GAP admits computations in abelian extension fields of the rational number field \(ℚ\), that is fields with abelian Galois group over \(ℚ\). These fields are subfields of cyclotomic fields \(ℚ(e_n)\) where \(e_n = \exp(2 \pi i/n)\) is a primitive complex \(n\)-th root of unity. The elements of these fields are called cyclotomics.
Information concerning operations for domains of cyclotomics, for example certain integral bases of fields of cyclotomics, can be found in Chapter 60. For more general operations that take a field extension as a –possibly optional– argument, e.g., Trace (58.3-5) or Coefficients (61.6-3), see Chapter 58.
‣ E( n ) | ( operation ) |
E returns the primitive n-th root of unity \(e_n = \exp(2\pi i/n)\). Cyclotomics are usually entered as sums of roots of unity, with rational coefficients, and irrational cyclotomics are displayed in such a way. (For special cyclotomics, see 18.4.)
gap> E(9); E(9)^3; E(6); E(12) / 3; -E(9)^4-E(9)^7 E(3) -E(3)^2 -1/3*E(12)^7
A particular basis is used to express cyclotomics, see 60.3; note that E(9) is not a basis element, as the above example shows.
‣ Cyclotomics | ( global variable ) |
is the domain of all cyclotomics.
gap> E(9) in Cyclotomics; 37 in Cyclotomics; true in Cyclotomics; true true false
As the cyclotomics are field elements, the usual arithmetic operators +, -, * and / (and ^ to take powers by integers) are applicable. Note that ^ does not denote the conjugation of group elements, so it is not possible to explicitly construct groups of cyclotomics. (However, it is possible to compute the inverse and the multiplicative order of a nonzero cyclotomic.) Also, taking the \(k\)-th power of a root of unity \(z\) defines a Galois automorphism if and only if \(k\) is coprime to the conductor (see Conductor (18.1-7)) of \(z\).
gap> E(5) + E(3); (E(5) + E(5)^4) ^ 2; E(5) / E(3); E(5) * E(3); -E(15)^2-2*E(15)^8-E(15)^11-E(15)^13-E(15)^14 -2*E(5)-E(5)^2-E(5)^3-2*E(5)^4 E(15)^13 E(15)^8 gap> Order( E(5) ); Order( 1+E(5) ); 5 infinity
‣ IsCyclotomic( obj ) | ( category ) |
‣ IsCyc( obj ) | ( category ) |
Every object in the family CyclotomicsFamily lies in the category IsCyclotomic. This covers integers, rationals, proper cyclotomics, the object infinity (18.2-1), and unknowns (see Chapter 74). All these objects except infinity (18.2-1) and unknowns lie also in the category IsCyc, infinity (18.2-1) lies in (and can be detected from) the category IsInfinity (18.2-1), and unknowns lie in IsUnknown (74.1-3).
gap> IsCyclotomic(0); IsCyclotomic(1/2*E(3)); IsCyclotomic( infinity ); true true true gap> IsCyc(0); IsCyc(1/2*E(3)); IsCyc( infinity ); true true false
‣ IsIntegralCyclotomic( obj ) | ( property ) |
A cyclotomic is called integral or a cyclotomic integer if all coefficients of its minimal polynomial over the rationals are integers. Since the underlying basis of the external representation of cyclotomics is an integral basis (see 60.3), the subring of cyclotomic integers in a cyclotomic field is formed by those cyclotomics for which the external representation is a list of integers. For example, square roots of integers are cyclotomic integers (see 18.4), any root of unity is a cyclotomic integer, character values are always cyclotomic integers, but all rationals which are not integers are not cyclotomic integers.
gap> r:= ER( 5 ); # The square root of 5 ... E(5)-E(5)^2-E(5)^3+E(5)^4 gap> IsIntegralCyclotomic( r ); # ... is a cyclotomic integer. true gap> r2:= 1/2 * r; # This is not a cyclotomic integer, ... 1/2*E(5)-1/2*E(5)^2-1/2*E(5)^3+1/2*E(5)^4 gap> IsIntegralCyclotomic( r2 ); false gap> r3:= 1/2 * r - 1/2; # ... but this is one. E(5)+E(5)^4 gap> IsIntegralCyclotomic( r3 ); true
‣ Int( cyc ) | ( method ) |
The operation Int can be used to find a cyclotomic integer near to an arbitrary cyclotomic, by applying Int (14.2-3) to the coefficients.
