Character Nonvanishing under Coprime Actions

arXiv:2609.16227 · math.GR · Submitted 2026-07-30 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: We're looking at the paper: "Character Nonvanishing under Coprime Actions".

Emmy: Here's the gist of the paper. Let A finite group act coprimely on a finite group G and put C = CG(A).

Ada: A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always C/C′?

Emmy: This would follow if one could show that χC is never zero if and only if the Glauberman–Isaacs correspondent χ∗ of χ is linear The Counterexample Construction Theorem A states that there exist a finite group A ∼= C21 and a finite 2-group G of order 2135, with A acting faithfully and coprimely on G, such that NA(G) = 1728 and CG(A)/CG(A)′ = 4096 In particular the equality of Problem 21.100 fails The group is G = B ⋊ V, where V = F24 ⊕ F8 and B is the group of Boolean functions on V, and B is the additive group of all Boolean functions on V The Mechanism of Failure The strategy for proving Theorem A involves isolating a counterexample where three facts do all the work First, C = CG(A) = BA is elementary abelian of rank the number of A-orbits on V (Proposition 2.2) Second, the little-groups method identifies IrrA(G) with BA (Proposition 2.5), so that IrrA(G) = C/C′ exactly; hence a single invariant character with a zero on C already refutes Problem 21.100 Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V, the associated χu ∈ IrrA(G) has degree V and satisfies χu(c) = V − 2supp c for c ∈ BA (Lemma 2.10) Everything therefore reduces to an arithmetic question about the multiset of A-orbit sizes on V: does some sub-multiset sum to V/2 The Arithmetic Constraint The condition for a character vanishing at some element of C is determined by Corollary 2.9, which states that if some χ ∈ IrrA(G) vanishes at some element of C, then some sub-multiset of the A-orbit sizes on V sums to V/2 The smallest admissible cyclic pair is A = 21 with dim V = 7, where the orbit sizes are forced to be 1, 3, 3, 3, 3, 3, 7, and the sub-multisets summing to V/2 are exactly the twenty of the forms The Final Count Theorem A concludes that for this action there is a χ ∈ IrrA(G) of degree 128 and a c ∈ C with χ(c) = 0 This implies NA(G) ≤ 4095 < 4096 = C/C′, and the equality of Problem 21.100 fails The exact count is given by Theorem 7.1, showing that NA(G) = 1728 and CG(A)/CG(A)′ = 4096 Consequences for Related Problems Corollary B shows that the proposed criterion of Problem 6.3 fails because there are 2368 characters χ ∈ IrrA(G) with χ∗(1) = 1 for which χC nevertheless has a zero Corollary D demonstrates that this failure persists when passing to direct powers, as the ratio NA(Gn)/Cn/(Cn)' approaches 0 as n → ∞ The smallest dimension for a balanced pair (A, V) is 7 The Exact Verification Verification confirms that three programs compute NA(G) by different routes—a 128-point integer Walsh transform, a direct computation in F24 ⊕ F8 from the radial correlation table, and a binary linear-algebra test built on the twenty zero patterns of Lemma 6.1—and all three return 1728 The final script confirms that K = C × A is a Carter subgroup of Γ, as Proposition 9.1 asserts The Minimal Case Remark 3.7 notes that the smallest balanced pair (A, V) has dim V = 8 with A = 15 and element orders This confirms that the construction of Section 4 produces a counterexample to Problem 21.100 if and only if V has an A-stable subset of size V/2 The smallest order of a balanced operator group is A = 15, attained at dim V = 8 The General Principle The paper establishes that nonvanishing on C is not determined by χ∗ alone, and that any repair of Problem 6.3 must either restrict the operator group — to a p-group, say — or use more of χ than its correspondent The p-group case is the one that survives, and it is the natural remaining question The Exact Zero Criterion Lemma 6.1 states that expression (10) vanishes if and only if (H0, H1) = (δ + εJ, ρR + εJ) for some ε ∈ F2 and some three-element subset R ⊆ This corresponds bijectively to the twenty A-stable subsets of V of size 64 found in Proposition 3.3 The final count is NA(G) = 1728 and CG(A)/CG(A)′ = 4096 The Carter Subgroup Analogue Theorem 9.4 shows that the number of characters of Γ that do not vanish at any element of K is 21 NA(G) = 36288, whereas K/K′ = 86016 This proves Problem 6.7 of

