Mod 2 Cohomological Invariants and Witt Invariants of Alternating Groups

Basic Information

• Paper ID: 2401.12738
• Title: Invariants cohomologiques mod 2 et invariants de Witt des groupes alternés
• Author: Jean-Pierre Serre
• Classification: math.GR (Group Theory)
• Publication Date: January 2024 (arXiv v4 updated November 7, 2025)
• Paper Link: https://arxiv.org/abs/2401.12738

Abstract

This paper determines the mod 2 cohomological invariants and Witt invariants of alternating groups $A_n$. This extends Serre's results in Se 03 on the corresponding invariants of symmetric groups $S_n$.

Research Background and Motivation

Research Questions

This paper investigates two important classes of algebraic invariants of alternating groups $A_n$:

1. Mod 2 cohomological invariants: Characterizing the cohomological properties of $A_n$-torsors under field extensions
2. Witt invariants: Characterizing the Witt ring properties of quadratic forms related to $A_n$

Significance of the Problem

1. Theoretical significance: Alternating groups are fundamental finite groups, and their invariant theory is a central problem in algebraic topology and algebraic geometry
2. Relationship to symmetric groups: Alternating groups are index 2 subgroups of symmetric groups, but their invariant structures differ essentially
3. Application of Milnor's conjecture: Using Voevodsky's proof of Milnor's conjecture to establish connections between cohomology and Witt theory

Limitations of Existing Methods

• The invariants of symmetric groups $S_n$ have been completely determined in Se 03
• As subgroups of $S_n$, alternating groups require new detection techniques
• Standard restriction map methods cannot be directly applied, since $[S_n:A_n]=2$ is even

Research Motivation

To extend the invariant theory of Coxeter groups (particularly symmetric groups) to alternating groups, and to develop a "splitting principle" applicable to more general groups.

Core Contributions

The main contributions of this paper include:

1. Complete determination of the structure of cohomological invariants of $A_n$ (Theorem 3.4.1):
   • Proves that $\text{Inv}_k(A_n)$ is a free $H(k)$-module of rank $1+[\frac{n}{4}]$
   • Provides an explicit basis: $w_{2i}(q^a)$, where $0 \leq i \leq [\frac{n}{4}]$
2. Complete determination of the structure of Witt invariants of $A_n$ (Theorem 3.4.2):
   • Proves that $\text{Inv}_k(A_n,W)$ is a free $W(k)$-module of rank $1+[\frac{n}{4}]$
   • Provides an explicit basis: $\lambda_i q^a$, where $0 \leq i \leq [\frac{n}{4}]$
3. Establishes a detection theorem (Theorem 4.2.1):
   • Constructs a special 2-subgroup $E \subset A_n$ (elementary abelian group of rank $2[\frac{n}{4}]$)
   • Proves that the restriction map $\text{Inv}_k(A_n,\mathcal{C}) \to \text{Inv}_k(E,\mathcal{C})$ is injective
4. Provides refined properties of trace forms (Theorems 6.1.1 and 6.1.2):
   • Proves that exterior powers $\lambda_j q_L$ can be expressed as integer linear combinations of $\lambda_i q_L$ ($i \leq [\frac{n}{4}]$)
   • Proves that the generating function $\lambda_t(q_L)$ is divisible by $(1+t)^{2m+c}$

Detailed Methods

Task Definition

Let $k$ be a field of characteristic $\neq 2$, and $G$ a finite group.

Cohomological invariants: For each field extension $K/k$, provide a map $a_K: \text{Tors}_G(K) \to H^r(K)$ satisfying functoriality compatible with field extensions.

Witt invariants: Similarly define maps to the Witt ring $W(K)$.

Core objects:

• $q^a \in \text{Inv}_k(A_n,\hat{W})$: trace form invariant
• $w_i(q^a) \in \text{Inv}_k^i(A_n)$: Stiefel-Whitney classes
• $\lambda_i q^a \in \text{Inv}_k(A_n,W)$: exterior powers

Method Architecture

Step One: Construction of Detection Subgroups

For the decomposition $n = 4m + c$ ($c \in \{0,1,2,3\}$), construct:

1. Partition: $X = \{1,\ldots,n\} = X_0 \cup X_1 \cup \cdots \cup X_m$
   • $X_i = \{4i-3, 4i-2, 4i-1, 4i\}$ for $i=1,\ldots,m$
   • $X_0 = \{4m+1,\ldots,n\}$
2. Local groups: $A(i) \cong A_4$ fixing $X \setminus X_i$
3. Detection group:
   • $E = \prod_{i=1}^m E(i)$, where $E(i)$ is a 2-Sylow subgroup of $A(i)$
   • $E$ is an elementary abelian 2-group of rank $2m$