gap> Int( E(5)+1/2*E(5)^2 ); Int( 2/3*E(7)-3/2*E(4) ); E(5) -E(4)
‣ String( cyc ) | ( method ) |
The operation String returns for a cyclotomic cyc a string corresponding to the way the cyclotomic is printed by ViewObj (6.3-5) and PrintObj (6.3-5).
gap> String( E(5)+1/2*E(5)^2 ); String( 17/3 ); "E(5)+1/2*E(5)^2" "17/3"
‣ Conductor( cyc ) | ( attribute ) |
‣ Conductor( C ) | ( attribute ) |
For an element cyc of a cyclotomic field, Conductor returns the smallest integer \(n\) such that cyc is contained in the \(n\)-th cyclotomic field. For a collection C of cyclotomics (for example a dense list of cyclotomics or a field of cyclotomics), Conductor returns the smallest integer \(n\) such that all elements of C are contained in the \(n\)-th cyclotomic field.
gap> Conductor( 0 ); Conductor( E(10) ); Conductor( E(12) ); 1 5 12
‣ AbsoluteValue( cyc ) | ( attribute ) |
returns the absolute value of a cyclotomic number cyc. At the moment only methods for rational numbers exist.
gap> AbsoluteValue(-3); 3
‣ RoundCyc( cyc ) | ( operation ) |
is a cyclotomic integer \(z\) (see IsIntegralCyclotomic (18.1-4)) near to the cyclotomic cyc in the following sense: Let c be the \(i\)-th coefficient in the external representation (see CoeffsCyc (18.1-10)) of cyc. Then the \(i\)-th coefficient in the external representation of \(z\) is Int( c + 1/2 ) or Int( c - 1/2 ), depending on whether c is nonnegative or negative, respectively.
Expressed in terms of the Zumbroich basis (see 60.3), rounding the coefficients of cyc w.r.t. this basis to the nearest integer yields the coefficients of \(z\).
gap> RoundCyc( E(5)+1/2*E(5)^2 ); RoundCyc( 2/3*E(7)+3/2*E(4) ); E(5)+E(5)^2 -2*E(28)^3+E(28)^4-2*E(28)^11-2*E(28)^15-2*E(28)^19-2*E(28)^23 -2*E(28)^27
‣ CoeffsCyc( cyc, N ) | ( function ) |
Let cyc be a cyclotomic with conductor \(n\) (see Conductor (18.1-7)). If N is not a multiple of \(n\) then CoeffsCyc returns fail because cyc cannot be expressed in terms of N-th roots of unity. Otherwise CoeffsCyc returns a list of length N with entry at position \(j\) equal to the coefficient of \(\exp(2 \pi i (j-1)/\textit{N})\) if this root belongs to the N-th Zumbroich basis (see 60.3), and equal to zero otherwise. So we have cyc = CoeffsCyc( cyc, N ) * List( [1..N], j -> E(N)^(j-1) ).
gap> cyc:= E(5)+E(5)^2; E(5)+E(5)^2 gap> CoeffsCyc( cyc, 5 ); CoeffsCyc( cyc, 15 ); CoeffsCyc( cyc, 7 ); [ 0, 1, 1, 0, 0 ] [ 0, -1, 0, 0, 0, 0, 0, 0, -1, 0, 0, -1, 0, -1, 0 ] fail
‣ DenominatorCyc( cyc ) | ( function ) |
For a cyclotomic number cyc (see IsCyclotomic (18.1-3)), this function returns the smallest positive integer \(n\) such that \(n\) * cyc is a cyclotomic integer (see IsIntegralCyclotomic (18.1-4)). For rational numbers cyc, the result is the same as that of DenominatorRat (17.2-5).
