Number Fields: Galois Groups and Class Groups

Galois Groups

We can compute the Galois group of a number field using the galois_group function, which by default calls Pari (http://pari.math.u-bordeaux.fr/). You do not have to worry about installing Pari, since Pari is part of Sage. In fact, despite appearances much of the difficult algebraic number theory in Sage is actually done by the Pari C library (be sure to also cite Pari in papers that use Sage).

Sage

sage: K.<alpha> = NumberField(x^6 + 40*x^3 + 1372)
sage: G = K.galois_group()
sage: G
Galois group 6T2 ([3]2) with order 6 of x^6 + 40*x^3 + 1372

Python

>>> from sage.all import *
>>> K = NumberField(x**Integer(6) + Integer(40)*x**Integer(3) + Integer(1372), names=('alpha',)); (alpha,) = K._first_ngens(1)
>>> G = K.galois_group()
>>> G
Galois group 6T2 ([3]2) with order 6 of x^6 + 40*x^3 + 1372

Sage Live

K.<alpha> = NumberField(x^6 + 40*x^3 + 1372)
G = K.galois_group()
G

Internally G is represented as a group of permutations, but we can also apply any element of G to any element of the field:

Sage

sage: G.order()
6
sage: G.gens()
[(1,2)(3,4)(5,6), (1,4,6)(2,5,3)]
sage: f = G.1; f(alpha)
1/36*alpha^4 + 1/18*alpha

Python

>>> from sage.all import *
>>> G.order()
6
>>> G.gens()
[(1,2)(3,4)(5,6), (1,4,6)(2,5,3)]
>>> f = G.gen(1); f(alpha)
1/36*alpha^4 + 1/18*alpha

Sage Live

G.order()
G.gens()
f = G.1; f(alpha)

Some more advanced number-theoretical tools are available via G:

Sage

sage: P = K.primes_above(2)[0]
sage: G.inertia_group(P)
Subgroup generated by [(1,4,6)(2,5,3)] of (Galois group 6T2 ([3]2) with order 6 of x^6 + 40*x^3 + 1372)
sage: sorted([G.artin_symbol(Q) for Q in K.primes_above(5)])  # random order, see Issue #18308
[(1,3)(2,6)(4,5), (1,2)(3,4)(5,6), (1,5)(2,4)(3,6)]

Python

>>> from sage.all import *
>>> P = K.primes_above(Integer(2))[Integer(0)]
>>> G.inertia_group(P)
Subgroup generated by [(1,4,6)(2,5,3)] of (Galois group 6T2 ([3]2) with order 6 of x^6 + 40*x^3 + 1372)
>>> sorted([G.artin_symbol(Q) for Q in K.primes_above(Integer(5))])  # random order, see Issue #18308
[(1,3)(2,6)(4,5), (1,2)(3,4)(5,6), (1,5)(2,4)(3,6)]

Sage Live

P = K.primes_above(2)[0]
G.inertia_group(P)
sorted([G.artin_symbol(Q) for Q in K.primes_above(5)])  # random order, see Issue #18308

If the number field is not Galois over $\mathbf{Q}$, then the galois_group command will construct its Galois closure and return the Galois group of that:

Sage

sage: K.<a> = NumberField(x^3 - 2)
sage: G = K.galois_group(names='b'); G
Galois group 3T2 (S3) with order 6 of x^3 - 2
sage: G.order()
6

Python

>>> from sage.all import *
>>> K = NumberField(x**Integer(3) - Integer(2), names=('a',)); (a,) = K._first_ngens(1)
>>> G = K.galois_group(names='b'); G
Galois group 3T2 (S3) with order 6 of x^3 - 2
>>> G.order()
6

Sage Live

K.<a> = NumberField(x^3 - 2)
G = K.galois_group(names='b'); G
G.order()

Some more Galois groups

We compute two more Galois groups of degree $5$ extensions, and see that one has Galois group $S_5$, so is not solvable by radicals. For these purposes we only need to know the structure of the Galois group as an abstract group, rather than as an explicit group of automorphisms of the splitting field:

Sage

sage: NumberField(x^5 - 2, 'a').galois_group()
Galois group 5T3 (5:4) with order 20 of x^5 - 2
sage: NumberField(x^5 - x + 2, 'a').galois_group()
Galois group 5T5 (S5) with order 120 of x^5 - x + 2