12: has a negative answer The head characters of Γ are not the Carter-nonvanishing irreducible characters of Γ The General Bound Theorem 9.7 shows that there are at most A NA(G) such characters, and since NA(G) 36288 This proves Problem 6.7 has a negative answer for Γ, and the head characters of Γ are not its Carternonvanishing characters The smallest we obtain is the one of Theorem 9.4, of order 2135 · 21 The Inequality The inequality NA(G) ≤ C/C′ always holds in our examples, and holds trivially whenever C is abelian, since then C/C′ = Irr(C) = IrrA(G) ≥ NA(G) We know of no general argument The Repair The paper suggests that a repair of Problem 6.3 must either restrict the operator group — to a p-group, say — or use more of χ than its correspondent The p-group case is the one that survives, and it is the natural remaining question The Verification Verification confirms that three programs compute NA(G) by different routes—a 128-point integer Walsh transform, a direct computation in F24 ⊕ F8 from the radial correlation table, and a binary linear-algebra test built on the twenty zero patterns of Lemma 6.1—and all three return 1728 The final script confirms that K = C × A is a Carter subgroup of Γ, as Proposition 9.1 asserts The smallest we obtain is the one of Theorem 3.5 and Corollary 3.6, which shows that dim V = 7 is the least attainable dimension in which the construction of Section 2 produces a counterexample, over all operator groups of odd order The smallest order of a balanced operator group is A = 15, attained at dim V = 8 The paper is 30 Jul 2026 The Scope Problem 6.3 sits inside a group of related problems in

12: , and Navarro raises it there as an instance of a general principle: for a distinguished subgroup H ≤ G and a natural bijection ∗ from a canonical subset of Irr(G) onto one of Irr(H), one expects χ∗ to be linear exactly when χH is never zero Corollary B refutes that principle for the Glauberman–Isaacs correspondence The hypothesis that A be a p-group cannot be dropped The Verification The proofs above are self-contained; the computations described here are independent checks and are used nowhere in them All of them are exact, in integer and F2 arithmetic with no floating point, and all are available in

5: The paper is arXiv:2609.16227v1

math.GR: 30 Jul 2026 The author used OpenAI’s ChatGPT (GPT-5.6 Thinking) during the preparation of this manuscript for mathematical exploration, checking intermediate arguments and edge cases, generating exact verification code, and editorial revision The author reviewed the resulting arguments and takes responsibility for the contents of the paper The acknowledgments thank Abhinav Namboori for helpful discussions and comments on the exposition The References The references include

1: A.

Ada: H.

Title and authors: Ada: Let's start with the title and who wrote this paper, "Character Nonvanishing under Coprime Actions." It sounds very specific, focusing on characters not vanishing on the fixed subgroup C when A acts coprimely on G.

Emmy: The authors are Eric Hou, and they're working with a situation where A acts faithfully and coprimely on a two-group G of order two thousand one hundred thirty-five <ref:2609.16227#pg1>.

Kolya: That specific setup is interesting because it sets up the counterexample they need to show the conjecture fails.

Ada: The authors are setting up a situation where they find that this identification between nonvanishing characters and linear characters of C doesn't always hold, which refutes what Isaacs suggested.

Emmy: It's a crucial distinction because it suggests that the correspondence isn't perfect in all cases, even for solvable A groups.

The paper's summary: Ada: So, let's move to the summary of "Character Nonvanishing under Coprime Actions." The paper describes a group A of order twenty-one acting on G, where G is a specific type of group called B direct product V <ref:2609.16227#pg1,of order 21 acting on G>.