‣ ExtRepOfObj( cyc ) | ( method ) |
The external representation of a cyclotomic cyc with conductor \(n\) (see Conductor (18.1-7) is the list returned by CoeffsCyc (18.1-10), called with cyc and \(n\).
gap> ExtRepOfObj( E(5) ); CoeffsCyc( E(5), 5 ); [ 0, 1, 0, 0, 0 ] [ 0, 1, 0, 0, 0 ] gap> CoeffsCyc( E(5), 15 ); [ 0, 0, 0, 0, 0, 0, 0, 0, -1, 0, 0, 0, 0, -1, 0 ]
‣ DescriptionOfRootOfUnity( root ) | ( function ) |
Given a cyclotomic root that is known to be a root of unity (this is not checked), DescriptionOfRootOfUnity returns a list \([ n, e ]\) of coprime positive integers such that root \(=\) E\((n)^e\) holds.
gap> E(9); DescriptionOfRootOfUnity( E(9) ); -E(9)^4-E(9)^7 [ 9, 1 ] gap> DescriptionOfRootOfUnity( -E(3) ); [ 6, 5 ]
‣ IsGaussInt( x ) | ( function ) |
IsGaussInt returns true if the object x is a Gaussian integer (see GaussianIntegers (60.5-1)), and false otherwise. Gaussian integers are of the form \(a + b\)*E(4), where \(a\) and \(b\) are integers.
‣ IsGaussRat( x ) | ( function ) |
IsGaussRat returns true if the object x is a Gaussian rational (see GaussianRationals (60.1-3)), and false otherwise. Gaussian rationals are of the form \(a + b\)*E(4), where \(a\) and \(b\) are rationals.
‣ DefaultField( list ) | ( function ) |
DefaultField for cyclotomics is defined to return the smallest cyclotomic field containing the given elements.
Note that Field (58.1-3) returns the smallest field containing all given elements, which need not be a cyclotomic field. In both cases, the fields represent vector spaces over the rationals (see 60.3).
gap> Field( E(5)+E(5)^4 ); DefaultField( E(5)+E(5)^4 ); NF(5,[ 1, 4 ]) CF(5)
‣ IsInfinity( obj ) | ( category ) |
‣ IsNegInfinity( obj ) | ( category ) |
‣ infinity | ( global variable ) |
‣ -infinity | ( global variable ) |
infinity and -infinity are special GAP objects that lie in CyclotomicsFamily. They are larger or smaller than all other objects in this family respectively. infinity is mainly used as return value of operations such as Size (30.4-6) and Dimension (57.3-3) for infinite and infinite dimensional domains, respectively.
Some arithmetic operations are provided for convenience when using infinity and -infinity as top and bottom element respectively.
gap> -infinity + 1; -infinity gap> infinity + infinity; infinity
Often it is useful to distinguish infinity from proper
cyclotomics. For that, infinity lies in the category IsInfinity but not in IsCyc (18.1-3), and the other cyclotomics lie in the category IsCyc (18.1-3) but not in IsInfinity.
gap> s:= Size( Rationals ); infinity gap> s = infinity; IsCyclotomic( s ); IsCyc( s ); IsInfinity( s ); true true false true gap> s in Rationals; s > 17; false true gap> Set( [ s, 2, s, E(17), s, 19 ] ); [ 2, 19, E(17), infinity ]
To compare cyclotomics, the operators <, <=, =, >=, >, and <> can be used, the result will be true if the first operand is smaller, smaller or equal, equal, larger or equal, larger, or unequal, respectively, and false otherwise.
Cyclotomics are ordered as follows: The relation between rationals is the natural one, rationals are smaller than irrational cyclotomics, and infinity (18.2-1) is the largest cyclotomic. For two irrational cyclotomics with different conductors (see Conductor (18.1-7)), the one with smaller conductor is regarded as smaller. Two irrational cyclotomics with same conductor are compared via their external representation (see ExtRepOfObj (18.1-12)).
For comparisons of cyclotomics and other GAP objects, see Section 4.13.
gap> E(5) < E(6); # the latter value has conductor 3 false gap> E(3) < E(3)^2; # both have conductor 3, compare the ext. repr. false gap> 3 < E(3); E(5) < E(7); true true
‣ EB( N ) | ( function ) |
‣ EC( N ) | ( function ) |
‣ ED( N ) | ( function ) |
‣ EE( N ) | ( function ) |
‣ EF( N ) | ( function ) |
‣ EG( N ) | ( function ) |
‣ EH( N ) | ( function ) |
For a positive integer N, let \(z =\) E(N) \(= \exp(2 \pi i/\textit{N})\). The following so-called atomic irrationalities (see [CCN+85, Chapter 7, Section 10]) can be entered using functions. (Note that the values are not necessary irrational.)
