Infinite Class Field Towers

• Infinite class field towers are sequences of number fields obtained by iteratively taking maximal unramified abelian (or pro‑p) extensions, ensuring the process never stabilizes.
• The Golod–Shafarevich criterion links the generators and relations of associated pro‑p groups to guarantee the infinitude of these towers.
• Explicit constructions, using techniques like Kummer and cyclotomic extensions, provide key examples with minimal ramification and controlled root discriminant bounds.

An infinite class field tower is a sequence of number fields (or global function fields) where each stage is the maximal unramified abelian (or pro-$p$) extension of its predecessor, such that the process never stabilizes—i.e., there is no finite stage where the field is its own Hilbert class field. The theory is deeply intertwined with the structure of class groups, pro-$p$ Galois groups, and group cohomology. The existence, construction, and properties of fields with infinite class field towers are central in algebraic number theory, with implications for root discriminant bounds, explicit field constructions, algorithmic number theory, and heuristics on class groups.

1. Definitions and Fundamental Structures

Let $K$ be a number field with ideal class group $\operatorname{Cl}_K$. Define the (narrow) Hilbert class field $H_0(K)$ as the maximal abelian unramified extension of $K$. Set $K_0=K$ and recursively define $K_{i+1} = H_0(K_i)$. The chain $K_0 \subset K_1 \subset K_2 \subset \cdots$ is called the class field tower of $K$ or the Hilbert class field tower.

For a fixed rational prime $p$0, the $p$1-class field tower is the sequence where each $p$2 is the maximal abelian unramified $p$3-extension of $p$4. The Galois group $p$5 of the full unramified pro-$p$6 extension encodes the tower in the structure of its abelianizations: $p$7 (Liu et al., 2024, Boston et al., 2017).

A class field tower is infinite if for all $p$8, $p$9, i.e., there is no finite maximal unramified abelian extension.

2. Golod–Shafarevich Criterion and Generalizations

The classical tool for proving infinitude of class field towers is the Golod–Shafarevich inequality. For a pro-$K$0 group $K$1 presented minimally with $K$2 generators and $K$3 relations, set $K$4. If $K$5 is finite, then $K$6 for all $K$7. If $K$8, $K$9 is infinite (Hajir et al., 2019, McLeman, 2010, Joshi et al., 2010, Liu et al., 2024).

Schoof's extension applies this by focusing on cyclic extensions $\operatorname{Cl}_K$0 of prime degree $\operatorname{Cl}_K$1, giving a numerical criterion involving the number of ramified places $\operatorname{Cl}_K$2 and ranks $\operatorname{Cl}_K$3 of certain unit and norm-indexed groups. If

$\operatorname{Cl}_K$4

then $\operatorname{Cl}_K$5 has infinite $\operatorname{Cl}_K$6-class tower (Liu et al., 2024, Leshin, 2013).

Liu–Xing further generalize to nonabelian Galois extensions, replacing the invariants with group cohomological bounds on local inertia and decomposition subgroups, and obtaining more flexible criteria for infinite $\operatorname{Cl}_K$7-class towers in nonabelian settings (Liu et al., 2024).

3. Explicit Constructions and Small Root Discriminant Examples

A major challenge is constructing explicit number fields with infinite class field towers, especially with minimal ramification (few ramified primes) and small root discriminant. Kummer and cyclotomic techniques exploit carefully chosen extensions:

• Kummer extensions of cyclotomic fields: For example, $\operatorname{Cl}_K$8 has infinite $\operatorname{Cl}_K$9-class tower if $H_0(K)$0 exceeds a threshold derived from the tower criterion. Optimized examples for $H_0(K)$1 achieve root discriminants $H_0(K)$2 respectively, below previous records for these characteristics (Liu et al., 2024).
• $H_0(K)$3-extensions: Leshin constructs number fields $H_0(K)$4 (Galois over $H_0(K)$5, ramified at 3, $H_0(K)$6, and $H_0(K)$7) with infinite 3-class field tower and minimal known root discriminant ($H_0(K)$8 for $H_0(K)$9) (Leshin, 2013).
• Prime-power discriminant fields: Hajir–Maire–Ramakrishna show for any $K$0 the existence of $K$1 ramified only at $K$2 and $K$3 with infinite $K$4-tower, exploiting the high class numbers in cyclotomic fields and manipulating local and global relations in their Galois groups (Hajir et al., 2019).