Emmy: This setup involves C being an elementary abelian group whose rank is determined by the number of orbits A has on V, and they use the little-groups method to link characters in Irr A(G) to characters in C.

Kolya: The core mechanism they use is showing that if a character vanishes somewhere on C, it forces some specific sub-multiset of the sizes of those orbits on V to sum up to exactly half of V.

Ada: That arithmetic constraint about summing to half the size of V is what they use to find the counterexample. They prove that this condition can only be met if V has an A-stable subset whose size is exactly half of V, and they show this happens for infinitely many pairs (A, V) but never for dimension less than seven.

Emmy: So, the summary boils down to finding a specific configuration where some character vanishes because of this geometric constraint on the orbits.

The paper's improvements: Ada: The authors suggest that their proof isolates a very short counterexample in Section four, which they claim can be checked by hand. This makes the construction quite concrete for verification.

Emmy: They break down the proof into three facts that work together: first, C is elementary abelian of rank equal to the number of A orbits on V; second, the little-groups method connects Irr A(G) to B direct product V; and third, they use a specific character associated with the indicator function of zero in V.

Kolya: The improvement here is really showing that this entire problem reduces to an arithmetic question about those orbit sizes summing up to half of V. That's a very clean way to frame the failure.

Ada: And for someone just listening, it means that if you can find an action where you can control the orbits on V, you can force a character to vanish by choosing A and V correctly.

Conclusion: Emmy: To wrap up, the paper concludes that for their specific group action—the C21 acting on G of order two thousand one hundred thirty-five—there is a character in Irr A(G) with degree one hundred twenty-eight that vanishes at some element of C.

Ada: That means the number of characters nowhere zero on C is one thousand seven hundred twenty-eight which is smaller than the size C/C Prime which they calculated to be four thousand ninety-six proving the equality fails for this case <ref:2609.16227#pg1>.

Kolya: From a probabilistic view, it suggests that as you increase the dimension of V or use larger groups A, this failure becomes even more pronounced because of how fast those ratios approach zero when you look at direct powers.

Ada: So, to summarize "Character Nonvanishing under Coprime Actions," they show that the expected correspondence between nonvanishing characters and linear characters is not always true.

Emmy: It confirms that for a general group A acting coprimely on G, we can't just rely on the Glauberman Isaacs correspondent to know if a character vanishes somewhere.

Eric Hou

math.GR

Submitted: 2026-07-30

Updated: 2026-11-09

Comments: 10 pages

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 83/100

The gist: Let A finite group act coprimely on a finite group G and put C = CG(A).

Key concepts

Coprime Action
This describes a situation where two groups, A and G, interact in a special way: they act on each other without sharing any common prime factors. This setup is used to study how characters of the larger group G behave when restricted to subgroups related to A.
Character Vanishing Nowhere
This refers to irreducible characters ($\chi$) of a group G that do not have a zero value at any element in the subgroup C = CG(A). The classical Burnside theorem suggests these characters should only be linear, but this paper tests if this holds under coprime actions.
Glauberman–Isaacs Correspondent
This is a specific mapping between certain irreducible characters of G and those of a subgroup A. The paper explores the hypothesis that $\chi$ does not vanish on C if and only if its correspondent character $\chi^*$ is linear, which is the core question being tested.

Terminology

Summary

Let A finite group act coprimely on a finite group G and put C = CG(A). p. 0

Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always C/C′? p. 0
Where the paper says it

A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21. 100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always |C/C′ |?