EB(N) |
= | \(b_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^2}} \right) / 2\) | , | \(\textit{N} \equiv 1 \pmod{2}\) |
EC(N) |
= | \(c_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^3}} \right) / 3\) | , | \(\textit{N} \equiv 1 \pmod{3}\) |
ED(N) |
= | \(d_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^4}} \right) / 4\) | , | \(\textit{N} \equiv 1 \pmod{4}\) |
EE(N) |
= | \(e_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^5}} \right) / 5\) | , | \(\textit{N} \equiv 1 \pmod{5}\) |
EF(N) |
= | \(f_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^6}} \right) / 6\) | , | \(\textit{N} \equiv 1 \pmod{6}\) |
EG(N) |
= | \(g_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^7}} \right) / 7\) | , | \(\textit{N} \equiv 1 \pmod{7}\) |
EH(N) |
= | \(h_{\textit{N}}\) | = | \(\left( \sum_{{j = 1}}^{{\textit{N}-1}} z^{{j^8}} \right) / 8\) | , | \(\textit{N} \equiv 1 \pmod{8}\) |
(Note that in EC(N), \(\ldots\), EH(N), N must be a prime.)
gap> EB(5); EB(9); E(5)+E(5)^4 1
‣ EI( N ) | ( function ) |
‣ ER( N ) | ( function ) |
For a rational number N, ER returns the square root \(\sqrt{{\textit{N}}}\) of N, and EI returns \(\sqrt{{-\textit{N}}}\). By the chosen embedding of cyclotomic fields into the complex numbers, ER returns the positive square root if N is positive, and if N is negative then ER(N) = EI(-N) holds. In any case, EI(N) = E(4) * ER(N).
ER is installed as method for the operation Sqrt (31.12-5), for rational argument.
From a theorem of Gauss we know that \(b_{\textit{N}} =\)
| \((-1 + \sqrt{{\textit{N}}}) / 2\) | if | \(\textit{N} \equiv 1 \pmod 4\) |
| \((-1 + i \sqrt{{\textit{N}}}) / 2\) | if | \(\textit{N} \equiv -1 \pmod 4\) |
So \(\sqrt{{\textit{N}}}\) can be computed from \(b_{\textit{N}}\), see EB (18.4-1).
gap> ER(3); EI(3); -E(12)^7+E(12)^11 E(3)-E(3)^2
‣ EY( N[, d] ) | ( function ) |
‣ EX( N[, d] ) | ( function ) |
‣ EW( N[, d] ) | ( function ) |
‣ EV( N[, d] ) | ( function ) |
‣ EU( N[, d] ) | ( function ) |
‣ ET( N[, d] ) | ( function ) |
‣ ES( N[, d] ) | ( function ) |
For the given integer N \(> 2\), let \(\textit{N}_k\) denote the first integer with multiplicative order exactly \(k\) modulo N, chosen in the order of preference
\[ 1, -1, 2, -2, 3, -3, 4, -4, \ldots . \]
We define (with \(z = \exp(2 \pi i/\textit{N})\))
EY(N) |
= | \(y_{\textit{N}}\) | = | \(z + z^n\) | \((n = \textit{N}_2)\) |
EX(N) |
= | \(x_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}}\) | \((n = \textit{N}_3)\) |
EW(N) |
= | \(w_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}} + z^{{n^3}}\) | \((n = \textit{N}_4)\) |
EV(N) |
= | \(v_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}} + z^{{n^3}} + z^{{n^4}}\) | \((n = \textit{N}_5)\) |
EU(N) |
= | \(u_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}} + \ldots + z^{{n^5}}\) | \((n = \textit{N}_6)\) |
ET(N) |
= | \(t_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}} + \ldots + z^{{n^6}}\) | \((n = \textit{N}_7)\) |
ES(N) |
= | \(s_{\textit{N}}\) | = | \(z + z^n + z^{{n^2}} + \ldots + z^{{n^7}}\) | \((n = \textit{N}_8)\) |
For the two-argument versions of the functions, see Section NK (18.4-5).
