KEVOS
ArticlesServicesCase studiesAboutContact
ArticlesServicesCase studiesAboutContact
← ArticlesSub-exponential Class Group Computation for Quadratic FieldsEngineering · Engineering MathematicsLesson 810/884← PrevNext →
ArticlePublished 7 Aug 20262 min readBy Kevin Joginsub-exponentialquadratic fieldclass grouprelation collection
On this page

Ask about this page

KEVOS AISub-exponential Class Group Computation for Quadratic Fields

KEVOS knowledge first · trusted web sources when needed

Quadratic Fields

Sub-exponential Class Group Computation for Quadratic Fields

Sub-exponential class group and regulator computation for quadratic fields by relation collection over a factor base.

Engineering / MathematicsQuadratic Fields2 min readKV-MATH-0608

Enumeration and cycle traversal both cost roughly the square root of the discriminant. The sub-exponential method collects relations over a factor base instead, reducing the exponent from one half to the sub-exponential range.

The method

Factor base→Collect smooth relations→Linear algebra→Class group and regulator

Sub-exponential class group computation

  1. Build the factor basePrime ideals of norm below a bound, equivalently forms with small first coefficient.
  2. Generate random formsCompose random products of factor base elements and reduce.
  3. Test smoothnessA reduced form whose first coefficient factors over the base gives a relation.
  4. Assemble the matrixRows are relations, columns are factor base primes.
  5. ReduceStructured elimination then Smith normal form.
  6. Read the resultsInvariant factors give the class group; the kernel gives the regulator.
Running time ~ L_D(1/2, c)Conditional on GRH; D the discriminant.

Why reduction produces smooth forms

Key point

Reduction bounds the first coefficient by roughly the square root of the discriminant. A random integer of that size is smooth with respect to a well-chosen base with useful probability, which is exactly the mechanism that makes the method work — see smoothness.

The real quadratic complication

Caution

For real fields, relations carry a real-valued component recording the distance travelled along the cycle. That component is what produces the regulator, and it must be tracked numerically with sufficient precision throughout — a considerable practical complication absent from the imaginary case.

Comparison

Quadratic class group methods compared
MethodCostRange
Form enumerationSquare root of the discriminantUp to about 12 digits
Cycle traversalSquare root of the discriminantReal fields, modest size
Analytic formulaGrows with required precisionVerification, small cases
Sub-exponential relationsL(1/2), conditional on GRHLarge discriminants

Relation to factoring

Key point

The structure of this algorithm is identical to that of the quadratic sieve: collect smooth relations over a factor base, then do sparse linear algebra. The kinship is not coincidental — see the class group factoring method, which runs the connection in the other direction.

The general case

The same approach extends to arbitrary number fields as Buchmann's algorithm, with ideals in place of forms and a more elaborate reduction step.

Source. Henri Cohen, A Course in Computational Algebraic Number Theory, Springer GTM 138 — 5.3.4-5.4.3. Structural reference unverified: the source file was not available during authoring; chapter and section numbers are taken from the published edition and have not been checked against a physical copy.

Related pages

  • Relation Matrix Construction
  • The Sub-exponential Algorithm in Practice
  • Quadratic Sieve Factor Base Selection
  • The Fundamental Unit of a Real Quadratic Field
  • Computing the Structure of Residue Rings

Continue learning

The Fundamental Unit of a Real Quadratic FieldArticle · Engineering MathematicsNEXT LESSON →Computing the Structure of Residue RingsArticle · Engineering MathematicsReduction of Indefinite Forms and the Cycle StructureArticle · Engineering MathematicsThe Maximal Order ProblemArticle · Engineering Mathematics
KEVOS · Engineering, manufacturing and project improvement
ArticlesServicesCase studiesAboutContact
© 2026 KEVOS®