Python

>>> from sage.all import *
>>> NumberField(x**Integer(5) - Integer(2), 'a').galois_group()
Galois group 5T3 (5:4) with order 20 of x^5 - 2
>>> NumberField(x**Integer(5) - x + Integer(2), 'a').galois_group()
Galois group 5T5 (S5) with order 120 of x^5 - x + 2

Sage Live

NumberField(x^5 - 2, 'a').galois_group()
NumberField(x^5 - x + 2, 'a').galois_group()

Magma’s Galois group command

Recent versions of Magma have an algorithm for computing Galois groups that in theory applies when the input polynomial has any degree. There are no open source implementation of this algorithm (as far as I know). If you have Magma, you can use this algorithm from Sage by calling the galois_group function and giving the algorithm='magma' option. The return value is one of the groups in the GAP transitive groups database.

Sage

sage: K.<a> = NumberField(x^3 - 2)
sage: K.galois_group(type="gap", algorithm='magma')  # optional - magma
Galois group Transitive group number 2 of degree 3 of
the Number Field in a with defining polynomial x^3 - 2

Python

>>> from sage.all import *
>>> K = NumberField(x**Integer(3) - Integer(2), names=('a',)); (a,) = K._first_ngens(1)
>>> K.galois_group(type="gap", algorithm='magma')  # optional - magma
Galois group Transitive group number 2 of degree 3 of
the Number Field in a with defining polynomial x^3 - 2

Sage Live

K.<a> = NumberField(x^3 - 2)
K.galois_group(type="gap", algorithm='magma')  # optional - magma

We emphasize that the above example should not work if you don’t have Magma.

Computing complex embeddings

You can also enumerate all complex embeddings of a number field:

Sage

sage: K.complex_embeddings()
[Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> -0.629960524947437 - 1.09112363597172*I,
 Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> -0.629960524947437 + 1.09112363597172*I,
 Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> 1.25992104989487]

Python

>>> from sage.all import *
>>> K.complex_embeddings()
[Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> -0.629960524947437 - 1.09112363597172*I,
 Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> -0.629960524947437 + 1.09112363597172*I,
 Ring morphism:
   From: Number Field in a with defining polynomial x^3 - 2
   To:   Complex Field with 53 bits of precision
   Defn: a |--> 1.25992104989487]

Sage Live

K.complex_embeddings()

Class Numbers and Class Groups

The class group $C_K$ of a number field $K$ is the group of fractional ideals of the maximal order $R$ of $K$ modulo the subgroup of principal fractional ideals. One of the main theorems of algebraic number theory asserts that $C_K$ is a finite group. For example, the quadratic number field $\mathbf{Q}(\sqrt{-23})$ has class number $3$, as we see using the Sage class number command.

Sage

sage: L.<a> = NumberField(x^2 + 23)
sage: L.class_number()
3

Python

>>> from sage.all import *
>>> L = NumberField(x**Integer(2) + Integer(23), names=('a',)); (a,) = L._first_ngens(1)
>>> L.class_number()
3

Sage Live

L.<a> = NumberField(x^2 + 23)
L.class_number()

Quadratic imaginary fields with class number 1

There are only 9 quadratic imaginary field $\mathbf{Q}(\sqrt{D})$ that have class number $1$:

$$D=-3,-4,-7,-8,-11,-19,-43,-67,-163$$

To find this list using Sage, we first experiment with making lists in Sage. For example, typing [1..10] makes the list of integers between $1$ and $10$.

Sage

sage: [1..10]
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

Python

>>> from sage.all import *
>>> (ellipsis_range(Integer(1),Ellipsis,Integer(10)))
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

Sage Live

[1..10]

We can also make the list of odd integers between $1$ and $11$, by typing [1,3,..,11], i.e., by giving the second term in the arithmetic progression.

Sage

sage: [1,3,..,11]
[1, 3, 5, 7, 9, 11]

Python

>>> from sage.all import *
>>> (ellipsis_range(Integer(1),Integer(3),Ellipsis,Integer(11)))
[1, 3, 5, 7, 9, 11]

Sage Live

[1,3,..,11]

Applying this idea, we make the list of negative numbers from $-1$ down to $-10$.