gap> EY(5); E(5)+E(5)^4 gap> EW(16,3); EW(17,2); 0 E(17)+E(17)^4+E(17)^13+E(17)^16
‣ EM( N[, d] ) | ( function ) |
‣ EL( N[, d] ) | ( function ) |
‣ EK( N[, d] ) | ( function ) |
‣ EJ( N[, d] ) | ( function ) |
Let N be an integer, N \(> 2\). We define (with \(z = \exp(2 \pi i/\textit{N})\))
EM(N) |
= | \(m_{\textit{N}}\) | = | \(z - z^n\) | \((n = \textit{N}_2)\) |
EL(N) |
= | \(l_{\textit{N}}\) | = | \(z - z^n + z^{{n^2}} - z^{{n^3}}\) | \((n = \textit{N}_4)\) |
EK(N) |
= | \(k_{\textit{N}}\) | = | \(z - z^n + \ldots - z^{{n^5}}\) | \((n = \textit{N}_6)\) |
EJ(N) |
= | \(j_{\textit{N}}\) | = | \(z - z^n + \ldots - z^{{n^7}}\) | \((n = \textit{N}_8)\) |
For the two-argument versions of the functions, see Section NK (18.4-5).
‣ NK( N, k, d ) | ( function ) |
Let \(\textit{N}_{\textit{k}}^{(\textit{d})}\) be the \((\textit{d}+1)\)-th integer with multiplicative order exactly k modulo N, chosen in the order of preference defined in Section 18.4-3; NK returns \(\textit{N}_{\textit{k}}^{(\textit{d})}\); if there is no integer with the required multiplicative order, NK returns fail.
We write \(\textit{N}_{\textit{k}} = \textit{N}_{\textit{k}}^{(0)}, \textit{N}_{\textit{k}}^{\prime} = \textit{N}_{\textit{k}}^{(1)}, \textit{N}_{\textit{k}}^{\prime\prime} = \textit{N}_{\textit{k}}^{(2)}\) and so on.
The algebraic numbers
\[ y_{N}^{\prime} = y_{N}^{(1)}, y_{N}^{\prime\prime} = y_{N}^{(2)}, \ldots, x_{N}^{\prime}, x_{N}^{\prime\prime}, \ldots, j_{N}^{\prime}, j_{N}^{\prime\prime}, \ldots \]
are obtained on replacing \(\textit{N}_{\textit{k}}\) in the definitions in the sections 18.4-3 and 18.4-4 by \(\textit{N}_{\textit{k}}^{\prime}, \textit{N}_{\textit{k}}^{\prime\prime}, \ldots\); they can be entered as
EY(N,d) |
= | \(y_{\textit{N}}^{(\textit{d})}\) |
EX(N,d) |
= | \(x_{\textit{N}}^{(\textit{d})}\) |
| \(\ldots\) | ||
EJ(N,d) |
= | \(j_{\textit{N}}^{(\textit{d})}\) |
‣ AtlasIrrationality( irratname ) | ( function ) |
Let irratname be a string that describes an irrational value as a linear combination in terms of the atomic irrationalities introduced in the sections 18.4-1, 18.4-2, 18.4-3, 18.4-4. These irrational values are defined in [CCN+85, Chapter 6, Section 10], and the following description is mainly copied from there. If \(q_N\) is such a value (e.g. \(y_{24}^{\prime\prime}\)) then linear combinations of algebraic conjugates of \(q_N\) are abbreviated as in the following examples:
2qN+3&5-4&7+&9 |
means | \(2 q_N + 3 q_N^{{*5}} - 4 q_N^{{*7}} + q_N^{{*9}}\) |
4qN&3&5&7-3&4 |
means | \(4 (q_N + q_N^{{*3}} + q_N^{{*5}} + q_N^{{*7}}) - 3 q_N^{{*11}}\) |
4qN*3&5+&7 |
means | \(4 (q_N^{{*3}} + q_N^{{*5}}) + q_N^{{*7}}\) |
To explain the ampersand
syntax in general we remark that &k
is interpreted as \(q_N^{{*k}}\), where \(q_N\) is the most recently named atomic irrationality, and that the scope of any premultiplying coefficient is broken by a \(+\) or \(-\) sign, but not by \(\&\) or \(*k\). The algebraic conjugations indicated by the ampersands apply directly to the atomic irrationality \(q_N\), even when, as in the last example, \(q_N\) first appears with another conjugacy \(*k\).
