Goto Chapter: Top 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 Bib Ind

14 Integers

One of the most fundamental datatypes in every programming language is the integer type. **GAP** is no exception.

**GAP** integers are entered as a sequence of decimal digits optionally preceded by a "`+`

" sign for positive integers or a "`-`

" sign for negative integers. The size of integers in **GAP** is only limited by the amount of available memory, so you can compute with integers having thousands of digits.

gap> -1234; -1234 gap> 123456789012345678901234567890123456789012345678901234567890; 123456789012345678901234567890123456789012345678901234567890

Note that in a few places, only certain small integer values can be used. A *small integer* (also referred to as *immediate integer*) is an integer \(n\) satisfying `INTOBJ_MIN`

\( \leq n \leq \) `INTOBJ_MAX`

, where `INTOBJ_MIN`

and `INTOBJ_MAX`

equal either \(-2^{28}\) and \(2^{28}-1\) (on 32-bit systems) or \(-2^{60}\) and \(2^{60}-1\) (on 64-bit systems). For example, the elements of a range are restricted to small integers (see 21.22).

Many more functions that are mainly related to the prime residue group of integers modulo an integer are described in chapter 15, and functions dealing with combinatorics can be found in chapter 16.

`‣ Integers` | ( global variable ) |

`‣ PositiveIntegers` | ( global variable ) |

`‣ NonnegativeIntegers` | ( global variable ) |

These global variables represent the ring of integers and the semirings of positive and nonnegative integers, respectively.

gap> Size( Integers ); 2 in Integers; infinity true

`Integers`

is a subset of `Rationals`

(17.1-1), which is a subset of `Cyclotomics`

(18.1-2). See Chapter 18 for arithmetic operations and comparison of integers.

`‣ IsIntegers` ( obj ) | ( category ) |

`‣ IsPositiveIntegers` ( obj ) | ( category ) |

`‣ IsNonnegativeIntegers` ( obj ) | ( category ) |

are the defining categories for the domains `Integers`

(14.1-1), `PositiveIntegers`

(14.1-1), and `NonnegativeIntegers`

(14.1-1).

gap> IsIntegers( Integers ); IsIntegers( Rationals ); IsIntegers( 7 ); true false false

`‣ IsInt` ( obj ) | ( category ) |

Every rational integer lies in the category `IsInt`

, which is a subcategory of `IsRat`

(17.2-1).

`‣ IsPosInt` ( obj ) | ( category ) |

Every positive integer lies in the category `IsPosInt`

.

`‣ Int` ( elm ) | ( attribute ) |

`Int`

returns an integer `int`

whose meaning depends on the type of `elm`. For example:

If `elm` is a rational number (see Chapter 17) then `int`

is the integer part of the quotient of numerator and denominator of `elm` (see `QuoInt`

(14.3-1)).

If `elm` is an element of a finite prime field (see Chapter 59) then `int`

is the smallest nonnegative integer such that

.`elm` = int * One( `elm` )

If `elm` is a string (see Chapter 27) consisting entirely of decimal digits `'0'`

, `'1'`

, \(\ldots\), `'9'`

, and optionally a sign `'-'`

(at the first position), then `int`

is the integer described by this string. For all other strings, `fail`

is returned. See `Int`

(27.9-1).

The operation `String`

(27.7-6) can be used to compute a string for rational integers, in fact for all cyclotomics.

gap> Int( 4/3 ); Int( -2/3 ); 1 0 gap> int:= Int( Z(5) ); int * One( Z(5) ); 2 Z(5) gap> Int( "12345" ); Int( "-27" ); Int( "-27/3" ); 12345 -27 fail

`‣ IsEvenInt` ( n ) | ( function ) |

tests if the integer `n` is divisible by 2.

`‣ IsOddInt` ( n ) | ( function ) |

tests if the integer `n` is not divisible by 2.

`‣ AbsInt` ( n ) | ( function ) |

`AbsInt`

returns the absolute value of the integer `n`, i.e., `n` if `n` is positive, -`n` if `n` is negative and 0 if `n` is 0.

`AbsInt`

is a special case of the general operation `EuclideanDegree`

(56.6-2).

See also `AbsoluteValue`

(18.1-8).

gap> AbsInt( 33 ); 33 gap> AbsInt( -214378 ); 214378 gap> AbsInt( 0 ); 0

`‣ SignInt` ( n ) | ( function ) |

`SignInt`

returns the sign of the integer `n`, i.e., 1 if `n` is positive, -1 if `n` is negative and 0 if `n` is 0.

gap> SignInt( 33 ); 1 gap> SignInt( -214378 ); -1 gap> SignInt( 0 ); 0

`‣ LogInt` ( n, base ) | ( function ) |

`LogInt`

returns the integer part of the logarithm of the positive integer `n` with respect to the positive integer `base`, i.e., the largest positive integer \(e\) such that \(\textit{base}^e \leq \textit{n}\). The function `LogInt`

will signal an error if either `n` or `base` is not positive.