Sage

sage: [-1,-2,..,-10]
[-1, -2, -3, -4, -5, -6, -7, -8, -9, -10]

Python

>>> from sage.all import *
>>> (ellipsis_range(-Integer(1),-Integer(2),Ellipsis,-Integer(10)))
[-1, -2, -3, -4, -5, -6, -7, -8, -9, -10]

Sage Live

[-1,-2,..,-10]

Enumerating quadratic imaginary fields with class number 1

The first two lines below makes a list $v$ of every $D$ from $-1$ down to $-200$ such that $D$ is a fundamental discriminant (the discriminant of a quadratic imaginary field).

Note

Note that you will not see the … in the output below; this … notation just means that part of the output is omitted below.

Sage

sage: w = [-1,-2,..,-200]
sage: v = [D for D in w if D.is_fundamental_discriminant()]
sage: v
[-3, -4, -7, -8, -11, -15, -19, -20, ..., -195, -199]

Python

>>> from sage.all import *
>>> w = (ellipsis_range(-Integer(1),-Integer(2),Ellipsis,-Integer(200)))
>>> v = [D for D in w if D.is_fundamental_discriminant()]
>>> v
[-3, -4, -7, -8, -11, -15, -19, -20, ..., -195, -199]

Sage Live

w = [-1,-2,..,-200]
v = [D for D in w if D.is_fundamental_discriminant()]
v

Finally, we make the list of $D$ in our list $v$ such that the quadratic number field $\mathbf{Q}(\sqrt{D})$ has class number $1$. Notice that QuadraticField(D) is a shorthand for NumberField(x^2 - D).

Sage

sage: [D for D in v if QuadraticField(D,'a').class_number()==1]
[-3, -4, -7, -8, -11, -19, -43, -67, -163]

Python

>>> from sage.all import *
>>> [D for D in v if QuadraticField(D,'a').class_number()==Integer(1)]
[-3, -4, -7, -8, -11, -19, -43, -67, -163]

Sage Live

[D for D in v if QuadraticField(D,'a').class_number()==1]

Of course, we have not proved that this is the list of all negative $D$ so that $\mathbf{Q}(\sqrt{D})$ has class number $1$.

Class number 1 fields

A frustrating open problem is to prove that there are infinitely many number fields with class number $1$. It is quite easy to be convinced that this is probably true by computing a bunch of class numbers of real quadratic fields. For example, over 58 percent of the real quadratic number fields with discriminant $D<1000$ have class number $1$!

Sage

sage: w = [1..1000]
sage: v = [D for D in w if D.is_fundamental_discriminant()]
sage: len(v)
302
sage: len([D for D in v if QuadraticField(D,'a').class_number() == 1])
176
sage: 176.0/302
0.582781456953642

Python

>>> from sage.all import *
>>> w = (ellipsis_range(Integer(1),Ellipsis,Integer(1000)))
>>> v = [D for D in w if D.is_fundamental_discriminant()]
>>> len(v)
302
>>> len([D for D in v if QuadraticField(D,'a').class_number() == Integer(1)])
176
>>> RealNumber('176.0')/Integer(302)
0.582781456953642

Sage Live

w = [1..1000]
v = [D for D in w if D.is_fundamental_discriminant()]
len(v)
len([D for D in v if QuadraticField(D,'a').class_number() == 1])
176.0/302

For more intuition about what is going on, read about the Cohen-Lenstra heuristics.

Class numbers of cyclotomic fields

Sage can also compute class numbers of extensions of higher degree, within reason. Here we use the shorthand CyclotomicField(n) to create the number field $\mathbf{Q}({\zeta }_n)$.

Sage

sage: CyclotomicField(7)
Cyclotomic Field of order 7 and degree 6
sage: for n in [2..15]:
....:     print("{} {}".format(n, CyclotomicField(n).class_number()))
2 1
3 1
...
15 1

Python

>>> from sage.all import *
>>> CyclotomicField(Integer(7))
Cyclotomic Field of order 7 and degree 6
>>> for n in (ellipsis_range(Integer(2),Ellipsis,Integer(15))):
...     print("{} {}".format(n, CyclotomicField(n).class_number()))
2 1
3 1
...
15 1

Sage Live

CyclotomicField(7)
for n in [2..15]:
    print("{} {}".format(n, CyclotomicField(n).class_number()))

In the code above, the notation for n in [2..15]: ... means “do … for $n$ equal to each of the integers $2,3,4,\dots ,15$.”