gap> AtlasIrrationality( "b7*3" ); E(7)^3+E(7)^5+E(7)^6 gap> AtlasIrrationality( "y'''24" ); E(24)-E(24)^19 gap> AtlasIrrationality( "-3y'''24*13&5" ); 3*E(8)-3*E(8)^3 gap> AtlasIrrationality( "3y'''24*13-2&5" ); -3*E(24)-2*E(24)^11+2*E(24)^17+3*E(24)^19 gap> AtlasIrrationality( "3y'''24*13-&5" ); -3*E(24)-E(24)^11+E(24)^17+3*E(24)^19 gap> AtlasIrrationality( "3y'''24*13-4&5&7" ); -7*E(24)-4*E(24)^11+4*E(24)^17+7*E(24)^19 gap> AtlasIrrationality( "3y'''24&7" ); 6*E(24)-6*E(24)^19
‣ GaloisCyc( cyc, k ) | ( operation ) |
‣ GaloisCyc( list, k ) | ( operation ) |
For a cyclotomic cyc and an integer k, GaloisCyc returns the cyclotomic obtained by raising the roots of unity in the Zumbroich basis representation of cyc to the k-th power. If k is coprime to the integer \(n\), GaloisCyc( ., k ) acts as a Galois automorphism of the \(n\)-th cyclotomic field (see 60.4); to get the Galois automorphisms themselves, use GaloisGroup (58.3-1).
The complex conjugate of cyc is GaloisCyc( cyc, -1 ), which can also be computed using ComplexConjugate (18.5-2).
For a list or matrix list of cyclotomics, GaloisCyc returns the list obtained by applying GaloisCyc to the entries of list.
‣ ComplexConjugate( z ) | ( attribute ) |
‣ RealPart( z ) | ( attribute ) |
‣ ImaginaryPart( z ) | ( attribute ) |
For a cyclotomic number z, ComplexConjugate returns GaloisCyc( z, -1 ), see GaloisCyc (18.5-1). For a quaternion \(\textit{z} = c_1 e + c_2 i + c_3 j + c_4 k\), ComplexConjugate returns \(c_1 e - c_2 i - c_3 j - c_4 k\), see IsQuaternion (62.8-8).
When ComplexConjugate is called with a list then the result is the list of return values of ComplexConjugate for the list entries in the corresponding positions.
When ComplexConjugate is defined for an object z then RealPart and ImaginaryPart return (z + ComplexConjugate( z )) / 2 and (z - ComplexConjugate( z )) / 2 i, respectively, where i denotes the corresponding imaginary unit.
gap> GaloisCyc( E(5) + E(5)^4, 2 ); E(5)^2+E(5)^3 gap> GaloisCyc( E(5), -1 ); # the complex conjugate E(5)^4 gap> GaloisCyc( E(5) + E(5)^4, -1 ); # this value is real E(5)+E(5)^4 gap> GaloisCyc( E(15) + E(15)^4, 3 ); E(5)+E(5)^4 gap> ComplexConjugate( E(7) ); E(7)^6
‣ StarCyc( cyc ) | ( function ) |
If the cyclotomic cyc is an irrational element of a quadratic extension of the rationals then StarCyc returns the unique Galois conjugate of cyc that is different from cyc, otherwise fail is returned. In the first case, the return value is often called cyc\(*\) (see 71.13).
gap> StarCyc( EB(5) ); StarCyc( E(5) ); E(5)^2+E(5)^3 fail
‣ Quadratic( cyc ) | ( function ) |
Let cyc be a cyclotomic integer that lies in a quadratic extension field of the rationals. Then we have cyc\( = (a + b \sqrt{{n}}) / d\), for integers \(a\), \(b\), \(n\), \(d\), such that \(d\) is either \(1\) or \(2\). In this case, Quadratic returns a record with the components a, b, root, d, ATLAS, and display; the values of the first four are \(a\), \(b\), \(n\), and \(d\), the ATLAS value is a (not necessarily shortest) representation of cyc in terms of the Atlas irrationalities \(b_{{|n|}}\), \(i_{{|n|}}\), \(r_{{|n|}}\), and the display value is a string that expresses cyc in GAP notation, corresponding to the value of the ATLAS component.
