1. Introduction
A classical theorem of Broline and Garrison implies that if an irreducible character χ of a finite group G has maximal degree then $\operatorname{Ker}\chi$ is nilpotent (Corollary 12.20 of [Reference Isaacs1]). This result was extended by Isaacs, who considered characters of nth maximal degree in [Reference Isaacs2], and proved that if
$\chi\in{{\operatorname{Irr}}}(G)$ has nth maximal degree, then the Fitting height of the solvable radical of
$\operatorname{Ker}\chi$ is at most n.
Our goal in this note is to consider irreducible characters at the other extreme. Of course, if $\chi\in{{\operatorname{Irr}}}(G)$ is linear, then
$G'\leq\operatorname{Ker}\chi$ and
$G/\operatorname{Ker}\chi$ is abelian. But can we restrict the structure of
$G/\operatorname{Ker}\chi$ if
$\chi(1)$ is “small”? This is the content of our main result. We write
$m(G)=\min\{\chi(1)\mid\chi\in{{\operatorname{Irr}}}(G), \chi(1) \gt 1\}$. We say that
$\chi\in{{\operatorname{Irr}}}(G)$ is a minimal character if
$\chi(1)=m(G)$.
Theorem A. Let G be a solvable finite group. Suppose that m(G) is odd. If $\chi\in{{\operatorname{Irr}}}(G)$ is a minimal character, then
$G/\operatorname{Ker}\chi$ is nilpotent-by-abelian.
As $\operatorname{GL}_2(3)$ shows, some hypothesis on m(G) is definitely necessary. The Frobenius group of order 20 acting faithfully on an extraspecial 2-group of order 25 is an example with faithful minimal characters of degree 4. We do not know whether it is enough to assume that m(G) is not a power of 2. On the other hand, some solvability hypothesis is definitely necessary: consider any non-abelian simple group with odd degree minimal characters (for instance,
$\mathsf{A}_5$). Theorem A follows from applying the next result to
$G/\operatorname{Ker}\chi$.
Theorem B. Let G be a finite solvable group. Suppose that $\chi\in{{\operatorname{Irr}}}(G)$ is a faithful minimal character. If
$\chi(1)$ is odd, then G is nilpotent-by-abelian.
Note that the structure of groups with all minimal characters faithful was described in detail by Robinson in [Reference Robinson6, Reference Robinson7]. In particular, as shown in Lemma 2.1 of [Reference Robinson6], solvable groups with all minimal characters faithful are nilpotent-by-abelian. The examples mentioned above show that this is not the case if we just assume that G has a minimal faithful character. Our proof of Theorem B relies on some of the ideas developed by Robinson.
2. Proofs
We argue as in Lemma 2 of [Reference Robinson7] to prove our first lemma.
Lemma 2.1. Let G be a finite group. Suppose that $\chi\in{{\operatorname{Irr}}}(G)$ is a primitive faithful minimal character of G. If
$N\trianglelefteq G$ is non-central, then
$\chi_N\in{{\operatorname{Irr}}}(N)$.
Proof. Suppose that $\chi_N\not\in{{\operatorname{Irr}}}(N)$. Then there exists a central extension
$G^*$ of G and
$\alpha,\beta\in{{\operatorname{Irr}}}(G^*)$ such that
$\chi=\alpha\beta$, where
$\alpha,\beta$ are primitive nonlinear irreducible characters of
$G^*$. Without loss of generality, we may assume that
$\alpha(1)\leq\chi(1)^{1/2}$. Since
$1_{G^*}$ is an irreducible constituent of
$\alpha\overline{\alpha}$, the minimality of
$\chi(1)$ implies that
$\alpha\overline{\alpha}$ is a sum of linear characters. Hence
$(G^*)'\leq\operatorname{Ker}(\alpha\overline{\alpha})$. In particular,
$(G^*)'\leq{\mathbf{Z}}(\alpha)$. By Lemma 2.27 of [Reference Isaacs1],
${\mathbf{Z}}(\alpha)/\operatorname{Ker}\alpha\leq{\mathbf{Z}}(G^*/\operatorname{Ker}\alpha)$. If follows that
$G^*/\operatorname{Ker}\alpha$ is nilpotent (of class at most 2). But α is nonlinear and primitive. This contradicts Theorem 6.22 of [Reference Isaacs1].
Next, we handle the primitive case of Theorem B. We refer the reader to [Reference Isaacs3] for the definition and basic properties of Gajendragadkar’s p-special characters.
Lemma 2.2. Let G be a solvable group. Suppose that $\chi\in{{\operatorname{Irr}}}(G)$ is a primitive faithful minimal character. Then
$\chi(1)$ is a power of a prime p. Furthermore, if p > 2, then G is nilpotent-by-abelian.
Proof. We may assume that $\chi(1) \gt 1$. By Theorem 2.17 of [Reference Isaacs3], χ factors as a product of p-special characters, where p runs over the set of prime divisors of
$\chi(1)$. Since
$\chi(1)=m(G)$, it follows that
$\chi(1)=p^n$ is a power of a prime p. This proves the first part of the lemma.