For `base` \(= 2\) this is very efficient because the internal binary representation of the integer is used.

gap> LogInt( 1030, 2 ); 10 gap> 2^10; 1024 gap> LogInt( 1, 10 ); 0

`‣ RootInt` ( n[, k] ) | ( function ) |

`RootInt`

returns the integer part of the `k`th root of the integer `n`. If the optional integer argument `k` is not given it defaults to 2, i.e., `RootInt`

returns the integer part of the square root in this case.

If `n` is positive, `RootInt`

returns the largest positive integer \(r\) such that \(r^{\textit{k}} \leq \textit{n}\). If `n` is negative and `k` is odd `RootInt`

returns `-RootInt( -`

. If `n`, `k` )`n` is negative and `k` is even `RootInt`

will cause an error. `RootInt`

will also cause an error if `k` is 0 or negative.

gap> RootInt( 361 ); 19 gap> RootInt( 2 * 10^12 ); 1414213 gap> RootInt( 17000, 5 ); 7 gap> 7^5; 16807

`‣ SmallestRootInt` ( n ) | ( function ) |

`SmallestRootInt`

returns the smallest root of the integer `n`.

The smallest root of an integer `n` is the integer \(r\) of smallest absolute value for which a positive integer \(k\) exists such that \(\textit{n} = r^k\).

gap> SmallestRootInt( 2^30 ); 2 gap> SmallestRootInt( -(2^30) ); -4

Note that \((-2)^{30} = +(2^{30})\).

gap> SmallestRootInt( 279936 ); 6 gap> LogInt( 279936, 6 ); 7 gap> SmallestRootInt( 1001 ); 1001

`‣ IsSquareInt` ( n ) | ( function ) |

`IsSquareInt`

tests whether the integer `n` is the square of an integer or not. This test is much faster than the simpler `RootInt`

\((n)^2=n\) because it first tests whether `n` is a square residue modulo some small integers.

`‣ ListOfDigits` ( n ) | ( function ) |

For a positive integer `n` this function returns a list `l`, consisting of the digits of `n` in decimal representation.

gap> ListOfDigits(3142); [ 3, 1, 4, 2 ]

`‣ Random` ( Integers ) | ( method ) |

`Random`

for integers returns pseudo random integers between \(-10\) and \(10\) distributed according to a binomial distribution. To generate uniformly distributed integers from a range, use the construction `Random( [ `

(see `low` .. `high` ] )`Random`

(30.7-1)).

`‣ QuoInt` ( n, m ) | ( function ) |

`QuoInt`

returns the integer part of the quotient of its integer operands.

If `n` and `m` are positive, `QuoInt`

returns the largest positive integer \(q\) such that \(q * \textit{m} \leq \textit{n}\). If `n` or `m` or both are negative the absolute value of the integer part of the quotient is the quotient of the absolute values of `n` and `m`, and the sign of it is the product of the signs of `n` and `m`.

`QuoInt`

is used in a method for the general operation `EuclideanQuotient`

(56.6-3).

gap> QuoInt(5,3); QuoInt(-5,3); QuoInt(5,-3); QuoInt(-5,-3); 1 -1 -1 1

`‣ BestQuoInt` ( n, m ) | ( function ) |

`BestQuoInt`

returns the best quotient \(q\) of the integers `n` and `m`. This is the quotient such that \(\textit{n}-q*\textit{m}\) has minimal absolute value. If there are two quotients whose remainders have the same absolute value, then the quotient with the smaller absolute value is chosen.

gap> BestQuoInt( 5, 3 ); BestQuoInt( -5, 3 ); 2 -2

`‣ RemInt` ( n, m ) | ( function ) |

`RemInt`

returns the remainder of its two integer operands.

If `m` is not equal to zero, `RemInt`

returns

. Note that the rules given for `n` - `m` * QuoInt( `n`, `m` )`QuoInt`

(14.3-1) imply that the return value of `RemInt`

has the same sign as `n` and its absolute value is strictly less than the absolute value of `m`. Note also that the return value equals

when both `n` mod `m``n` and `m` are nonnegative. Dividing by `0`

signals an error.

`RemInt`

is used in a method for the general operation `EuclideanRemainder`

(56.6-4).

gap> RemInt(5,3); RemInt(-5,3); RemInt(5,-3); RemInt(-5,-3); 2 -2 2 -2

`‣ GcdInt` ( m, n ) | ( function ) |

`GcdInt`

returns the greatest common divisor of its two integer operands `m` and `n`, i.e., the greatest integer that divides both `m` and `n`. The greatest common divisor is never negative, even if the arguments are. We define `GcdInt( `

and `m`, 0 ) = GcdInt( 0, `m` ) = AbsInt( `m` )`GcdInt( 0, 0 ) = 0`

.