If cyc is not a cyclotomic integer or does not lie in a quadratic extension field of the rationals then fail is returned.
If the denominator \(d\) is \(2\) then necessarily \(n\) is congruent to \(1\) modulo \(4\), and \(r_n\), \(i_n\) are not possible; we have cyc = x + y * EB( root ) with y = b, x = ( a + b ) / 2.
If \(d = 1\), we have the possibilities \(i_{{|n|}}\) for \(n < -1\), \(a + b * i\) for \(n = -1\), \(a + b * r_n\) for \(n > 0\). Furthermore if \(n\) is congruent to \(1\) modulo \(4\), also cyc \(= (a+b) + 2 * b * b_{{|n|}}\) is possible; the shortest string of these is taken as the value for the component ATLAS.
gap> Quadratic( EB(5) ); Quadratic( EB(27) ); rec( ATLAS := "b5", a := -1, b := 1, d := 2, display := "(-1+Sqrt(5))/2", root := 5 ) rec( ATLAS := "1+3b3", a := -1, b := 3, d := 2, display := "(-1+3*Sqrt(-3))/2", root := -3 ) gap> Quadratic(0); Quadratic( E(5) ); rec( ATLAS := "0", a := 0, b := 0, d := 1, display := "0", root := 1 ) fail
‣ GaloisMat( mat ) | ( attribute ) |
Let mat be a matrix of cyclotomics. GaloisMat calculates the complete orbits under the operation of the Galois group of the (irrational) entries of mat, and the permutations of rows corresponding to the generators of the Galois group.
If some rows of mat are identical, only the first one is considered for the permutations, and a warning will be printed.
GaloisMat returns a record with the components mat, galoisfams, and generators.
mata list with initial segment being the rows of mat (not shallow copies of these rows); the list consists of full orbits under the action of the Galois group of the entries of mat defined above. The last rows in the list are those not contained in mat but must be added in order to complete the orbits; so if the orbits were already complete, mat and mat have identical rows.
galoisfamsa list that has the same length as the mat component, its entries are either 1, 0, -1, or lists.
galoisfams[i] = 1means that mat[i] consists of rationals, i.e., [ mat[i] ] forms an orbit;
galoisfams[i] = -1means that mat[i] contains unknowns (see Chapter 74); in this case [ mat[i] ] is regarded as an orbit, too, even if mat[i] contains irrational entries;
galoisfams[i] = \([ l_1, l_2 ]\)(a list) means that mat[i] is the first element of its orbit in mat, \(l_1\) is the list of positions of rows that form the orbit, and \(l_2\) is the list of corresponding Galois automorphisms (as exponents, not as functions); so we have mat\([ l_1[j] ][k] = \) GaloisCyc( mat\([i][k], l_2[j]\) );
galoisfams[i] = 0means that mat[i] is an element of a nontrivial orbit but not the first element of it.
generatorsa list of permutations generating the permutation group corresponding to the action of the Galois group on the rows of mat.
gap> GaloisMat( [ [ E(3), E(4) ] ] ); rec( galoisfams := [ [ [ 1, 2, 3, 4 ], [ 1, 7, 5, 11 ] ], 0, 0, 0 ], generators := [ (1,2)(3,4), (1,3)(2,4) ], mat := [ [ E(3), E(4) ], [ E(3), -E(4) ], [ E(3)^2, E(4) ], [ E(3)^2, -E(4) ] ] ) gap> GaloisMat( [ [ 1, 1, 1 ], [ 1, E(3), E(3)^2 ] ] ); rec( galoisfams := [ 1, [ [ 2, 3 ], [ 1, 2 ] ], 0 ], generators := [ (2,3) ], mat := [ [ 1, 1, 1 ], [ 1, E(3), E(3)^2 ], [ 1, E(3)^2, E(3) ] ] )
‣ RationalizedMat( mat ) | ( attribute ) |
returns the list of rationalized rows of mat, which must be a matrix of cyclotomics. This is the set of sums over orbits under the action of the Galois group of the entries of mat (see GaloisMat (18.5-5)), so the operation may be viewed as a kind of trace on the rows.