This would follow if one could show that χC is never zero if and only if the Glauberman–Isaacs correspondent χ∗ of χ is linear

The Counterexample Construction

Theorem A states that there exist a finite group A ∼= C21 and a finite 2-group G of order 2135, with A acting faithfully and coprimely on G, such that NA(G) = 1728 and CG(A)/CG(A)′ = 4096 In particular the equality of Problem 21.100 fails The group is G = B ⋊ V, where V = F24 ⊕ F8 and B is the group of Boolean functions on V, and B is the additive group of all Boolean functions on V

The Mechanism of Failure

The strategy for proving Theorem A involves isolating a counterexample where three facts do all the work First, C = CG(A) = BA is elementary abelian of rank the number of A-orbits on V (Proposition 2.2) Second, the little-groups method identifies IrrA(G) with BA (Proposition 2. p. 2

Where the paper says it

2); in particular C ′ = 1 and |C/C′ | = |BA|. Second, the little-groups method identifies IrrA(G) with BA (Proposition 2.

p. 2
Where the paper says it

2); in particular C ′ = 1 and |C/C′ | = |BA|. Second, the little-groups method identifies IrrA(G) with BA (Proposition 2.

5), so that IrrA(G) = C/C′ exactly; hence a single invariant character with a zero on C already refutes Problem 21. p. 2
Where the paper says it

Second, the little-groups method identifies IrrA(G) with BA (Proposition 2. 5), so that |IrrA(G)| = |C/C′ | exactly; hence a single invariant character with a zero on C already refutes Problem 21.

p. 2
Where the paper says it

Second, the little-groups method identifies IrrA(G) with BA (Proposition 2. 5), so that |IrrA(G)| = |C/C′ | exactly; hence a single invariant character with a zero on C already refutes Problem 21.

100 Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V, the associated χu ∈ IrrA(G) has degree V and satisfies χu(c) = V − 2supp c for c ∈ BA (Lemma 2. p. 2
Where the paper says it

5), so that |IrrA(G)| = |C/C′ | exactly; hence a single invariant character with a zero on C already refutes Problem 21. Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V , the associated χu ∈ IrrA(G) has degree |V | and satisfies χu(c) = |V | − 2|supp c| for c ∈ BA (Lemma 2.

p. 2
Where the paper says it

5), so that |IrrA(G)| = |C/C′ | exactly; hence a single invariant character with a zero on C already refutes Problem 21. Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V , the associated χu ∈ IrrA(G) has degree |V | and satisfies χu(c) = |V | − 2|supp c| for c ∈ BA (Lemma 2.

10) Everything therefore reduces to an arithmetic question about the multiset of A-orbit sizes on V: does some sub-multiset sum to V/2

The Arithmetic Constraint

The condition for a character vanishing at some element of C is determined by Corollary 2.9, which states that if some χ ∈ IrrA(G) vanishes at some element of C, then some sub-multiset of the A-orbit sizes on V sums to V/2 The smallest admissible cyclic pair is A = 21 with dim V = 7, where the orbit sizes are forced to be 1, 3, 3, 3, 3, 3, 7, and the sub-multisets summing to V/2 are exactly the twenty of the forms

The Final Count

Theorem A concludes that for this action there is a χ ∈ IrrA(G) of degree 128 and a c ∈ C with χ(c) = 0 This implies NA(G) ≤ 4095 < 4096 = C/C′, and the equality of Problem 21.100 fails The exact count is given by Theorem 7. p. 1

Where the paper says it

The criterion in the second sentence of Problem 21. 100 therefore predicts that no invariant character vanishes anywhere on C, and it fails as badly as it can.

1, showing that NA(G) = 1728 and CG(A)/CG(A)′ = 4096

Consequences for Related Problems

Corollary B shows that the proposed criterion of Problem 6.3 fails because there are 2368 characters χ ∈ IrrA(G) with χ∗(1) = 1 for which χC nevertheless has a zero Corollary D demonstrates that this failure persists when passing to direct powers, as the ratio NA(Gn)/Cn/(Cn)' approaches 0 as n → ∞ The smallest dimension for a balanced pair (A, V) is 7

The Exact Verification

Verification confirms that three programs compute NA(G) by different routes—a 128-point integer Walsh transform, a direct computation in F24 ⊕ F8 from the radial correlation table, and a binary linear-algebra test built on the twenty zero patterns of Lemma 6.1—and all three return 1728 The final script confirms that K = C × A is a Carter subgroup of Γ, as Proposition 9. p. 5

Where the paper says it

Thus ℓ = 0 and µ is trivial, giving θu,µ = λeu. The final count is Proposition 2.

p. 5
Where the paper says it

Thus ℓ = 0 and µ is trivial, giving θu,µ = λeu. The final count is Proposition 2.

