiatkowski; Mark Pankov

REVIEW 2 major objections 4 minor 6 references

Graphs related to $2$-dimensional simplex codes

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper determines the distance relation on the graph of 2-dimensional 4-ary simplex codes: connected of diameter 3, with 162 vertices, degree 25, and distance layers of sizes 6 and 130 with an S5 symmetry group.

desk verdict Solid, narrow structural result for q=4 simplex-code graphs; the main case analysis holds, but a typo in Lemma 8 and several unshown finite checks need attention before acceptance. read the letter →

arxiv 1908.03703 v1 pith:TJIVJJ5G submitted 2019-08-10 math.CO

Mariusz Kwiatkowski , Mark Pankov This is my paper · ORCID

classification math.CO MSC 05C1294B0551E20

keywords simplexcodesGrassmanngraphdistancefinitefieldF4projectivegeometryHamminglinessymmetryorbits

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 the graph whose vertices are 2-dimensional simplex codes over the four-element field, with two codes adjacent when their intersection is a 1-dimensional subspace. It proves that this graph is connected, has diameter exactly 3, and has 162 vertices, each adjacent to 25 others. From any fixed vertex $L$, the six vertices at distance 3 form a set sharply 3-transitively permuted by the symmetry group fixing $L$, and the 130 vertices at distance 2 split into three classes of sizes 20, 90, and 20 with explicitly described adjacency. The description is complete because of a local rigidity that holds only for $q=4$: no three pairwise-adjacent simplex lines meet in three distinct points.

What carries the argument

The argument is carried by a local restriction proved for the field $\mathbb{F}_4$: there is no triple of mutually adjacent simplex lines whose three pairwise intersections are three distinct points (Proposition 3), so every maximal clique of the graph is the star of all simplex lines through one point. Combined with the adjacency equation of Proposition 2, this yields the dichotomy $n(L')\in\{0,1,3\}$ of Lemma 7, where $n(L')$ counts how many of the five hyperplanes $H_i$ a non-adjacent line $L'$ meets inside the coordinate hyperplane $C_i$. The three possible values of $n(L')$ separate the 130 distance-2 lines into $X^3_{20}$, $X^1_{90}$, and $X^0_{20}$, and the same local picture drives the $S_5$ orbit computation.

What would settle it

Enumerate the 135 simplex points in $\mathrm{PG}(4,4)$, form the 162 simplex lines, join two lines when they share a point, and check that every line has degree 25, exactly 6 vertices at distance 3, and exactly 130 at distance 2 with the claimed $X^3_{20}$, $X^1_{90}$, $X^0_{20}$ partition. The appendix's matrix lists make this a finite verification; a discrepancy in any count would refute Theorem 1.

Watch

Extended reading notes

Core claim

For $q=4$, the paper gives a complete description of the distance relation on the graph of 2-dimensional simplex codes, i.e. lines in $\mathrm{PG}(4,4)$ all of whose points are simplex points. The graph has 162 vertices, is connected, and has diameter 3. Fixing a vertex $L$, there are exactly 6 vertices at distance 3 from $L$, and exactly 130 at distance 2, partitioned as $X^3_{20}$ (20 vertices adjacent to three of the six), $X^1_{90}$ (90 vertices adjacent to one of them), and $X^0_{20}$ (20 vertices adjacent to none). The set $\{L,L_1,\ldots,L_6\}\cup X^0_{20}$ is a spread of all 135 simplex points. The stabilizer of $L$ is isomorphic to $S_5$: it acts sharply 3-transitively on the six distance-3 vertices, and its orbits on the remaining vertices have sizes 10, 15, 20, 20, 30, and 60.

Load-bearing premise

The whole description rests on one local fact about the four-element field: there is no triple of mutually touching simplex lines whose three pairwise meeting points are all different. If that fact fails, the proof's classification of distance-2 lines into 20, 90, and 20 collapses, and it does fail for every larger field.

Editorial extensions

If this is right

For every fixed vertex $L$, the graph is explicitly layered: 25 neighbors, 130 vertices at distance 2, 6 vertices at distance 3, and nothing farther away.
The 20 vertices in $X^0_{20}$, together with $L$ and the six $L_i$, form a spread of all 135 simplex points, so every simplex point lies on exactly one line of this 27-line spread.
The vertex stabilizer is $S_5$ and acts sharply 3-transitively on the six distance-3 vertices; the full orbit sizes on the graph are 1, 6, 10, 15, 20, 20, 30, and 60.
By duality, the same distance description transfers to the graph of 3-dimensional 4-ary Hamming codes, since simplex codes and Hamming codes are dual and adjacency is preserved under duality.
The appendix gives explicit generator matrices for every vertex in the nontrivial orbits, so the whole distance map can be verified line by line.

