REVIEW 5 minor 20 references
Flag complexes and homology
T0 review · 0 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read Every flag complex's f-vector can be realized by a balanced complex with no smaller top homology, yielding sharp bounds on face numbers from Betti numbers.
desk verdict A genuinely new homology-strengthening of Frohmader's theorem with sharp bounds; the proofs are sound and the paper deserves refereeing. 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 Turán complex $\Delta(T_d(n))$, the clique complex of the complete $d$-partite graph on $n$ vertices with parts as equal as possible; its face numbers are denoted $\binom{n}{k}_d$ and appear in the canonical representations that parametrize the $f$-vectors of balanced complexes. The paper's main technical theorem (Theorem 3.5) says that, among balanced complexes with a fixed number of $(k-1)$-faces, the color-shifted revlex one has the largest top Betti number, with value given explicitly by the canonical-representation expression $\sum_j\binom{N_{d-j}-(d-j)}{d-j}_{d-j}$. The proof pivots on a formula (Theorem 3.1) that identifies the top Betti number of a pure color-shifted balanced complex with the number of top faces avoiding the least vertex of each color class; the paper extends this to non-pure complexes by observing that top-degree chains and cycles depend only on the top faces. A second bridge, a theorem that every flag complex's $f$-vector is realized by a revlex balanced complex, lets these balanced-complex bounds be transported back to flag complexes.
What would settle it
Enumerate all small color-shifted balanced complexes and compare, for each one, the top reduced Betti number computed over a field with the number of top faces that avoid the least vertex in every color class. Any complex where the two numbers differ would refute the non-pure extension of the formula (Corollary 3.3) on which Theorem 3.5, and hence Theorems 1.2, 1.3, and 1.5, depend; the paper's own construction of $\hat\Delta$ inside the proof of Theorem 3.5 is the natural place to probe for such a divergence.
Extended reading notes
Core claim
On the paper's own terms, the central discovery is that the top reduced Betti number of a flag complex is governed by its face numbers through canonical representations. Theorem 1.1 asserts that the known realization of flag-complex $f$-vectors by balanced complexes can be upgraded: the balanced realization $\Gamma$ can be chosen with the same $f$-vector and $\beta_d(\Gamma)\ge\beta_d(\Delta)$. The engine behind the numerical consequences is Theorem 3.5: among all balanced complexes with a fixed number $N$ of $(k-1)$-faces, the revlex balanced complex maximizes the top Betti number, and the maximum is the explicit sum $\sum_j\binom{N_{d-j}-(d-j)}{d-j}_{d-j}$ read off from the $(k,d)$-canonical representation of $N$. Applying this to flag complexes gives Theorem 1.2's upper bound on $\beta_{d-1}(\Delta)$ and, by inverting it, Theorem 1.3's lower bounds on $f_{i-1}(\Delta)$ in terms of $\beta_{d-1}(\Delta)=a$. The continuous inequality $f_\Delta(x)\ge(1+(\sqrt[d]{a}+1)x)^d$ is the polynomial shadow of these bounds, and its equality cases are Turán complexes.
Load-bearing premise
The chain of bounds rests on one exact formula: for a color-shifted balanced complex, the top reduced Betti number equals the number of top-dimensional faces avoiding the smallest vertex of every color class. The formula was originally stated without the purity condition and later corrected, so the paper's non-pure extension depends on the observation that top homology sees only top faces; if that extension is false, the upper bounds and the lower bounds built from them collapse.
Editorial extensions
If this is right
- Because every flag complex shares its $f$-vector with a balanced complex of no smaller top Betti number, any homology-aware face-number inequality proved for balanced complexes applies verbatim to flag complexes.
- The top Betti number $\beta_{d-1}(\Delta)$ is bounded above by the explicit canonical-representation sum determined by any face number $f_{k-1}(\Delta)$, so large homology is impossible without many faces.
- Given $\beta_{d-1}(\Delta)=a$, each $f_{i-1}(\Delta)$ is at least $\sum_j\binom{a_{d-j}+d-j}{i-j}_{d-j}$; if equality holds at one index $i\ge s+1$, equality holds at all larger indices, forcing the whole tail of the $f$-vector.