1 asserts

The Minimal Case

Remark 3.7 notes that the smallest balanced pair (A, V) has dim V = 8 with A = 15 and element orders This confirms that the construction of Section 4 produces a counterexample to Problem 21. p. 9

Where the paper says it

Then (A, V ) is not balanced. Consequently dim V = 7 is the least dimension in which the construction of Section 2 produces a counterexample, over all operator groups of odd order.

100 if and only if V has an A-stable subset of size V/2 The smallest order of a balanced operator group is A = 15, attained at dim V = 8

The General Principle

The paper establishes that nonvanishing on C is not determined by χ∗ alone, and that any repair of Problem 6.3 must either restrict the operator group — to a p-group, say — or use more of χ than its correspondent The p-group case is the one that survives, and it is the natural remaining question

The Exact Zero Criterion

Lemma 6.1 states that expression (10) vanishes if and only if (H0, H1) = (δ + εJ, ρR + εJ) for some ε ∈ F2 and some three-element subset R ⊆ This corresponds bijectively to the twenty A-stable subsets of V of size 64 found in Proposition 3. p. 2

Where the paper says it

Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V , the associated χu ∈ IrrA(G) has degree |V | and satisfies χu(c) = |V | − 2|supp c| for c ∈ BA (Lemma 2. 10), so χu vanishes as soon as V has an A-stable subset of size |V |/2.

3 The final count is NA(G) = 1728 and CG(A)/CG(A)′ = 4096

The Carter Subgroup Analogue

Theorem 9.4 shows that the number of characters of Γ that do not vanish at any element of K is 21 NA(G) = 36288, whereas K/K′ = 86016 This proves Problem 6. p. 1

Where the paper says it

Then Γ is solvable of order 2 135 · 21, the subgroup K = CG(A) × A is a Carter subgroup of Γ, and exactly 36288 irreducible characters of Γ do not vanish on K, while |K/K′ | = 86016. Consequently Problem 6.

7 of [12] has a negative answer The head characters of Γ are not the Carter-nonvanishing irreducible characters of Γ

The General Bound

Theorem 9.7 shows that there are at most A NA(G) such characters, and since NA(G) 36288 This proves Problem 6. p. 1

Where the paper says it

Then Γ is solvable of order 2 135 · 21, the subgroup K = CG(A) × A is a Carter subgroup of Γ, and exactly 36288 irreducible characters of Γ do not vanish on K, while |K/K′ | = 86016. Consequently Problem 6.

7 has a negative answer for Γ, and the head characters of Γ are not its Carternonvanishing characters The smallest we obtain is the one of Theorem 9. p. 1
Where the paper says it

Consequently Problem 6. 7 of [12] has a negative answer, and the head characters of Γ in the sense of [8] are not its Carter-nonvanishing characters.

4, of order 2135 · 21

The Inequality

The inequality NA(G) ≤ C/C′ always holds in our examples, and holds trivially whenever C is abelian, since then C/C′ = Irr(C) = IrrA(G) ≥ NA(G) We know of no general argument

The Repair

The paper suggests that a repair of Problem 6.3 must either restrict the operator group — to a p-group, say — or use more of χ than its correspondent The p-group case is the one that survives, and it is the natural remaining question

The Verification

Verification confirms that three programs compute NA(G) by different routes—a 128-point integer Walsh transform, a direct computation in F24 ⊕ F8 from the radial correlation table, and a binary linear-algebra test built on the twenty zero patterns of Lemma 6.1—and all three return 1728 The final script confirms that K = C × A is a Carter subgroup of Γ, as Proposition 9.1 asserts The smallest we obtain is the one of Theorem 3. p. 2

Where the paper says it

2); no group of odd order at all, of any structure, admits one of dimension less than 7 (Theorem 3. 5); and the smallest admissible cyclic pair is |A| = 21 with dim V = 7, where the orbit sizes are forced to be 1, 3, 3, 3, 3, 3, 7, 21, 21, 21, 21, 21 and 1 + 21 + 21 + 21 = 64 = 1 2 |V | (Proposition 3.

