REVIEW 3 major objections 5 minor 15 references
Region crossing change on surfaces
T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read On any closed orientable surface, region crossing change equivalence reduces to a homology rank computation.
desk verdict A useful generalization of region crossing change to surfaces, but the proof of the main rank formula has a real gap in the upper-bound half that needs repair before the counting formula can be trusted. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the modified incidence matrix $M_L$ over $\mathbb{Z}_2$: its rows are regions of $\Sigma_g \setminus L$, its columns are crossings, and entry $(i,j)$ records modulo 2 how many times region $i$ touches crossing $j$. Applying a modified region crossing change to a region corresponds exactly to adding that region's row; hence a sum of rows lists the crossings switched, so the row space generates all reachable diagrams. The formula for the row space's rank is controlled by $N_L$, the $n \times 2g$ matrix whose rows are the $\mathbb{Z}_2$-homology classes of the components. Lemmas 4.2 and 4.3 provide the bridge: a zero-sum dependency among regions is equivalent to a 2-colorable sublink, and a sublink is 2-colorable exactly when the sum of its component homology classes is trivial.
What would settle it
Take any link diagram on a genus-2 surface with specified component homology classes, write down its modified incidence matrix $M_L$, and compute $\operatorname{rank}_{\mathbb{Z}_2}(M_L)$ by elimination over $\mathbb{Z}_2$; compare it with $r - n - 1 + \operatorname{rank}_{\mathbb{Z}_2}(N_L)$. A mismatch, or an explicit diagram in which the $k+1$ dependency relations in the proof of Theorem 4.1 are linearly dependent, would refute the formula.
Extended reading notes
Core claim
On the paper's own terms, the discovery is Theorem 4.1: for a link diagram $L = K_1 \cup \cdots \cup K_n$ on $\Sigma_g$, the nullity $r - \operatorname{rank}_{\mathbb{Z}_2}(M_L)$ equals $n + 1 - \operatorname{rank}_{\mathbb{Z}_2}(N_L)$. Here $N_L$ is the $n \times 2g$ matrix over $\mathbb{Z}_2$ whose rows are the homology classes $[K_i]$ in $H_1(\Sigma_g; \mathbb{Z}_2)$. Since Proposition 2.4 counts the equivalence classes of diagrams with a fixed projection as $2^{c - \operatorname{rank}(M_L)}$, the formula gives $2^{c - r + n + 1 - \operatorname{rank}(N_L)}$ equivalence classes under modified region crossing change. The result extends the planar rank formula $r - n - 1$ and needs neither checkerboard colorability nor cellular embedding of the projection. Lemmas 4.2 and 4.3 carry the argument by translating zero-sum relations among region rows into 2-colorable sublinks.
Load-bearing premise
The load-bearing premise is that the $k$ zero-sum relations among regions coming from 2-colorable sublinks, together with the relation saying all regions sum to zero, are linearly independent; the paper only sketches this independence, and if it fails the lower bound on nullity, and with it the rank formula, could be wrong.
Editorial extensions
If this is right
- For a fixed projection with $c$ crossings and $r$ regions, the number of equivalence classes under modified region crossing change is $2^{\,c - r + n + 1 - \operatorname{rank}_{\mathbb{Z}_2}(N_L)}$.
- The quantity $r - \operatorname{rank}_{\mathbb{Z}_2}(M_L)$ is invariant under Reidemeister moves and depends only on the $\mathbb{Z}_2$-homology classes of the components, not on how the diagram is drawn.
- On the torus, the graph $G_K$ of a knot diagram is connected exactly when the knot bounds a disk (if the gcd of its winding coefficients is even) or lies in an annulus (if the gcd is odd).
- If a knot projection is cellularly embedded on a surface of genus $g > 0$, its equivalence graph $G_K$ is never connected; connectivity forces the surface to be the sphere.
Reading between the lines
- The same incidence-matrix method would apply to link diagrams on nonorientable surfaces or to virtual diagrams, with $N_L$ replaced by the appropriate $\mathbb{Z}_2$-homology matrix; the proof structure of Theorem 4.1 suggests the formula could survive in the same shape.