- The $f$-polynomial inequality $f_\Delta(x)\ge(1+(\sqrt[d]{a}+1)x)^d$ holds coefficientwise; when $\sqrt[d]{a}$ is an integer the unique equality case is the Turán complex $\Delta(T_d(d(\sqrt[d]{a}+1)))$.
- Comparing with any Turán complex $T$ whose top Betti number is at most $a$ yields $f_i(\Delta)\ge f_i(T)$ for all $i$; equality in $f_0$ alone forces $\Delta\cong T$ (provided $\beta_{d-1}(T)=a$).
Reading between the lines
- A natural testable extension is to check Conjecture 6.2, that the same canonical-representation lower bound holds for homology in every dimension $k$, not just the top; a computer search over small flag complexes with prescribed $\beta_{k-1}$ would give evidence before any proof.
- The theorem effectively turns a homology computation into a face-count test: to rule out $\beta_{d-1}\ge a$, it suffices to check that some face number lies below the Turán-type bound, which could make homology estimation for large clique complexes purely combinatorial.
- Equality-case rigidity suggests a stability phenomenon: flag complexes with top Betti number close to $a$ should have face vectors close to the Turán complex; quantifying that slack, rather than exact equality, is an open direction the paper does not pursue.
- Because the balanced realization preserves the $f$-vector, a future characterization of $(f,\beta)$-vectors of balanced complexes would immediately produce restrictions on flag complexes; the authors explicitly leave this as Problem 6.1.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves structural results linking the f-vector and the top-dimensional reduced Betti number of flag complexes. Theorem 1.1 states that every d-dimensional flag complex has the same f-vector as some balanced complex whose d-th reduced Betti number is at least that of the flag complex, thereby extending Frohmader's f-vector realization theorem. The paper then derives Theorem 1.2, a sharp upper bound on beta_{d-1}(Delta) in terms of the (k,d)-canonical representation of f_{k-1}(Delta); Theorem 1.3, a sharp lower bound on the face numbers of a flag complex in terms of beta_{d-1}(Delta), refining Meshulam's theorem; and Theorem 1.5, the coefficient-wise inequality f_Delta(x) >= (1 + (d-th root of a + 1)x)^d when beta_{d-1}(Delta)=a. The proofs combine color-shifted balanced complexes, the Frankl-Furedi-Kalai characterization, Murai's correction of the Babson-Novik formula, and an inductive Mayer-Vietoris construction.
Significance. If accepted, the results give the first homology-aware extension of Frohmader's theorem, with sharp bounds and rigidity statements in terms of Turan complexes. Strengths of the manuscript are its detailed proofs, its explicit treatment of the Babson-Novik non-pure pitfall, and the fully worked canonical-representation lemmas (Lemma 2.5). The paper's reliance on the Murai-corrected formula is contained to the top degree, where the non-pure extension via Corollary 3.3 is justified by the observation that top homology depends only on top faces; the stress-test concern about this point does not land. I found no load-bearing gap.
minor comments (5)
- [Section 3, proof of Theorem 3.5] The text refers to 'Proposition 2.3' when deriving the bound on L; this should be 'Lemma 2.3'.
- [Section 4, Theorem 4.2] The complex Sigma_0 is used in equation (18) before it is defined; please define Sigma_0 immediately after equation (16) as the revlex d-colorable complex obtained from the induction hypothesis applied to Lk_Delta(v_0).
- [Section 4, proof of Theorem 1.1] Theorem 2.9 (Frohmader) is stated as an existence result, so the word '(unique)' in 'there exists a (unique) revlex balanced complex Gamma' is not justified by the cited theorem; either remove it or provide a citation for uniqueness.
- [Abstract and Theorem 1.5] The abstract and Theorem 1.5 use the notation \sqrt[d]{a}, while the body uses 'd\sqrt{a}'; please standardize the root notation.
- [Section 5, proof of Theorem 1.5] The coefficient-wise inequality is derived by summing over k in [0,d]; it would help readers if the role of the empty face (f_{-1}=1) in the constant coefficient were stated explicitly.