`GcdInt`

is a method used by the general function `Gcd`

(56.7-1).

gap> GcdInt( 123, 66 ); 3

`‣ Gcdex` ( m, n ) | ( function ) |

returns a record `g`

describing the extended greatest common divisor of `m` and `n`. The component `gcd`

is this gcd, the components `coeff1`

and `coeff2`

are integer cofactors such that `g.gcd = g.coeff1 * `

, and the components `m` + g.coeff2 * `n``coeff3`

and `coeff4`

are integer cofactors such that `0 = g.coeff3 * `

.`m` + g.coeff4 * `n`

If `m` and `n` both are nonzero, `AbsInt( g.coeff1 )`

is less than or equal to `AbsInt(`

, and `n`) / (2 * g.gcd)`AbsInt( g.coeff2 )`

is less than or equal to `AbsInt(`

.`m`) / (2 * g.gcd)

If `m` or `n` or both are zero `coeff3`

is `-`

and `n` / g.gcd`coeff4`

is

.`m` / g.gcd

The coefficients always form a unimodular matrix, i.e., the determinant `g.coeff1 * g.coeff4 - g.coeff3 * g.coeff2`

is \(1\) or \(-1\).

gap> Gcdex( 123, 66 ); rec( coeff1 := 7, coeff2 := -13, coeff3 := -22, coeff4 := 41, gcd := 3 )

This means \(3 = 7 * 123 - 13 * 66\), \(0 = -22 * 123 + 41 * 66\).

gap> Gcdex( 0, -3 ); rec( coeff1 := 0, coeff2 := -1, coeff3 := 1, coeff4 := 0, gcd := 3 ) gap> Gcdex( 0, 0 ); rec( coeff1 := 1, coeff2 := 0, coeff3 := 0, coeff4 := 1, gcd := 0 )

`GcdRepresentation`

(56.7-3) provides similar functionality over arbitrary Euclidean rings.

`‣ LcmInt` ( m, n ) | ( function ) |

returns the least common multiple of the integers `m` and `n`.

`LcmInt`

is a method used by the general operation `Lcm`

(56.7-6).

gap> LcmInt( 123, 66 ); 2706

`‣ CoefficientsQadic` ( i, q ) | ( operation ) |

returns the `q`-adic representation of the integer `i` as a list \(l\) of coefficients satisfying the equality \(\textit{i} = \sum_{{j = 0}} \textit{q}^j \cdot l[j+1]\) for an integer \(\textit{q} > 1\).

gap> l:=CoefficientsQadic(462,3); [ 0, 1, 0, 2, 2, 1 ]

`‣ CoefficientsMultiadic` ( ints, int ) | ( function ) |

returns the multiadic expansion of the integer `int` modulo the integers given in `ints` (in ascending order). It returns a list of coefficients in the *reverse* order to that in `ints`.

`‣ ChineseRem` ( moduli, residues ) | ( function ) |

`ChineseRem`

returns the combination of the `residues` modulo the `moduli`, i.e., the unique integer `c`

from `[0..Lcm(`

such that `moduli`)-1]`c = `

modulo `residues`[i]

for all `moduli`[i]`i`

, if it exists. If no such combination exists `ChineseRem`

signals an error.

Such a combination does exist if and only if

for every pair `residues`[i] = `residues`[k] mod Gcd( `moduli`[i], `moduli`[k] )`i`

, `k`

. Note that this implies that such a combination exists if the moduli are pairwise relatively prime. This is called the Chinese remainder theorem.

gap> ChineseRem( [ 2, 3, 5, 7 ], [ 1, 2, 3, 4 ] ); 53 gap> ChineseRem( [ 6, 10, 14 ], [ 1, 3, 5 ] ); 103

gap> ChineseRem( [ 6, 10, 14 ], [ 1, 2, 3 ] ); Error, the residues must be equal modulo 2 called from <function>( <arguments> ) called from read-eval-loop Entering break read-eval-print loop ... you can 'quit;' to quit to outer loop, or you can 'return;' to continue brk> gap>

`‣ PowerModInt` ( r, e, m ) | ( function ) |

returns \(\textit{r}^{\textit{e}} \pmod{\textit{m}}\) for integers `r`, `e` and `m`.

Note that `PowerModInt`

can reduce intermediate results and thus will generally be faster than using `r``^`

`e`` mod `

`m`, which would compute \(\textit{r}^{\textit{e}}\) first and reduces the result afterwards.

`PowerModInt`

is a method for the general operation `PowerMod`

(56.7-9).