Function field analogs (over $K$5) leverage Artin–Schreier and Carlitz–cyclotomic extensions, making construction of towers with prescribed ramification particularly tractable (Hoelscher, 2011).

4. Structural Invariants, Successive Approximations, and Classification

The structure of infinite class field towers is controlled by the interplay of class groups, Galois module structure, and the classification of pro-$K$6 groups arising as tower groups.

• Invariants computed: Successive Approximation Theorem (Mayer) shows that the abelian type invariants (ATI) of $K$7 for all intermediate unramified $K$8-extensions $K$9 up to a given stage almost determine the structure of the tower group (Mayer, 2017).
• Transfer kernels and Artin patterns: Artin transfer homomorphisms and their kernels, encapsulated in Artin patterns or transfer kernel types, allow for fine classification and exclusion of potential tower group candidates.
• Metabelian and extraspecial groups: Explicit group-theoretical parametrizations (e.g., coclass-1 $K_0=K$0-groups) and their maximal subgroups are crucial for understanding which group types can occur as Galois groups for towers over particular number fields, especially in dihedral and quadratic settings.

Cohomological refinements—such as those involving triple Massey products in the case of $K_0=K$1—can decide infinitude in borderline rank cases (McLeman, 2010).

5. Root Discriminant Bounds and Martinet's Conjecture

Root discriminant, $K_0=K$2, is preserved along an unramified tower. Odlyzko's analytic lower bounds imply a universal lower bound for infinite tower fields: $K_0=K$3 unconditionally, $K_0=K$4 under GRH (Boston et al., 2017, Bleher et al., 2023, Liu et al., 2024). Constructing infinite towers with $K_0=K$5 as small as possible is a major open challenge.

The existence and length of 2-class field towers for imaginary quadratic fields with 2-class group of rank 4 is the content of Martinet's conjecture. Finite 2-towers for such fields would yield a negative answer and preclude improved upper bounds for the lim inf of root discriminants, while an infinite tower would establish new records (Boston et al., 2017).

6. Infinite Towers from Galois Representations and Density Theorems

Fields with infinite class field towers arise as fixed fields of Galois representations:

• Modular forms and elliptic curves: For suitable $K_0=K$6 and $K_0=K$7, the fixed fields of the mod-$K_0=K$8 Galois representation associated to a weight-$K_0=K$9 modular eigenform or the $K_{i+1} = H_0(K_i)$0-division field of a non-CM elliptic curve typically admit infinite towers, contingent upon cycling the problem to cyclotomic fields with infinite towers (Joshi et al., 2010).
• Density results: For "most" $K_{i+1} = H_0(K_i)$1, the field $K_{i+1} = H_0(K_i)$2 has an infinite class field tower; similarly, fields constructed from Galois representations of modular forms provide infinite towers for a set of primes of density one among those coprime to a fixed compositum.

These constructions point to the ubiquity of infinite class field towers in large families of number fields.

7. Open Problems, Heuristics, and Further Developments

Key open questions include:

• Minimization of root discriminant for infinite-tower fields, especially in odd residue characteristic (no known examples with $K_{i+1} = H_0(K_i)$3 in $K_{i+1} = H_0(K_i)$4).
• Classification of fields with prescribed ramification sets admitting infinite towers (e.g., "tame" versus "wild" ramification, $K_{i+1} = H_0(K_i)$5-towers).
• Structural heuristics and measure constructions on pro-$K_{i+1} = H_0(K_i)$6 groups guide predictions about the distribution and typicality of infinite towers in number field families. Non-abelian Cohen–Lenstra heuristics (Boston–Bush–Hajir and extensions) conjecture probability measures on possible pro-$K_{i+1} = H_0(K_i)$7 tower groups, with positive measure contribution to infinite groups only for high class-group ranks (Boston et al., 2018).
• The search for refined group-theoretic or cohomological criteria (beyond Golod–Shafarevich) is ongoing, motivated by "near-miss" cases and the role of higher $K_{i+1} = H_0(K_i)$8-group ranks (e.g., $K_{i+1} = H_0(K_i)$9-rank, Rédei matrices, Massey products) (Wang, 2015, McLeman, 2010).

Extension to function field analogs demonstrates systematic, explicit constructions with prescribed ramification, highlighting contrasts and parallels with the number field case (Hoelscher, 2011). These advances continue to enrich the structural understanding and landscape of infinite class field towers in arithmetic geometry and algebraic number theory.