- The Section 5 examples show that for the original, single-counted region crossing change the rank depends on arithmetic details such as divisibility by 3, so a homology-only formula is unlikely; a precise count would need an invariant recording how regions self-touch.
- Because all connected components of $G_L$ are isomorphic, the component count supplied by the theorem determines the full equivalence-class structure: knowing the number of classes is knowing everything about reachability by modified region crossing changes.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies region crossing change for link diagrams on closed orientable surfaces, using a modified version in which a region incident to a crossing point multiple times switches that crossing multiple times mod 2. The main object is the Z_2 incidence matrix M_L whose row space records the effect of region crossing changes. The central result, Theorem 4.1, states that for an n-component link diagram L on Sigma_g, the rank of M_L equals r - n - 1 + rank_{Z_2}(N_L), where r is the number of regions and N_L is the n x 2g matrix of Z_2-homology classes of the components. Together with Proposition 2.4, this yields the number of equivalence classes of diagrams with the same projection under modified region crossing changes. The paper also proves Reidemeister invariance of r - rank(M_L) in Theorem 3.4, gives a homological characterization of 2-colorability in Lemma 4.2, and studies the original (unmodified) region crossing change in Section 5, including a lower bound on rank(M_L).
Significance. If the main formula is correct, Theorem 4.1 is a clean and useful homological description of the rank of the region-crossing-change incidence matrix on surfaces, extending the planar formula of Cheng and Gao. The counting consequence, Proposition 2.4 combined with Theorem 4.1, gives a direct way to compute the number of equivalence classes of link diagrams with a fixed projection under modified region crossing changes. The paper has concrete strengths: Theorem 3.4 is proved by explicit matrix checks that appear checkable; the examples in Section 4 illustrate the formula; and Section 5 honestly identifies that the formula fails for the original region crossing change, with an explicit counterexample. There are no fitted parameters or post-hoc selections, and the argument is derived from definitions and homology. The main weakness is a genuine gap in the proof of the upper-bound half of Theorem 4.1, which is the load-bearing step for the counting formula. The result is plausible and likely repairable, but the manuscript as written does not fully establish it.
major comments (3)
- [Section 4, proof of Theorem 4.1 (upper bound)] The upper-bound direction contains an unjustified assertion. Starting from k = r - rank(M_L), the proof invokes Lemma 4.3 to obtain k 2-colorable sublinks B_1, ..., B_k, then states: 'The reason why we drop the last equality is, the last equality \sum_{i in B_k} [K_i] = 0 can be obtained from the first k-1 equalities.' No derivation of this claim is given. The subsequent sentence about replacing R_k by the all-rows relation does not establish that B_k is redundant as a homology relation. If the remaining k-1 relations among the rows of N_L are dependent, then n - rank(N_L) < k-1, and the desired inequality r - rank(M_L) <= n + 1 - rank(N_L) is not obtained. Since this inequality is half of the central formula and drives the counting consequence in Proposition 2.4, this gap is load-bearing and must be fixed.
- [Section 4, proof of Theorem 4.1 (lower bound)] The lower-bound direction also relies on an unproved independence claim. After constructing k+1 linear relations coming from k 2-colorable sublinks together with the all-rows relation, the proof asserts that 'any equality of these can not be derived from the rest k equalities' and illustrates this by an arc of K_1. This is only a sketch: it does not prove that the coefficient vectors of the k+1 relations are linearly independent as elements of Z_2^r. A proper argument should show that no nontrivial Z_2-linear combination of the coefficient vectors vanishes, perhaps using the parity of region multiplicities along the components. As written, this is an assertion rather than a proof, and the lower bound r - rank(M_L) >= n + 1 - rank(N_L) depends on it.