`‣ Primes` | ( global variable ) |

`Primes`

is a strictly sorted list of the 168 primes less than 1000.

This is used in `IsPrimeInt`

(14.4-2) and `FactorsInt`

(14.4-7) to cast out small primes quickly.

gap> Primes[1]; 2 gap> Primes[100]; 541

`‣ IsPrimeInt` ( n ) | ( function ) |

`‣ IsProbablyPrimeInt` ( n ) | ( function ) |

`IsPrimeInt`

returns `false`

if it can prove that the integer `n` is composite and `true`

otherwise. By convention `IsPrimeInt(0) = IsPrimeInt(1) = false`

and we define `IsPrimeInt(-`

`n``) = IsPrimeInt(`

`n``)`

.

`IsPrimeInt`

will return `true`

for every prime `n`. `IsPrimeInt`

will return `false`

for all composite `n` \(< 10^{18}\) and for all composite `n` that have a factor \(p < 1000\). So for integers `n` \(< 10^{18}\), `IsPrimeInt`

is a proper primality test. It is conceivable that `IsPrimeInt`

may return `true`

for some composite `n` \(> 10^{18}\), but no such `n` is currently known. So for integers `n` \(> 10^{18}\), `IsPrimeInt`

is a probable-primality test. `IsPrimeInt`

will issue a warning when its argument is probably prime but not a proven prime. (The function `IsProbablyPrimeInt`

will do a similar calculation but not issue a warning.) The warning can be switched off by `SetInfoLevel( InfoPrimeInt, 0 );`

, the default level is \(1\) (also see `SetInfoLevel`

(7.4-3) ).

If composites that fool `IsPrimeInt`

do exist, they would be extremely rare, and finding one by pure chance might be less likely than finding a bug in **GAP**. We would appreciate being informed about any example of a composite number `n` for which `IsPrimeInt`

returns `true`

.

`IsPrimeInt`

is a deterministic algorithm, i.e., the computations involve no random numbers, and repeated calls will always return the same result. `IsPrimeInt`

first does trial divisions by the primes less than 1000. Then it tests that `n` is a strong pseudoprime w.r.t. the base 2. Finally it tests whether `n` is a Lucas pseudoprime w.r.t. the smallest quadratic nonresidue of `n`. A better description can be found in the comment in the library file `primality.gi`

.

The time taken by `IsPrimeInt`

is approximately proportional to the third power of the number of digits of `n`. Testing numbers with several hundreds digits is quite feasible.

`IsPrimeInt`

is a method for the general operation `IsPrime`

(56.5-8).

Remark: In future versions of **GAP** we hope to change the definition of `IsPrimeInt`

to return `true`

only for proven primes (currently, we lack a sufficiently good primality proving function). In applications, use explicitly `IsPrimeInt`

or `IsProbablyPrimeInt`

with this change in mind.

gap> IsPrimeInt( 2^31 - 1 ); true gap> IsPrimeInt( 10^42 + 1 ); false

`‣ PrimalityProof` ( n ) | ( function ) |

Construct a machine verifiable proof of the primality of (the probable prime) `n`, following the ideas of [BLS75]. The proof consists of various Fermat and Lucas pseudoprimality tests, which taken as a whole prove the primality. The proof is represented as a list of witnesses of two kinds. The first kind, `[ "F", divisor, base ]`

, indicates a successful Fermat pseudoprimality test, where `n` is a strong pseudoprime at `base`

with order not divisible by \((\textit{n}-1)/divisor\). The second kind, `[ "L", divisor, discriminant, P ]`

indicates a successful Lucas pseudoprimality test, for a quadratic form of given `discriminant`

and middle term `P`

with an extra check at \((\textit{n}+1)/divisor\).

`‣ IsPrimePowerInt` ( n ) | ( function ) |

`IsPrimePowerInt`

returns `true`

if the integer `n` is a prime power and `false`

otherwise.

An integer \(n\) is a *prime power* if there exists a prime \(p\) and a positive integer \(i\) such that \(p^i = n\). If \(n\) is negative the condition is that there must exist a negative prime \(p\) and an odd positive integer \(i\) such that \(p^i = n\). The integers 1 and -1 are not prime powers.

Note that `IsPrimePowerInt`

uses `SmallestRootInt`

(14.2-10) and a probable-primality test (see `IsPrimeInt`

(14.4-2)).

gap> IsPrimePowerInt( 31^5 ); true gap> IsPrimePowerInt( 2^31-1 ); # 2^31-1 is actually a prime true gap> IsPrimePowerInt( 2^63-1 ); false gap> Filtered( [-10..10], IsPrimePowerInt ); [ -8, -7, -5, -3, -2, 2, 3, 4, 5, 7, 8, 9 ]

`‣ NextPrimeInt` ( n ) | ( function ) |

`NextPrimeInt`

returns the smallest prime which is strictly larger than the integer `n`.