Step Two: Proof of Detection Theorem

Key Lemma (Proposition 4.3.1):

• Construct $D = C_{A_n}(s)$ (centralizer of $s$)
• Prove that $D$ is a Coxeter group of type $D_r$ ($r = n/2$)
• $E$ is a maximal cube of $D$
• $(A_n : D)$ is odd

Application of splitting principle:

1. From $(A_n:D)$ odd $\Rightarrow$ $\text{Res}: \text{Inv}_k(A_n) \to \text{Inv}_k(D)$ is injective
2. From Weyl group theory $\Rightarrow$ $\text{Res}: \text{Inv}_k(D) \to \text{Inv}_k(E)$ is injective
3. Composition gives $\text{Res}: \text{Inv}_k(A_n) \to \text{Inv}_k(E)$ is injective

Step Three: Computation of Invariants

For the case $n = 4m$:

1. Decomposition: $A = \prod_{i=1}^m A(i) \cong A_4^m$
2. Tensor product structure (Proposition 1.8.3): $\text{Inv}_k(A,\mathcal{C}) = \bigotimes_{i=1}^m \text{Inv}_k(A(i),\mathcal{C})$
3. Symmetrization of invariants:
   • Define $q^a(d) = \sum_{|I|=d} \bigotimes_{i \in I} q_i^a$
   • Prove that $\{q^a(d): 0 \leq d \leq m\}$ is a basis for the invariant part of the normalizer
4. Computation of restriction maps (Proposition 5.3.2): $\lambda_t(\text{Res}(q^a)) = (1+t)^{2m} \sum_{d \leq m} t^d(1-t)^{2m-2d} q^a(d)$

For general $n$: Use induction noting that the detection groups for $A_{4m}, A_{4m+1}, A_{4m+2}, A_{4m+3}$ are the same.

Technical Innovations

1. Use of bitranspositions:
   • Unlike symmetric groups which use transpositions, alternating groups require bitranspositions
   • Bitranspositions have support sets of 4 elements, corresponding to biquadratic extensions
2. $D$-type structure of Coxeter groups:
   • Clever construction making the centralizer a group of type $D_r$
   • Utilization of cube theory for $D$-type groups
3. Generating function method:
   • Exploitation of divisibility by $(1+t)^{2m+c}$
   • Establishment of recurrence relations via Taylor expansion
4. Deep application of Milnor's conjecture:
   • Use of the reduction formula for symmetric functions (1.1.6)
   • Bridging cohomology and Witt theory (§2)

Experimental Setup

This is a pure mathematics theoretical paper with no numerical experiments. All results are obtained through rigorous mathematical proofs.

Verification Methods

• Explicit calculations for small values of $n$ (such as $n=4,6,7,10$)
• Verification of general cases via induction
• Comparison with known results for $S_n$

Main Results

Theorem 3.4.1 (Cohomological Invariants)

$\text{Inv}_k(A_n) \cong H(k) \oplus H(k) \cdot w_2(q^a) \oplus \cdots \oplus H(k) \cdot w_{2[\frac{n}{4}]}(q^a)$

Key properties:

• Rank $= 1 + [\frac{n}{4}]$ equals the number of conjugacy classes of involutions in $A_n$
• $w_i(q^a) = 0$ when $i$ is odd or $i > 2[\frac{n}{4}]$

Theorem 3.4.2 (Witt Invariants)

$\text{Inv}_k(A_n,W) \cong W(k) \oplus W(k) \cdot \lambda_1 q^a \oplus \cdots \oplus W(k) \cdot \lambda_{[\frac{n}{4}]} q^a$

Comparison with $S_n$:

• $S_n$: rank $= 1 + [\frac{n}{2}]$, basis detected by quadratic extensions
• $A_n$: rank $= 1 + [\frac{n}{4}]$, basis detected by biquadratic extensions

Corollary 3.4.3 (Surjectivity)

The restriction map $\text{Res}: \text{Inv}_k(S_n,\mathcal{C}) \to \text{Inv}_k(A_n,\mathcal{C})$ is surjective.

Theorem 6.1.2 (Divisibility of Generating Functions)

For an étale algebra $L$ of rank $n$ with discriminant 1, $\lambda_t(q_L) \equiv 0 \pmod{(1+t)^{2m+c}}$ where $m = [\frac{n}{4}]$ and $c = n - 4m$.

Concrete Example ($n=7$)