Note that no two rows of mat should be equal.
gap> mat:= [ [ 1, 1, 1 ], [ 1, E(3), E(3)^2 ], [ 1, E(3)^2, E(3) ] ];; gap> RationalizedMat( mat ); [ [ 1, 1, 1 ], [ 2, -1, -1 ] ]
The implementation of an internally represented cyclotomic is based on a list of length equal to its conductor. This means that the internal representation of a cyclotomic does not refer to the smallest number field but the smallest cyclotomic field containing it. The reason for this is the wish to reflect the natural embedding of two cyclotomic fields into a larger one that contains both. With such embeddings, it is easy to construct the sum or the product of two arbitrary cyclotomics (in possibly different fields) as an element of a cyclotomic field.
The disadvantage of this approach is that the arithmetical operations are quite expensive, so the use of internally represented cyclotomics is not recommended for doing arithmetics over number fields, such as calculations with matrices of cyclotomics. But internally represented cyclotomics are good enough for dealing with irrationalities in character tables (see chapter 71).
For the representation of cyclotomics one has to recall that the \(n\)-th cyclotomic field \(ℚ(e_n)\) is a vector space of dimension \(\varphi(n)\) over the rationals where \(\varphi\) denotes Euler's phi-function (see Phi (15.2-2)).
A special integral basis of cyclotomic fields is chosen that allows one to easily convert arbitrary sums of roots of unity into the basis, as well as to convert a cyclotomic represented w.r.t. the basis into the smallest possible cyclotomic field. This basis is accessible in GAP, see 60.3 for more information and references.
Note that the set of all \(n\)-th roots of unity is linearly dependent for \(n > 1\), so multiplication is not the multiplication of the group ring \(ℚ\langle e_n \rangle\); given a \(ℚ\)-basis of \(ℚ(e_n)\) the result of the multiplication (computed as multiplication of polynomials in \(e_n\), using \((e_n)^n = 1\)) will be converted to the basis.
gap> E(5) * E(5)^2; ( E(5) + E(5)^4 ) * E(5)^2; E(5)^3 E(5)+E(5)^3 gap> ( E(5) + E(5)^4 ) * E(5); -E(5)-E(5)^3-E(5)^4
An internally represented cyclotomic is always represented in the smallest cyclotomic field it is contained in. The internal coefficients list coincides with the external representation returned by ExtRepOfObj (18.1-12).
To avoid calculations becoming unintentionally very long, or consuming very large amounts of memory, there is a limit on the conductor of internally represented cyclotomics, by default set to one million. This can be raised (although not lowered) using SetCyclotomicsLimit (18.6-1) and accessed using GetCyclotomicsLimit (18.6-1). The maximum value of the limit is \(2^{28}-1\) on \(32\) bit systems, and \(2^{32}-1\) on \(64\) bit systems. So the maximal cyclotomic field implemented in GAP is not really the field \(ℚ^{ab}\).
It should be emphasized that one disadvantage of representing a cyclotomic in the smallest cyclotomic field (and not in the smallest field) is that arithmetic operations in a fixed small extension field of the rational number field are comparatively expensive. For example, take a prime integer \(p\) and suppose that we want to work with a matrix group over the field \(ℚ(\sqrt{{p}})\). Then each matrix entry could be described by two rational coefficients, whereas the representation in the smallest cyclotomic field requires \(p-1\) rational coefficients for each entry. So it is worth thinking about using elements in a field constructed with AlgebraicExtension (67.1-1) when natural embeddings of cyclotomic fields are not needed.
‣ SetCyclotomicsLimit( newlimit ) | ( function ) |
‣ GetCyclotomicsLimit( ) | ( function ) |
GetCyclotomicsLimit returns the current limit on conductors of internally represented cyclotomic numbers
SetCyclotomicsLimit can be called to increase the limit on conductors of internally represented cyclotomic numbers. Note that computing in large cyclotomic fields using this representation can be both slow and memory-consuming, and that other approaches may be better for some problems. See 18.6.
generated by GAPDoc2HTML