Reading between the lines

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

An intersection array is not computed in the paper; computing it would test whether this 162-vertex graph is distance-regular, since the layer sizes and orbit sizes are compatible with such a structure.
The mechanism is tied to $\mathbb{F}_4$ having exactly three nonzero elements, where a sum of three nonzero elements vanishes exactly when they are distinct; for $q\ge 5$, the paper's Example 3 shows the local restriction fails, so any analogue of Theorem 1 for larger fields would need a new counting principle rather than a minor modification.
The sharply 3-transitive $S_5$ action on the six distance-3 vertices suggests that the graph could be reconstructed as a Cayley graph on $S_5$ with a prescribed connection set, a compact presentation not given in the paper.

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

Here's my take on this one. The paper closes a small, clean case: for q=4 and k=2, it gives the complete distance-layer partition of the simplex-code graph and the orbit decomposition under G(L). The counts are new, and the structural work is largely sound. I agree with the reader's conditional verdict, and I'd want the typo and a few black-box checks addressed, but the mathematical core holds up.

What's genuinely new: previous work by these authors covered q=2,3 and abstract distance comparisons; the q=4 case is a real new computation. Theorem 1's 162 vertices, degree 25, diameter 3, distance layers 1/25/20/90/20/6, and the spread statement, plus Theorem 2's orbit sizes, are explicitly exhibited with matrices. The appendix's tables and the counting identity 162 = 1+25+20+20+90+6 close the enumeration. The load-bearing restrictions—Proposition 3's 'no triangle of lines meeting in three distinct points' and Lemma 7's n(L') in {0,1,3}—are argued in enough detail to be persuasive. I don't see a circular argument or fitted parameters.

Now the soft spots. The stress-test caught a real slip in Lemma 8's proof: after showing u is in G' but not in G(L), the text concludes G = G'. The correct conclusion is G = G'' (the order-6 subgroup), which is exactly what the lemma needs. So the claimed order 6 is recovered by repairing the typo; minor, but it shows the proofs are not fully polished. More substantively, several finite checks are asserted rather than displayed: the order-6 generation in Lemma 8, and the permutation/orbit-size claims in Propositions 9 and 10. The appendix gives full matrices, so an independent checker could confirm, but no machine-readable certificate or script is provided. If any of those matrix entries is wrong, Theorem 2's orbit sizes would change, though Theorem 1's distance-layer counts would not. The method is explicitly limited to q=4; that's a scope limitation, not a defect.

Who this is for: finite geometers and coding theorists working on Grassmann-graph restrictions. It is not a landmark, but it is a solid, defensible computation within an established program. It deserves a serious referee, not a desk rejection. I'd recommend sending it out, with the referee asked to check Lemma 8's repair and either verify the direct assertions or provide a short verification script. I'd cite it if I worked on simplex-code graphs.

Referee Report

2 major / 4 minor

Summary. The paper studies the graph Γ whose vertices are the 2-dimensional q-ary simplex codes for q = 4, viewed as simplex lines in PG(4,4). Theorem 1 claims that Γ has 162 vertices, is connected of diameter 3, is regular of degree 25, and that for every vertex L the 130 lines at distance 2 split into three disjoint classes X^3_20, X^1_90, X^0_20 with prescribed intersection behaviour with the six lines at distance 3; together with L and the six distance-3 lines, the class X^0_20 forms a spread of the 135 simplex points. Theorem 2 describes the action of the group G(L) on these classes, asserting orbits of sizes 6, 20, 20, 10, 15, 30 and 60. The proof is a finite coordinate-based case analysis over F_4, with explicit matrices for all exceptional lines collected in the appendix.