Note

Exercise: Compute what is omitted (replaced by …) in the output of the previous example.

Assuming conjectures to speed computations

Computations of class numbers and class groups in Sage is done by the Pari C library, and unlike in Pari, by default Sage tells Pari not to assume any conjectures. This can make some commands vastly slower than they might be directly in Pari, which does assume unproved conjectures by default. Fortunately, it is easy to tell Sage to be more permissive and allow Pari to assume conjectures, either just for this one call or henceforth for all number field functions. For example, with proof=False it takes only a few seconds to verify, modulo the conjectures assumed by Pari, that the class number of $\mathbf{Q}({\zeta }_{23})$ is $3$.

Sage

sage: CyclotomicField(23).class_number(proof=False)
3

Python

>>> from sage.all import *
>>> CyclotomicField(Integer(23)).class_number(proof=False)
3

Sage Live

CyclotomicField(23).class_number(proof=False)

Note

Exercise: What is the smallest $n$ such that $\mathbf{Q}({\zeta }_n)$ has class number bigger than $1$?

Class group structure

In addition to computing class numbers, Sage can also compute the group structure and generators for class groups. For example, the quadratic field $\mathbf{Q}(\sqrt{-30})$ has class group $C=(\mathbf{Z}/2\mathbf{Z}{)}^{\oplus 2}$, with generators the ideal classes containing $(5,\sqrt{-30})$ and $(3,\sqrt{-30})$.

Sage

sage: K.<a> = QuadraticField(-30)
sage: C = K.class_group()
sage: C
Class group of order 4 with structure C2 x C2 of Number Field in a with defining polynomial x^2 + 30 with a = 5.477225575051661?*I
sage: category(C)
Category of finite enumerated commutative groups
sage: C.gens()
(Fractional ideal class (5, a), Fractional ideal class (3, a))

Python

>>> from sage.all import *
>>> K = QuadraticField(-Integer(30), names=('a',)); (a,) = K._first_ngens(1)
>>> C = K.class_group()
>>> C
Class group of order 4 with structure C2 x C2 of Number Field in a with defining polynomial x^2 + 30 with a = 5.477225575051661?*I
>>> category(C)
Category of finite enumerated commutative groups
>>> C.gens()
(Fractional ideal class (5, a), Fractional ideal class (3, a))

Sage Live

K.<a> = QuadraticField(-30)
C = K.class_group()
C
category(C)
C.gens()

Arithmetic in the class group

In Sage, the notation C.i means “the $i^{th}$ generator of the object $C$,” where the generators are indexed by numbers $0,1,2,\dots$. Below, when we write C.0 \* C.1, this means “the product of the 0th and 1st generators of the class group $C$.”

Sage

sage: K.<a> = QuadraticField(-30)
sage: C = K.class_group()
sage: C.0
Fractional ideal class (5, a)
sage: C.0.ideal()
Fractional ideal (5, a)
sage: I = C.0 * C.1
sage: I
Fractional ideal class (2, a)

Python

>>> from sage.all import *
>>> K = QuadraticField(-Integer(30), names=('a',)); (a,) = K._first_ngens(1)
>>> C = K.class_group()
>>> C.gen(0)
Fractional ideal class (5, a)
>>> C.gen(0).ideal()
Fractional ideal (5, a)
>>> I = C.gen(0) * C.gen(1)
>>> I
Fractional ideal class (2, a)

Sage Live

K.<a> = QuadraticField(-30)
C = K.class_group()
C.0
C.0.ideal()
I = C.0 * C.1
I

Next we find that the class of the fractional ideal $(2,\sqrt{-30}+4/3)$ is equal to the ideal class $C.0\ast C.1$.

Sage

sage: A = K.ideal([2, a+4/3])
sage: J = C(A)
sage: J
Fractional ideal class (2/3, 1/3*a)
sage: J == C.0*C.1
True

Python

>>> from sage.all import *
>>> A = K.ideal([Integer(2), a+Integer(4)/Integer(3)])
>>> J = C(A)
>>> J
Fractional ideal class (2/3, 1/3*a)
>>> J == C.gen(0)*C.gen(1)
True

Sage Live

A = K.ideal([2, a+4/3])
J = C(A)
J
J == C.0*C.1

Unfortunately, there is currently no Sage function that writes a fractional ideal class in terms of the generators for the class group.

Make live