Suppose now that p > 2. Let q ≠ p be a prime. Then $\chi_{\mathbf{O}_q(G)}$ is not irreducible. It follows from Lemma 2.1 that
$\mathbf{O}_q(G)$ is central in G. Hence
${\mathbf{F}}(G)=E{\mathbf{Z}}(G)$, where
$E=\mathbf{O}_p(G)$. Furthermore, using again Lemma 2.1, every normal abelian subgroup of G is central. Since G has a faithful irreducible character, Theorem 2.32 of [Reference Isaacs1] implies that
${\mathbf{Z}}(G)$ is cyclic. Now, Corollary 1.10 of [Reference Manz and Wolf5] implies that E is extraspecial of exponent p. Since
$\chi_E\in{{\operatorname{Irr}}}(E)$, we necessarily have that
$|E|=p^{2n+1}$.
Note that ${\mathbf{C}}_G(E)={\mathbf{C}}_G({\mathbf{F}}(G))={\mathbf{Z}}(G)$, so
$G/{\mathbf{Z}}(G)$ is isomorphic to a subgroup of
${{\operatorname{Aut}}}_{{\mathbf{Z}}(E)}(E)$. By [Reference Winter8], using again that p > 2, we deduce that
$G/{\mathbf{F}}(G)$ is isomorphic to a subgroup of
$\operatorname{Sp}(2n,p)$. By [Reference Landazuri and Seitz4],
$\operatorname{Sp}(2n,p)$ has a faithful irreducible representation of dimension
$(p^n-1)/2$. Hence
$G/{\mathbf{F}}(G)$ has a faithful character of degree
$\leq (p^n-1)/2$. Since
$m(G)=p^n$, we conclude that
$G/{\mathbf{F}}(G)$ has a faithful character that is a sum of linear characters. We conclude that
$G/{\mathbf{F}}(G)$ is abelian, as we wanted to prove.
Now, we complete the proof of a slightly strengthened version of Theorem B.
Theorem 2.3. Let G be a solvable group. Suppose that $\chi\in{{\operatorname{Irr}}}(G)$ is a faithful minimal character. If χ is induced from an odd degree character, then G is nilpotent-by-abelian.
Proof. Let $H\leq G$ and
$\beta\in{{\operatorname{Irr}}}(H)$ primitive such that
$\beta^G=\chi$. Suppose first that
$\beta(1)=1$, so that
$\chi(1)=|G:H|$. Since 1G is an irreducible constituent of
$(1_H)^G$ and
$m(G)=|G:H|=(1_H)^G(1)$, we deduce that
$(1_H)^G$ is a sum of linear characters. Hence
$G'\leq\operatorname{Ker}(1_H)^G\leq H$. Thus
$H\trianglelefteq G$ and by Clifford’s theorem (Theorem 6.2 of [Reference Isaacs1]), χH is a sum of conjugates of β. In particular, χH is a sum of linear characters, so
$H'\leq\operatorname{Ker}\chi=1$. Hence G is metabelian and the result follows.
Now, we may assume that $\beta(1) \gt 1$ is odd. First, we will see that
$H\trianglelefteq G$ and
$G'=H'$. Note that
$\chi(1)=|G:H|\beta(1) \gt |G:H|$. Hence
$(1_H)^G$ is a sum of linear characters and
$G'\leq H$, as before. In particular,
$H\trianglelefteq G$. Thus H ʹ is also normal in G and all the irreducible characters of
$G/H'$ have degree divisible by
$|G:H| \lt \chi(1)=m(G)$. Hence,
${{\operatorname{Irr}}}(G/H')$ is a set of linear characters, and we conclude that
$G'=H'$, as desired.
Now, we claim that $\beta(1)=m(H)$. Let
$\mu\in{{\operatorname{Irr}}}(H)$ be nonlinear. Hence, there exists
$\nu\in{{\operatorname{Irr}}}(G)$ nonlinear such that
$[\mu^G,\nu]\neq 0$. Thus
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20241202182032052-0997:S0013091524000427:S0013091524000427_eqnU1.png?pub-status=live)
We conclude that $\beta(1)\leq\mu(1)$. The claim follows.
Thus β is a primitive faithful minimal character of $H/\operatorname{Ker}\beta$. By Lemma 2.2, we have that
$\beta(1)$ is a power of a prime p. Let
$\beta=\beta_1,\dots,\beta_t$ be the G-conjugates of β. Let
$K_i=\operatorname{Ker}\beta_i$. Since
$\chi=\beta^G$ is faithful, Lemma 5.11 of [Reference Isaacs1] implies that
$\bigcap_{i=1}^tK_i=1$. Since p > 2, by the second part of Lemma 2.2, we have that
$H/K_i$ is nilpotent-by-abelian. Write
$F_i/K_i={\mathbf{F}}(H/K_i)$, so that
$G'=H'\leq F_i$ for every i. By Proposition 9.5 of [Reference Manz and Wolf5],
$\bigcap_{i=1}^tF_i={\mathbf{F}}(H)$. Therefore,
$G'\leq {\mathbf{F}}(H)$ and we conclude that G is nilpotent-by-abelian, as wanted.
Funding Statement
Research supported by Ministerio de Ciencia e Innovación (Grant PID2019-103854GB-I00 funded by MCIN/AEI/10.13039/501100011033).