Circularity Check
No circularity: all main results reduce to external theorems (Zykov, Frankl-Furedi-Kalai, Frohmader, Babson-Novik-Murai), with one non-load-bearing self-citation.
full rationale
The paper's derivation chain is self-contained against external results. Theorem 1.1 follows from Theorem 4.2 and Frohmader's theorem; Theorem 4.2 is an induction that uses Lemma 4.1 (Mayer-Vietoris) and Theorem 3.5; Theorem 3.5 uses the Frankl-Furedi-Kalai characterization of f-vectors of colored complexes, Zykov's theorem, and Murai's colored algebraic shifting inequality together with the corrected Babson-Novik formula in top degree. None of these inputs contains the paper's conclusions. The non-pure extension in Corollary 3.3 is justified by the explicit observation that top-degree chains and cycles depend only on top faces, so it is not an assumption of the conclusion. The single self-citation [7] appears only as a pointer for canonical representations, alongside external references [14] and [13], and is never used as an argument in the proofs of Theorems 1.1-1.5. Thus the central claims have independent content and the minor self-citation is not load-bearing.
Assumptions & free parameters
assumptions (6)
- standard math Zykov's generalization of Turán's theorem (Theorem 2.4): for a flag complex with n vertices, f_i(Δ) ≤ f_i(Δ(T_d(n))).
- standard math Frankl-Füredi-Kalai theorem (Theorem 2.8): a vector is an f-vector of an r-colorable complex iff it satisfies certain shadow inequalities, equivalently iff it is realizable by a revlex r-colorable complex.
- standard math Frohmader's theorem (Theorem 2.9): every flag complex has the same f-vector as a revlex d-colorable complex.
- standard math Babson-Novik/Murai formula (Theorem 3.1): for a pure color-shifted balanced (d-1)-complex, the top reduced Betti number equals the number of top faces avoiding all minimal vertices of each color.
- standard math Murai's theorem: for any balanced complex Γ, β_i(Γ) ≤ β_i(Δ_≺(Γ)) for colored algebraic shifting.
- standard math Frankl-Füredi-Kalai continuous analog (Theorem 5.3): if (r choose k) α^k = f_{k-1}(Δ), then f_{j-1}(Δ) ≥ (r choose j) α^j.
Cite this review
Pith. "Pith review of Flag complexes and homology." pith.science (2026). https://pith.science/paper/JMHF5B3Q
@misc{pith2026190808308,
author = {Pith},
title = {Pith review of: Flag complexes and homology},
year = {2026},
howpublished = {\url{https://pith.science/paper/JMHF5B3Q}},
note = {Machine review of arXiv:1908.08308}
}
abstract
We prove several relations on the $f$-vectors and Betti numbers of flag complexes. For every flag complex $\Delta$, we show that there exists a balanced complex with the same $f$-vector as $\Delta$, and whose top-dimensional Betti number is at least that of $\Delta$, thereby extending a theorem of Frohmader by additionally taking homology into consideration. We obtain upper bounds on the top-dimensional Betti number of $\Delta$ in terms of its face numbers. We also give a quantitative refinement of a theorem of Meshulam by establishing lower bounds on the $f$-vector of $\Delta$, in terms of the top-dimensional Betti number of $\Delta$. This result has a continuous analog: If $\Delta$ is a $(d-1)$-dimensional flag complex whose $(d-1)$-th reduced homology group has dimension $a\geq 0$ (over some field), then the $f$-polynomial of $\Delta$ satisfies the coefficient-wise inequality $f_{\Delta}(x) \geq (1 + (\sqrt[d]{a}+1)x)^d$.