Note that `NextPrimeInt`

uses a probable-primality test (see `IsPrimeInt`

(14.4-2)).

gap> NextPrimeInt( 541 ); NextPrimeInt( -1 ); 547 2

`‣ PrevPrimeInt` ( n ) | ( function ) |

`PrevPrimeInt`

returns the largest prime which is strictly smaller than the integer `n`.

Note that `PrevPrimeInt`

uses a probable-primality test (see `IsPrimeInt`

(14.4-2)).

gap> PrevPrimeInt( 541 ); PrevPrimeInt( 1 ); 523 -2

`‣ FactorsInt` ( n ) | ( function ) |

`‣ FactorsInt` ( n: RhoTrials := trials ) | ( function ) |

`FactorsInt`

returns a list of factors of a given integer `n` such that `Product( FactorsInt( `

. If \(|n| \leq 1\) the list `n` ) ) = `n``[`

is returned. Otherwise the result contains probable primes, sorted by absolute value. The entries will all be positive except for the first one in case of a negative `n`]`n`.

See `PrimeDivisors`

(14.4-8) for a function that returns a set of (probable) primes dividing `n`.

Note that `FactorsInt`

uses a probable-primality test (see `IsPrimeInt`

(14.4-2)). Thus `FactorsInt`

might return a list which contains composite integers. In such a case you will get a warning about the use of a probable prime. You can switch off these warnings by `SetInfoLevel( InfoPrimeInt, 0 );`

(also see `SetInfoLevel`

(7.4-3)).

The time taken by `FactorsInt`

is approximately proportional to the square root of the second largest prime factor of `n`, which is the last one that `FactorsInt`

has to find, since the largest factor is simply what remains when all others have been removed. Thus the time is roughly bounded by the fourth root of `n`. `FactorsInt`

is guaranteed to find all factors less than \(10^6\) and will find most factors less than \(10^{10}\). If `n` contains multiple factors larger than that `FactorsInt`

may not be able to factor `n` and will then signal an error.

`FactorsInt`

is used in a method for the general operation `Factors`

(56.5-9).

In the second form, `FactorsInt`

calls `FactorsRho`

with a limit of `trials` on the number of trials it performs. The default is 8192. Note that Pollard's Rho is the fastest method for finding prime factors with roughly 5-10 decimal digits, but becomes more and more inferior to other factorization techniques like e.g. the Elliptic Curves Method (ECM) the bigger the prime factors are. Therefore instead of performing a huge number of Rho `trials`, it is usually more advisable to install the **FactInt** package and then simply to use the operation `Factors`

(56.5-9). The factorization of the 8-th Fermat number by Pollard's Rho below takes already a while.

gap> FactorsInt( -Factorial(6) ); [ -2, 2, 2, 2, 3, 3, 5 ] gap> Set( FactorsInt( Factorial(13)/11 ) ); [ 2, 3, 5, 7, 13 ] gap> FactorsInt( 2^63 - 1 ); [ 7, 7, 73, 127, 337, 92737, 649657 ] gap> FactorsInt( 10^42 + 1 ); [ 29, 101, 281, 9901, 226549, 121499449, 4458192223320340849 ] gap> FactorsInt(2^256+1:RhoTrials:=100000000); [ 1238926361552897, 93461639715357977769163558199606896584051237541638188580280321 ]

`‣ PrimeDivisors` ( n ) | ( attribute ) |

`PrimeDivisors`

returns for a non-zero integer `n` a set of its positive (probable) primes divisors. In rare cases the result could contain a composite number which passed certain primality tests, see `IsProbablyPrimeInt`

(14.4-2) and `FactorsInt`

(14.4-7) for more details.

gap> PrimeDivisors(-12); [ 2, 3 ] gap> PrimeDivisors(1); [ ]

`‣ PartialFactorization` ( n[, effort] ) | ( operation ) |

`PartialFactorization`

returns a partial factorization of the integer `n`. No assertions are made about the primality of the factors, except of those mentioned below.

The argument `effort`, if given, specifies how intensively the function should try to determine factors of `n`. The default is `effort` = 5.