5 and Corollary 3. p. 1
Where the paper says it

Is it true that χ ∗ (1) = 1 if and only if χC is never zero? Corollary B.

p. 1
Where the paper says it

Is it true that χ ∗ (1) = 1 if and only if χC is never zero? Corollary B.

6, which shows that dim V = 7 is the least attainable dimension in which the construction of Section 2 produces a counterexample, over all operator groups of odd order The smallest order of a balanced operator group is A = 15, attained at dim V = 8 The paper is 30 Jul 2026

The Scope

Problem 6.3 sits inside a group of related problems in [12], and Navarro raises it there as an instance of a general principle: for a distinguished subgroup H ≤ G and a natural bijection ∗ from a canonical subset of Irr(G) onto one of Irr(H), one expects χ∗ to be linear exactly when χH is never zero Corollary B refutes that principle for the Glauberman–Isaacs correspondence The hypothesis that A be a p-group cannot be dropped

The proofs above are self-contained; the computations described here are independent checks and are used nowhere in them All of them are exact, in integer and F2 arithmetic with no floating point, and all are available in [5] The paper is arXiv:2609.16227v1 [math. p. 0

Where the paper says it

1 arXiv:2609. 16227v1 [math.

p. 0
Where the paper says it

1 arXiv:2609. 16227v1 [math.

GR] 30 Jul 2026 The author used OpenAI’s ChatGPT (GPT-5. p. 0
Where the paper says it

16227v1 [math. GR] 30 Jul 2026

6 Thinking) during the preparation of this manuscript for mathematical exploration, checking intermediate arguments and edge cases, generating exact verification code, and editorial revision The author reviewed the resulting arguments and takes responsibility for the contents of the paper The acknowledgments thank Abhinav Namboori for helpful discussions and comments on the exposition

The References

The references include [1] A. H. Clifford, Representations induced in an invariant subgroup, Ann. p. 9

Where the paper says it

By the Feit–Thompson theorem [2] A is solvable, so it has a nontrivial abelian normal subgroup; let Z be one maximal among these. By Clifford’s theorem [1] V is a direct sum of k Z-isotypic components, which A permutes transitively; they have equal dimension d/k, and k divides |A|, so k is odd.

of Math. (2) 38 (1937), 533–550 [2] W. Feit and J. p. 9
Where the paper says it

Then there is an odd integer dA ≥ 3 that divides the size of every A-orbit on V \ {0}, and: (i) either there are a normal subgroup Z ≤ A and a divisor e > 1 of d with F2[Z] ∼= F2 e , such that V is an F2 e -vector space of dimension d/e on which Z acts by scalar multiplication; in this case dA = |Z| satisfies orddA (2) = e, so that e 2 3 4 5 6 7 dA 3 7 5, 15 31 9, 21, 63 127 (ii) or d = 6 and A is a 3-group; in this case dA = 3. By the Feit–Thompson theorem [2] A is solvable, so it has a nontrivial abelian normal subgroup; let Z be one maximal among these.

G. Thompson, Solvability of groups of odd order, Pacific J. p. 0
Where the paper says it

The construction needs only that V have an A-stable subset of half its size. This holds for infinitely many pairs (A, V ), and for none of dimension below 7, so the example is minimal over all operator groups of odd order.

Math. 13 (1963), 775–1029 [3] The GAP Group, GAP – Groups, Algorithms, and Programming, Version 4. p. 2
Where the paper says it

Notation Groups are finite and characters are complex. We write Cn for the cyclic group of order n, Irr(G) for the irreducible characters, G′ for the derived subgroup, and IndG H for induction.

14.0, 2024 [4] G. Glauberman, Correspondences of characters for relatively prime operator groups, Canad. p. 7
Where the paper says it

Then (A, V ) is not balanced. In particular no construction of Section 2 whose operator group has prime power order gives a counterexample.