Significance. If correct, the paper gives a complete and non-trivial description of the distance layers and their symmetries for a subgraph of the Grassmann graph at a parameter value where earlier general theorems do not apply. The counting identity 162 = 1 + 25 + 20 + 20 + 90 + 6 closes the enumeration, and the appendix lists the exceptional lines explicitly, making the claims concretely checkable. The structural result Proposition 3, that every maximal clique is a star of lines through one point, is an interesting use of the special properties of F_4. On the other hand, the paper does not provide machine-checked proofs, and the proofs of the orbit decompositions in Theorem 2 are the least developed part of the argument.

major comments (2)

[§4.2, Proposition 9] The proof that the set A of lines adjacent to L is the union of two orbits of sizes 10 and 15 is incomplete. The text establishes that through each point of L there are two lines meeting some Lijk and three lines meeting no Lijk, and it exhibits local permutations of the five lines through P5. Transitivity of G(L) on the points of L makes the local counts constant, but it does not imply that the two lines through different points lie in the same orbit, nor that the three lines through different points lie in the same orbit. To justify the sentence 'The set A is the union of two orbits', the authors need to show transitivity of the point stabilizer on the relevant two lines (and on the relevant three lines) through a point, or supply a direct computational verification of the two orbits.
[§4.2, Proposition 10] The same transitivity gap occurs in the description of X^1_90. From the facts that G(L)∩G(L1) acts transitively on the points of L1 and that G(L) is transitive on {L1,...,L6}, it follows only that the two classes (lines intersecting some Lijk, and lines intersecting no Lijk) have constant size across the six Li; it does not follow that each class is a single G(L)-orbit. The conclusion that X^1_90 is the union of orbits of sizes 60 and 30 therefore needs an explicit transitivity proof or an independent computer check of the claimed orbit sizes.

minor comments (4)

[§4.2, Lemma 8] The concluding sentence 'Therefore, G = G′' contradicts the immediately preceding statement that u ∈ G′\G′′ and u ∉ G(L); the correct conclusion is G = G′′. The intended conclusion that |G| = 6 is recovered after this correction, but as written the proof contains a logical slip.
[§4.2, page 16] There is a typo in 'the projective transformation indued by p(2,3)(4,5)'; it should read 'induced'.
[§4.1, Lemma 3] In the sentence 'The fist two vectors satisfy (H4) and (H5)', 'fist' should be 'first'.
[§4.1, Lemma 7] The phrase 'If a,b,c are mutually distinct non-zero-elements of the field' should read 'non-zero elements of the field'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the q=4 distance-layer and G(L)-orbit classification is derived from the adjacency equation and explicit finite-field computations, with self-citations used only for background.

full rationale

The paper's central results (Theorem 1 and Theorem 2) are derived inside the paper from the definition of the Grassmann graph restricted to simplex lines. Proposition 2 gives an explicit coordinate equation for adjacency in F4, and Proposition 3 (no triangle of mutually adjacent lines with distinct pairwise intersections) is proved by normalizing one line and checking surviving possibilities with Proposition 2. Lemmas 3, 6, and 7 then establish the structural keystone n(L') in {0,1,3} by direct algebra over F4, and the counts 20, 90, 20 and the orbit sizes 10/15 and 30/60 follow by counting arguments and explicit monomial/semilinear transformations preserving L. No parameter is fitted from the target classification, and no asserted distance-layer property is used to define the objects whose properties are then 'predicted'. The self-citations [3] and [4] supply background on distances in subgraphs of the Grassmann graph and on projective codes; the q=4 claims do not reduce to those results. The use of PΓL(2,4) = S5 is a standard external group-theoretic fact, not an imported uniqueness theorem from the authors. A typographical slip in the last sentence of Lemma 8 ('Therefore, G = G′' where the preceding argument gives G = G′′) is a proof-correctness issue, not circularity, and does not feed any claimed result back into its own input.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No free parameters are fitted to data and no new physical or combinatorial entities are postulated. The only special ingredients are standard facts about F4 and PΓL(2,4), plus the paper's own proved lemmas, none of which are assumed without proof.

assumptions (6)