- [Section 4, final independence paragraph] The final paragraph of the proof of Theorem 4.1 is internally unclear. It says 'if we put all the regions corresponding to ... together, counted with multiplicity, it is not difficult to observe that if two regions are adjacent then the number of times they appear in this multiset have the same parity' and then states 'we know that R_1 appears once in this multiset but R_k does not appear'. However, R_1 and R_k have not been defined at that point in the proof, and the parity claim is not demonstrated. Since this paragraph is the only argument for the independence used to conclude n - rank(N_L) >= k - 1, it needs to be rewritten as a explicit proof with all objects defined.
minor comments (5)
- [Corollary 3.8] The Euler characteristic arguments in Corollary 3.8 are informal. Phrases such as 'it is not difficult to observe' and 'Due to the Euler characteristic reason' occur several times; the case analysis in the first bullet, in particular, should spell out why the listed possibilities exhaust all configurations and why the excluded configurations contradict connectedness of K.
- [Corollary 3.8, second bullet] In the second bullet, 'R_1 is homeomorphic to T^2 with a disk moved' appears to be a typo: it should read 'removed'. Please correct this.
- [Section 3, Figure 4] The text says 'a region can appears i times around a crossing point' and the example lists i running over {0,1,2,3,4}; it would help to clarify whether i=0 means the region does not touch the crossing point at all, and to label the regions in Figure 4 so that the claimed multiplicities are visible.
- [Section 4, Example 4.4] In Example 4.4, the sentence 'rank_{Z_2}(N_L) = 2. According to Theorem 4.1, we obtain rank_{Z_2}(M_L) = 8 - (4+1-2) = 5' uses the formula correctly, but the preceding figure and the calculation of r = 8 are not fully documented; adding a brief explanation of how r is read from Figure 11 would improve readability.
- [Section 5, Remark 5.2] Remark 5.2 states that the regions in C are linearly independent 'as can be verified directly on the diagram' but does not provide the verification. Since this remark is not used in the main results, it can remain heuristic, but a short justification would be helpful.
Circularity Check
No circular derivation found; the surface rank formula is derived from incidence-matrix linear algebra and homology lemmas, not from its own conclusion.
full rationale
The paper's central claim, Theorem 4.1, computes rank_{Z2}(M_L) as r - n - 1 + rank_{Z2}(N_L). The proof does not fit parameters or rename a known result; it derives the lower bound from row dependencies among regions and Lemmas 4.2-4.3, and the upper bound from 2-colorable sub-links and homology relations in N_L. Lemma 4.2 characterizes 2-colorability by vanishing sum of homology classes, and Lemma 4.3 converts row dependencies into 2-colorable sub-links. These lemmas are proved directly from the definitions of the incidence matrix and region coloring, not from Theorem 4.1. The planar result Proposition 2.5 cites the authors' earlier paper [3], but the surface theorem does not assume the target formula; [3] is an external prior theorem whose planar statement is weaker than the main surface conclusion. The proof does contain an asserted linear-independence step in the upper-bound half of Theorem 4.1 that is only sketched, and a skeptical reader may question the 'drop the last equality' claim, but that is a completeness or correctness concern, not circularity. No quantity that is being predicted is used to define an input, and no fitted input is relabeled as a prediction. Accordingly, the circularity score is 0.
Assumptions & free parameters
assumptions (3)
- standard math The planar rank formula rank_{Z2}(M_L) = c - n + 1 for connected projections, and rank = r - n - 1 for planar links, proved in [3] and restated as Proposition 2.5.
- standard math For a graph embedded in a surface, V - E + sum_i chi(face_i) = chi(S), where faces may have nonzero Euler characteristic.
- standard math In Z2-homology, a 1-cycle is null-homologous if and only if it bounds a 2-chain formed by regions of the complement of the link diagram.
Cite this review
Pith. "Pith review of Region crossing change on surfaces." pith.science (2026). https://pith.science/paper/CJHQKU67
@misc{pith2026190806864,
author = {Pith},
title = {Pith review of: Region crossing change on surfaces},
year = {2026},
howpublished = {\url{https://pith.science/paper/CJHQKU67}},
note = {Machine review of arXiv:1908.06864}
}
abstract
Region crossing change is a local operation on link diagrams. The behavior of region crossing change on $S^2$ is well understood. In this paper, we study the behavior of (modified) region crossing change on higher genus surfaces.