J. Math. 20 (1968), 1465–1488 [5] E. Hou, Coprime actions and characters nonvanishing on the fixed-point subgroup: manuscript and verification code, Zenodo, 2026 [6] I. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

M. Isaacs, Characters of solvable and symplectic groups, Amer. p. 1
Where the paper says it

Glauberman [4] for solvable A and extended to arbitrary A by I. Isaacs [6]; see also [11, Chapter 2].

J. Math. 95 (1973), 594–635 [7] I. M. Isaacs, Character Theory of Finite Groups, Academic Press, New York, 1976 [8] I. p. 6
Where the paper says it

10 the character χδ0 ∈ IrrA(G) satisfies χδ0 (c) = 0, so NA(G) is a proper subset of IrrA(G). Since C is abelian every Glauberman–Isaacs correspondent is linear, so χδ0 has linear correspondent and a zero on C, so the criterion of Problem 6.

M. Isaacs, Carter subgroups, characters and composition series, Trans. p. 1
Where the paper says it

Let Γ = G ⋊ A for the action of Theorem A. Then Γ is solvable of order 2 135 · 21, the subgroup K = CG(A) × A is a Carter subgroup of Γ, and exactly 36288 irreducible characters of Γ do not vanish on K, while |K/K′ | = 86016.

Amer. Math. Soc. Ser. B 9 (2022), 470–498 [9] I. M. Isaacs, G. Navarro and T. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

R. Wolf, Finite group elements where no irreducible character vanishes, J. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

Algebra 222 (1999), no. 2, 413–423 [10] E. I. Khukhro and V. D. Mazurov (eds.), Unsolved Problems in Group Theory: The Kourovka Notebook, No. p. 0
Where the paper says it

A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21. 100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always |C/C′ |?

21, Sobolev Institute of Mathematics, Novosibirsk, 2026, Problem 21. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

100 [11] G. p. 0
Where the paper says it

A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21. 100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always |C/C′ |?

Navarro, Character Theory and the McKay Conjecture, Cambridge Studies in Advanced Mathematics 175 [12] G. p. 0
Where the paper says it

3 of Navarro’s problem list, and with it the corresponding statement about the head characters of Isaacs; passing to G⋊A refutes Problem 6. 7, a conjecture Isaacs reports is supported by abundant computational evidence.

Navarro, Problems on characters: solvable groups, Publ. p. 2
Where the paper says it

Section 8 proves Corollaries B and D and records what we do not know; Section 9 proves Corollary C; Section 10 delimits the scope of the refutation among the problems of [12]; and Section 11 describes machine checks, which are independent of the proofs. Notation Groups are finite and characters are complex.

Mat. 67 (2023), 173–198, Problem 6. p. 0
Where the paper says it

Coprime Actions and Characters Non-vanishing on the Fixed-point Subgroup Eric Hou July 30, 2026 Abstract Let a finite group A act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.

3

The Summary

Let A finite group act coprimely on a finite group G and put C = CG(A). A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21.100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always C/C′ First, C = CG(A) = BA is elementary abelian of rank the number of A-orbits on V (Proposition 2. p. 0

Where the paper says it

A classical theorem of Burnside asserts that the irreducible characters of a group vanishing nowhere are exactly the linear ones, and Navarro asked in Problem 21. 100 of the 21st Kourovka Notebook whether the coprime analogue holds: is the number of A-invariant χ ∈ Irr(G) with χC nowhere zero always |C/C′ |?