standard math For q=4, the q-1 non-zero field elements sum to zero, and a sum of three non-zero elements of F4 is zero if and only if they are mutually distinct.
Used in Proposition 2 to turn equation (1) into an exact adjacency criterion for simplex points over F4; this is the special field property that makes the q=4 case tractable and which fails for q>=5 (Remark 1).
standard math PΓL(2,4) is isomorphic to the symmetric group S5, and PGL(2,4) is isomorphic to A5, so every permutation of the five points of a simplex line extends to a unique projective transformation from the code automorphism group.
Used before Theorem 2 and throughout Section 4.2 to compute orbit sizes of the action of G(L) on distance layers; standard finite group fact.
standard math Maximal cliques of the Grassmann graph Γ_k(V) are stars and tops (Pankov, Geometry of Semilinear Embeddings, Proposition 3.3).
Used in Section 3 to frame the clique structure of Γ; the q=4-specific refinement is Proposition 3.
domain assumption A k-dimensional subspace C is a q-ary simplex code if and only if its projective system consists of all points of PG(k-1,q), equivalently all columns of a generator matrix are mutually non-proportional, and every non-zero codeword has exactly [k-1]_q zero coordinates.
Standard coding-theory characterization from Section 2.2, used to translate codes into lines formed by simplex points.
standard math Monomial linear automorphisms of V act transitively on simplex points and on simplex lines, and the code automorphism group of a simplex code is ΓL(k,q), with linear part GL(k,q).
Used to reduce proofs to a single line L and to count simplex lines via Proposition 1; the relevant count appears in Section 2.2.
standard math For two k-dimensional subspaces of V, the distance in the Grassmann graph Γ_k(V) is k - dim(X∩Y).
Section 2.1 defines the ambient distance that Γ restricts; needed to identify adjacency as intersection in a point for k=2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Graphs related to $2$-dimensional simplex codes." pith.science (2026). https://pith.science/paper/TJIVJJ5G

@misc{pith2026190803703,
  author       = {Pith},
  title        = {Pith review of: Graphs related to $2$-dimensional simplex codes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TJIVJJ5G}},
  note         = {Machine review of arXiv:1908.03703}
}

abstract

We give a complete description of the distance relation on the graph of $4$-ary simplex codes of dimension $2$. This is a connected graph of diameter $3$. For every vertex we determine the sets of all vertices at distance $i\in\{1,2,3\}$ and describe their symmetries.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

[1]

Bonisoli, Every equidistant linear code is a sequence of dual Hamming codes , Ars Combin

A. Bonisoli, Every equidistant linear code is a sequence of dual Hamming codes , Ars Combin. 18(1984), 181–186

work page 1984
[2]

Huﬀman, V

W.C. Huﬀman, V. Pless, Fundamentals of Error-Correcting Codes , Cambridge University Press, 2003

work page 2003
[3]

Kwiatkowski, M

M. Kwiatkowski, M. Pankov, On the distance between linear codes , Finite Fields Appl. 39 (2016), 251–263

work page 2016
[4]

Kwiatkowski, M

M. Kwiatkowski, M. Pankov, A. Pasini, The graphs of projective codes , Finite Fields and Their Applications 54(2018), 15–29

work page 2018
[5]

Pankov, Geometry of Semilinear Embeddings

M. Pankov, Geometry of Semilinear Embeddings. Relations to Graphs and Codes , World Scientiﬁc, 2015

work page 2015
[6]

Tsfasman, S

M. Tsfasman, S. Vlˇ adut ¸, D. Nogin,Algebraic Geometry Codes. Basic notions , Amer. Math. Soc., Providence, 2007. Faculty of Mathematics and Computer Science, University of Warmia and Mazury, S loneczna 54, Olsztyn, Poland E-mail address : mkw@matman.uwm.edu.pl, pankov@matman.uwm.edu.pl

work page 2007

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Bonisoli, Every equidistant linear code is a sequence of dual Hamming codes , Ars Combin

A. Bonisoli, Every equidistant linear code is a sequence of dual Hamming codes , Ars Combin. 18(1984), 181–186

work page 1984

[2] [2]

Huﬀman, V

W.C. Huﬀman, V. Pless, Fundamentals of Error-Correcting Codes , Cambridge University Press, 2003

work page 2003

[3] [3]

Kwiatkowski, M

M. Kwiatkowski, M. Pankov, On the distance between linear codes , Finite Fields Appl. 39 (2016), 251–263

work page 2016

[4] [4]

Kwiatkowski, M

M. Kwiatkowski, M. Pankov, A. Pasini, The graphs of projective codes , Finite Fields and Their Applications 54(2018), 15–29

work page 2018

[5] [5]

Pankov, Geometry of Semilinear Embeddings

M. Pankov, Geometry of Semilinear Embeddings. Relations to Graphs and Codes , World Scientiﬁc, 2015

work page 2015

[6] [6]

Tsfasman, S

M. Tsfasman, S. Vlˇ adut ¸, D. Nogin,Algebraic Geometry Codes. Basic notions , Amer. Math. Soc., Providence, 2007. Faculty of Mathematics and Computer Science, University of Warmia and Mazury, S loneczna 54, Olsztyn, Poland E-mail address : mkw@matman.uwm.edu.pl, pankov@matman.uwm.edu.pl

work page 2007