If

`effort`= 0, trial division by the primes below 100 is done. Returned factors below \(10^4\) are guaranteed to be prime.If

`effort`= 1, trial division by the primes below 1000 is done. Returned factors below \(10^6\) are guaranteed to be prime.If

`effort`= 2, additionally trial division by the numbers in the lists`Primes2`

and`ProbablePrimes2`

is done, and perfect powers are detected. Returned factors below \(10^6\) are guaranteed to be prime.If

`effort`= 3, additionally`FactorsRho`

(Pollard's Rho) with`RhoTrials`

= 256 is used.If

`effort`= 4, as above, but`RhoTrials`

= 2048.If

`effort`= 5, as above, but`RhoTrials`

= 8192. Returned factors below \(10^{12}\) are guaranteed to be prime, and all prime factors below \(10^6\) are guaranteed to be found.If

`effort`= 6 and the package**FactInt**is loaded, in addition to the above quite a number of special cases are handled.If

`effort`= 7 and the package**FactInt**is loaded, the only thing which is not attempted to obtain a full factorization into Baillie-Pomerance-Selfridge-Wagstaff pseudoprimes is the application of the MPQS to a remaining composite with more than 50 decimal digits.

Increasing the value of the argument `effort` by one usually results in an increase of the runtime requirements by a factor of (very roughly!) 3 to 10. (Also see `CheapFactorsInt`

(EDIM: CheapFactorsInt)).

gap> List([0..5],i->PartialFactorization(97^35-1,i)); [ [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 2446338959059521520901826365168917110105972824229555319002965029 ], [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 967, 2529823122088440042297648774735177983563570655873376751812787 ], [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 967, 2529823122088440042297648774735177983563570655873376751812787 ], [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 967, 39761, 262321, 242549173950325921859769421435653153445616962914227 ], [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 967, 39761, 262321, 687121, 352993394104278463123335513593170858474150787 ], [ 2, 2, 2, 2, 2, 3, 11, 31, 43, 967, 39761, 262321, 687121, 20241187, 504769301, 34549173843451574629911361501 ] ]

`‣ PrintFactorsInt` ( n ) | ( function ) |

prints the prime factorization of the integer `n` in human-readable form. See also `StringPP`

(27.7-9).

gap> PrintFactorsInt( Factorial( 7 ) ); Print( "\n" ); 2^4*3^2*5*7

`‣ PrimePowersInt` ( n ) | ( function ) |

returns the prime factorization of the integer `n` as a list \([ p_1, e_1, \ldots, p_k, e_k ]\) with `n` = \(p_1^{{e_1}} \cdot p_2^{{e_2}} \cdot ... \cdot p_k^{{e_k}}\).

For negative integers, the absolute value is taken. Zero is not allowed as input.

gap> PrimePowersInt( Factorial( 7 ) ); [ 2, 4, 3, 2, 5, 1, 7, 1 ] gap> PrimePowersInt( 1 ); [ ]

`‣ DivisorsInt` ( n ) | ( function ) |

`DivisorsInt`

returns a list of all divisors of the integer `n`. The list is sorted, so that it starts with 1 and ends with `n`. We define that `DivisorsInt( -`

.`n` ) = DivisorsInt( `n` )

Since the set of divisors of 0 is infinite calling `DivisorsInt( 0 )`

causes an error.

`DivisorsInt`

may call `FactorsInt`

(14.4-7) to obtain the prime factors. `Sigma`

(15.5-1) and `Tau`

(15.5-2) compute the sum and the number of positive divisors, respectively.

gap> DivisorsInt( 1 ); DivisorsInt( 20 ); DivisorsInt( 541 ); [ 1 ] [ 1, 2, 4, 5, 10, 20 ] [ 1, 541 ]

`ZmodnZ`

(14.5-2) returns a residue class ring of `Integers`

(14) modulo an ideal. These residue class rings are rings, thus all operations for rings (see Chapter 56) apply. See also Chapters 59 and 15.

`14.5-1 \mod`

`‣ \mod` ( r/s, n ) | ( operation ) |

If `r`, `s` and `n` are integers,

as a reduced fraction is `r` / `s``p/q`

, where `q`

and `n` are coprime, then

is defined to be the product of `r` / `s` mod `n``p`

and the inverse of `q`

modulo `n`. See Section 4.14 for more details and definitions.

With the above definition, `4 / 6 mod 32`

equals `2 / 3 mod 32`

and hence exists (and is equal to 22), despite the fact that 6 has no inverse modulo 32.

`‣ ZmodnZ` ( n ) | ( function ) |

`‣ ZmodpZ` ( p ) | ( function ) |

`‣ ZmodpZNC` ( p ) | ( function ) |

`ZmodnZ`

returns a ring \(R\) isomorphic to the residue class ring of the integers modulo the ideal generated by `n`. The element corresponding to the residue class of the integer \(i\) in this ring can be obtained by `i * One( R )`

, and a representative of the residue class corresponding to the element \(x \in R\) can be computed by `Int`

\(( x )\).

`ZmodnZ( `

is equal to `n` )`Integers mod `

.`n`

`ZmodpZ`

does the same if the argument `p` is a prime integer, additionally the result is a field. `ZmodpZNC`

omits the check whether `p` is a prime.

Each ring returned by these functions contains the whole family of its elements if `n` is not a prime, and is embedded into the family of finite field elements of characteristic `n` if `n` is a prime.