2) Second, the little-groups method identifies IrrA(G) with BA (Proposition 2.5), so that IrrA(G) = C/C′ exactly; hence a single invariant character with a zero on C already refutes Problem 21.100 Third, if u = δ0 ∈ BA is the indicator of 0 ∈ V, the associated χu ∈ IrrA(G) has degree V and satisfies χu(c) = V − 2supp c for c ∈ BA (Lemma 2.10)

The condition for a character vanishing at some element of C is determined by Corollary 2.9, which states that if some χ ∈ IrrA(G) vanishes at some element of C, then some sub-multiset of the A-orbit sizes on V sums to V/2 The smallest admissible cyclic pair is A = 21 with dim V = 7, where the orbit sizes are forced to be 1, 3, 3, 3, 3, 3, and the sub-multisets summing to V/2 are exactly the twenty of the forms

Theorem A concludes that for this action there is a χ ∈ IrrA(G) of degree 128 and a c ∈ C with χ(c) = 0 This implies NA(G) ≤ 4095 < 4096 = C/C′, and the equality of Problem 21.100 fails Corollary D demonstrates that this failure persists when passing to direct powers, as the ratio NA(Gn)/Cn/(Cn)' approaches 0 as n → ∞ <ref:30 Jul 26 The smallest dimension for a balanced pair (A, V) is 7 <ref:30 Jul 26 The smallest order of a balanced operator group is A = 15, attained at dim V = 8 <ref:30 Jul 26

Verification confirms that three programs compute NA(G) by different routes—a 128-point integer Walsh transform, a direct computation in F24 ⊕ F8 from the radial correlation table, and a binary linear-algebra test built on the twenty zero patterns of Lemma 6.1—and all three return 1728 <ref:30 Jul 26 The final script confirms that K = C × A is a Carter subgroup of Γ, as Proposition 9.1 asserts <ref:30 Jul 26 The smallest we obtain is the one of Theorem 3. p. 2

Where the paper says it

2); no group of odd order at all, of any structure, admits one of dimension less than 7 (Theorem 3. 5); and the smallest admissible cyclic pair is |A| = 21 with dim V = 7, where the orbit sizes are forced to be 1, 3, 3, 3, 3, 3, 7, 21, 21, 21, 21, 21 and 1 + 21 + 21 + 21 = 64 = 1 2 |V | (Proposition 3.

5 and Corollary 3.6, which shows that dim V = 7 is the least attainable dimension in which the construction of Section 2 produces a counterexample, over all operator groups of odd order <ref:30 Jul 26 The smallest order of a balanced operator group is A = 15, attained at dim V = 8 <ref:30 Jul 26 The paper is arXiv:2609. p. 9
Where the paper says it

Then (A, V ) is not balanced. Consequently dim V = 7 is the least dimension in which the construction of Section 2 produces a counterexample, over all operator groups of odd order.

16227v1 [math.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI system can definitively determine whether a character vanishing on a fixed-point subgroup C is determined solely by its Glauberman Isaacs correspondent, addressing Problem 21.100 directly using Theorem A as a counterexample to the proposed criterion.

  2. The system can calculate the exact ratio of nonvanishing irreducible characters to the size of the group structure for specific coprime actions, such as determining that for Theorem A, NA(G) = 1728 and CG(A)/CG(A)' = 4096, which is crucial for verifying results like Corollary D.

  3. The system can identify the minimal necessary structural constraints (like the dimension of V being at least 7) required for a coprime action to yield a counterexample, based on Theorem 3.5 and Proposition 3.2, narrowing the search space for such examples significantly.

  4. The system can verify if a given pair of groups (A, V) is balanced by checking if some sub-multiset of the A-orbit sizes on V sums to V/2, as described in Corollary 2.9 and Lemma 2.10, which is the core arithmetic condition for constructing counterexamples.

  5. The system can compute the exact count of nonvanishing characters over a Carter subgroup K for groups G split by A, yielding 21 NA(G) = 36288 versus K/K' = 86016, thereby proving that head characters are not always Carter-nonvanishing.

Abstract

Let a finite group A act coprimely on a finite group G, and let C=C G(A). Navarro asked whether an A-invariant irreducible character has linear Glauberman--Isaacs correspondent if and only if it is nonvanishing on C. We give counterexamples to this equivalence. We also prove that when G is nilpotent, nonvanishing on C implies that the Glauberman--Isaacs correspondent is linear.

Sources

Related papers