Reference graph
Works this paper leans on
-
[1]
Knot Theory Ramifica- tions 21 (2012), no
Kazushi Ahara and Masaaki Suzuki, An integral region choice problem on knot projection , J. Knot Theory Ramifica- tions 21 (2012), no. 11, 1250119, 20 pp
work page 2012
-
[2]
Aida, Unknotting operation for Polygonal type , Tokyo J
H. Aida, Unknotting operation for Polygonal type , Tokyo J. Math. 15 (1992), no. 1, 111–121
work page 1992
-
[3]
7, 1487–1495
Zhiyun Cheng and Hongzhu Gao, On region crossing change and incidence matrix , Science China Mathematics 55 (2012), no. 7, 1487–1495
2012
-
[4]
, Mutation on knots and Whitneys 2-isomorphism theorem, Acta Mathematica Sinica, English Series 29 (2013), no. 6, 1219–1230
work page 2013
-
[5]
Zhiyun Cheng, When is region crossing change an unknotting operation? , Math. Proc. Cambridge Philos. Soc 155 (2013), no. 2, 257–269
2013
-
[6]
Russell, Equivalence of edge bicolored graphs on surfaces , Electron
Oliver Dasbach and Heather M. Russell, Equivalence of edge bicolored graphs on surfaces , Electron. J. Combin. 25 (2018), no. 1, Paper 1.59, 15 pp. 20 JIAWEI CHENG, ZHIYUN CHENG, JINWEN XU, AND JIEYAO ZHENG
work page 2018
-
[7]
207, Springer-V erlag, New York, 2001
Chris Godsil and Gordon Royle, Algebraic graph theory, Graduate Texts in Mathematics, vol. 207, Springer-V erlag, New York, 2001
work page 2001
-
[8]
Knot Theory Ramifications 24 (2015), no
Kenta Hayano, Ayaka Shimizu, and Reiko Shinjo, Region crossing change on spatial-graph diagrams, J. Knot Theory Ramifications 24 (2015), no. 8, 1550045, 12 pp
work page 2015
Show all 15 references
-
[9]
Knot Theory Ramifications 25 (2016), no
Ayumu Inoue and Ryo Shimizu, A subspecies of region crossing change, region freeze cross ing change, J. Knot Theory Ramifications 25 (2016), no. 14, 1650075, 9 pp
2016
-
[10]
Murakami, Some metrics on classical knots , Math
H. Murakami, Some metrics on classical knots , Math. Ann. 270 (1985), 35–45
1985
-
[11]
Murakami and Y
H. Murakami and Y . Nakanishi, On a certain move generating link-homology , Math. Ann. 284 (1989), 75–89
1989
-
[12]
4-5, 693–704
Martin Scharlemann, Crossing changes, Chaos Solitons Fractals 9 (1998), no. 4-5, 693–704
1998
-
[13]
Ayaka Shimizu, Region crossing change is an unknotting operation , J. Math. Soc. Japan 66 (2014), no. 3, 693–708
2014
-
[14]
Knot Theory Ramifica- tions 24 (2015), no
Daniel Silver and Susan Williams, On the component number of links from plane graphs , J. Knot Theory Ramifica- tions 24 (2015), no. 1, 1520002, 5 pp
2015
-
[15]
Hassler Whitney, 2-isomorphic graphs, Amer. J. Math. 55 (1933), 245–254. SCHOOL OF MATHEMATICAL SCIENCES , B EIJING NORMAL UNIVERSITY , B EIJING 100875, C HINA E-mail address: 201711130214@mail.bnu.edu.cn SCHOOL OF MATHEMATICAL SCIENCES , B EIJING NORMAL UNIVERSITY , B EIJING ...
1933
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.