`‣ ZmodnZObj` ( Fam, r ) | ( operation ) |

`‣ ZmodnZObj` ( r, n ) | ( operation ) |

If the first argument is a residue class family `Fam` then `ZmodnZObj`

returns the element in `Fam` whose coset is represented by the integer `r`.

If the two arguments are an integer `r` and a positive integer `n` then `ZmodnZObj`

returns the element in `ZmodnZ( `

(see `n` )`ZmodnZ`

(14.5-2)) whose coset is represented by the integer `r`.

gap> r:= ZmodnZ(15); (Integers mod 15) gap> fam:=ElementsFamily(FamilyObj(r));; gap> a:= ZmodnZObj(fam,9); ZmodnZObj( 9, 15 ) gap> a+a; ZmodnZObj( 3, 15 ) gap> Int(a+a); 3

`‣ IsZmodnZObj` ( obj ) | ( category ) |

`‣ IsZmodnZObjNonprime` ( obj ) | ( category ) |

`‣ IsZmodpZObj` ( obj ) | ( category ) |

`‣ IsZmodpZObjSmall` ( obj ) | ( category ) |

`‣ IsZmodpZObjLarge` ( obj ) | ( category ) |

The elements in the rings \(Z / n Z\) are in the category `IsZmodnZObj`

. If \(n\) is a prime then the elements are of course also in the category `IsFFE`

(59.1-1), otherwise they are in `IsZmodnZObjNonprime`

. `IsZmodpZObj`

is an abbreviation of `IsZmodnZObj and IsFFE`

. This category is the disjoint union of `IsZmodpZObjSmall`

and `IsZmodpZObjLarge`

, the former containing all elements with \(n\) at most `MAXSIZE_GF_INTERNAL`

.

The reasons to distinguish the prime case from the nonprime case are

that objects in

`IsZmodnZObjNonprime`

have an external representation (namely the residue in the range \([ 0, 1, \ldots, n-1 ]\)),that the comparison of elements can be defined as comparison of the residues, and

that the elements lie in a family of type

`IsZmodnZObjNonprimeFamily`

(note that for prime \(n\), the family must be an`IsFFEFamily`

).

The reasons to distinguish the small and the large case are that for small \(n\) the elements must be compatible with the internal representation of finite field elements, whereas we are free to define comparison as comparison of residues for large \(n\).

Note that we *cannot* claim that every finite field element of degree 1 is in `IsZmodnZObj`

, since finite field elements in internal representation may not know that they lie in the prime field.

`‣ CheckDigitISBN` ( n ) | ( function ) |

`‣ CheckDigitISBN13` ( n ) | ( function ) |

`‣ CheckDigitPostalMoneyOrder` ( n ) | ( function ) |

`‣ CheckDigitUPC` ( n ) | ( function ) |

These functions can be used to compute, or check, check digits for some everyday items. In each case what is submitted as input is either the number with check digit (in which case the function returns `true`

or `false`

), or the number without check digit (in which case the function returns the missing check digit). The number can be specified as integer, as string (for example in case of leading zeros) or as a sequence of arguments, each representing a single digit. The check digits tested are the 10-digit ISBN (International Standard Book Number) using `CheckDigitISBN`

(since arithmetic is module 11, a digit 11 is represented by an X); the newer 13-digit ISBN-13 using `CheckDigitISBN13`

; the numbers of 11-digit US postal money orders using `CheckDigitPostalMoneyOrder`

; and the 12-digit UPC bar code found on groceries using `CheckDigitUPC`

.

gap> CheckDigitISBN("052166103"); Check Digit is 'X' 'X' gap> CheckDigitISBN("052166103X"); Checksum test satisfied true gap> CheckDigitISBN(0,5,2,1,6,6,1,0,3,1); Checksum test failed false gap> CheckDigitISBN(0,5,2,1,6,6,1,0,3,'X'); # note single quotes! Checksum test satisfied true gap> CheckDigitISBN13("9781420094527"); Checksum test satisfied true gap> CheckDigitUPC("07164183001"); Check Digit is 1 1 gap> CheckDigitPostalMoneyOrder(16786457155); Checksum test satisfied true

`‣ CheckDigitTestFunction` ( l, m, f ) | ( function ) |

This function creates check digit test functions such as `CheckDigitISBN`

(14.6-1) for check digit schemes that use the inner products with a fixed vector modulo a number. The scheme creates will use strings of `l` digits (including the check digits), the check consists of taking the standard product of the vector of digits with the fixed vector `f` modulo `m`; the result needs to be 0. The function returns a function that then can be used for testing or determining check digits.

gap> isbntest:=CheckDigitTestFunction(10,11,[1,2,3,4,5,6,7,8,9,-1]); function( arg... ) ... end gap> isbntest("038794680"); Check Digit is 2 2

**GAP** provides `Random`

(30.7-1) methods for many collections of objects. On a lower level these methods use *random sources* which provide random integers and random choices from lists.