Reference graph
Works this paper leans on
-
[7]
Kai Fong Ernest Chong. Hilbert functions of colored quot ient rings and a generalization of the Clements-Lindstr¨ om theorem. J. Algebraic Combin. , 42(1):1–23, 2015
work page 2015
-
[1]
Face numbers and nongene ric initial ideals
Eric Babson and Isabella Novik. Face numbers and nongene ric initial ideals. Electron. J. Combin. , 11(2):Research Paper 25, 23 pp. (electronic), 2004/06
work page 2004
-
[2]
An extended Euler-Poinca r´ e theorem.Acta Math
Anders Bj¨ orner and Gil Kalai. An extended Euler-Poinca r´ e theorem.Acta Math. , 161(3-4):279–303, 1988
work page 1988
-
[3]
Extended Euler-Poincar´ e relations for cell complexes
Anders Bj¨ orner and Gil Kalai. Extended Euler-Poincar´ e relations for cell complexes. In Applied geometry and discrete mathematics, volume 4 of DIMACS Ser. Discrete Math. Theoret. Comput. Sci. , pages 81–89. Amer. Math. Soc., Providence, RI, 1991
work page 1991
-
[4]
Anders Bj¨ orner and Michelle L. Wachs. Shellable nonpur e complexes and posets. I. Trans. Amer. Math. Soc. , 348(4):1299–1327, 1996
work page 1996
-
[5]
Anders Bj¨ orner and Michelle L. Wachs. Shellable nonpur e complexes and posets. II. Trans. Amer. Math. Soc. , 349(10):3945–3975, 1997
work page 1997
-
[6]
Cohen-Macaulay rings , volume 39 of Cambridge Studies in Advanced Mathe- matics
Winfried Bruns and J¨ urgen Herzog. Cohen-Macaulay rings , volume 39 of Cambridge Studies in Advanced Mathe- matics. Cambridge University Press, Cambridge, 1993
work page 1993
-
[8]
On the h-vectors of Cohen-Macaulay flag complexes
Alexandru Constantinescu and Matteo Varbaro. On the h-vectors of Cohen-Macaulay flag complexes. Math. Scand., 112(1):86–111, 2013
work page 2013
Show all 20 references
-
[9]
Art M. Duval. On f -vectors and relative homology. J. Algebraic Combin. , 9(3):215–232, 1999
1999
-
[10]
A new Tur´ an-type theorem for cliques i n graphs
J¨ urgen Eckhoff. A new Tur´ an-type theorem for cliques i n graphs. Discrete Math. , 282(1-3):113–122, 2004
2004
-
[11]
P. Erd˝ os. On the number of complete subgraphs containe d in certain graphs. Magyar Tud. Akad. Mat. Kutat´ o Int. K¨ ozl., 7:459–464, 1962
1962
-
[12]
Shadows of colored complexes
Peter Frankl, Zolt´ an F¨ uredi, and Gil Kalai. Shadows of colored complexes. Math. Scand. , 63(2):169–178, 1988
1988
-
[13]
Face vectors of flag complexes
Andrew Frohmader. Face vectors of flag complexes. Israel J. Math. , 164:153–164, 2008
2008
-
[14]
Goodman and Joseph O’Rourke, editors
Jacob E. Goodman and Joseph O’Rourke, editors. Handbook of discrete and computational geometry . Discrete Mathematics and its Applications (Boca Raton). Chapman & Ha ll/CRC, Boca Raton, FL, second edition, 2004
2004
-
[15]
Domination numbers and homology
Roy Meshulam. Domination numbers and homology. J. Combin. Theory Ser. A , 102(2):321–330, 2003
2003
-
[16]
Betti numbers of strongly color-stable ideals and squarefree strongly color-stable ideals
Satoshi Murai. Betti numbers of strongly color-stable ideals and squarefree strongly color-stable ideals. J. Algebraic Combin., 27(3):383–398, 2008
2008
-
[17]
Richard P. Stanley. Combinatorics and commutative algebra , volume 41 of Progress in Mathematics . Birkh¨ auser Boston Inc., Boston, MA, 1983
1983
-
[18]
Eine Extremalaufgabe aus der Graphentheo rie
Paul Tur´ an. Eine Extremalaufgabe aus der Graphentheo rie. Mat. Fiz. Lapok , 48:436–452, 1941
1941
-
[19]
G¨ unter M. Ziegler. Lectures on polytopes , volume 152 of Graduate Texts in Mathematics . Springer-Verlag, New York, 1995
1995
-
[20]
A. A. Zykov. On some properties of linear complexes. Mat. Sbornik N.S. , 24(66):163–188, 1949. Singapore University of Technology and Design, Singapore E-mail address : ernest chong@sutd.edu.sg Einstein Institute of Mathematics, Hebrew University of Je rusalem, Israel E-mail a...
1949
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.