See `IsRandomSource`

(14.7-1) for the user interface for random sources, and Section 14.7-6 for information about developing new kinds of random sources.

`‣ IsRandomSource` ( obj ) | ( category ) |

This is the category of random source objects. The *user interface* for these objects consists of the following functions.

`RandomSource`

(14.7-5) creates a new random source `rs`, say.

`Random( `

yields a random element of the list `rs`, `list` )`list`, and `Random( `

yields a random integer between `rs`, `low`, `high` )`low` and `high` (inclusive), see `Random`

(14.7-2).

If `rs` supports resetting (see `State`

(14.7-3)) then `State( `

yields a copy `rs` )`state`, say, of the current state of `rs` such that `Reset( `

resets `rs`, `state` )`rs` to the given state.

One idea behind providing several independent (pseudo) random sources is to make algorithms which use some sort of random choices deterministic. They can use their own new random source created with a fixed seed and so do exactly the same in different calls.

Random source objects lie in the family `RandomSourcesFamily`

.

`‣ Random` ( rs, list ) | ( operation ) |

`‣ Random` ( rs, coll ) | ( operation ) |

`‣ Random` ( rs, low, high ) | ( operation ) |

This operation returns a random element from the dense, nonempty list `list` or the nonempty collection `coll`, or an integer in the range from the given (possibly large) integers `low` to `high`, respectively.

The choice should only depend on the random source `rs` and have no effect on other random sources.

It is not defined what happens if `list` or `coll` is empty, `list` is not dense, or `low` is larger than `high`.

gap> mysource := RandomSource(IsMersenneTwister, 42);; gap> Random(mysource, 1, 10^60); 999331861769949319194941485000557997842686717712198687315183

`‣ State` ( rs ) | ( operation ) |

`‣ Reset` ( rs[, seed] ) | ( operation ) |

These are the basic operations for random sources (see `IsRandomSource`

(14.7-1)).

`State`

returns a data structure which admits recovering the state of the random source such that a sequence of random calls using this random source can be reproduced. If a random source cannot be reset (say, it uses truly random physical data) then `State`

returns `fail`

.

`Reset( `

resets the random source `rs`, `seed` )`rs` to a state described by `seed`, if the random source can be reset; otherwise it does nothing. Here `seed` can be an output of `State`

and then `rs` gets reset to that state. For historical reasons, random sources accept integer values as `seed`. We recommend that new code should not rely on this; always use the output of a prior call to `State`

as `seed`, or omit it. Without the `seed` argument a fixed default seed is used. `Reset`

returns the state of `rs` before the call.

Most methods for `Random`

(30.7-1) in the **GAP** library that do not take a random source as argument use the `GlobalMersenneTwister`

(14.7-4) as random source. It can be reset into a known state as in the following example.

gap> seed := Reset(GlobalMersenneTwister);; gap> seed = State(GlobalMersenneTwister); true gap> List([1..10],i->Random(Integers)); [ -3, 2, -1, -2, -1, -1, 1, -4, 1, 0 ] gap> List([1..10],i->Random(Integers)); [ -1, -1, -1, 1, -1, 1, -2, -1, -2, 0 ] gap> Reset(GlobalMersenneTwister, seed);; gap> List([1..10],i->Random(Integers)); [ -3, 2, -1, -2, -1, -1, 1, -4, 1, 0 ]

`‣ IsMersenneTwister` ( rs ) | ( category ) |

`‣ IsGAPRandomSource` ( rs ) | ( category ) |

`‣ IsGlobalRandomSource` ( rs ) | ( category ) |

`‣ GlobalMersenneTwister` | ( global variable ) |

`‣ GlobalRandomSource` | ( global variable ) |

Currently, the **GAP** library provides three types of random sources, distinguished by the three listed categories.

`IsMersenneTwister`

are random sources which use a fast random generator of 32 bit numbers, called the Mersenne twister. The pseudo random sequence has a period of \(2^{19937}-1\) and the numbers have a \(623\)-dimensional equidistribution. For more details and the origin of the code used in the **GAP** kernel, see: http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html.

Use the Mersenne twister if possible, in particular for generating many large random integers.

There is also a predefined global random source `GlobalMersenneTwister`

which is used as the default random source by those library methods for `Random`

(30.7-1) that do not take a random source as an argument.

`IsGAPRandomSource`

uses the same number generator as `IsGlobalRandomSource`

, but you can create several of these random sources which generate their random numbers independently of all other random sources.

`IsGlobalRandomSource`

gives access to the *classical* global random generator which was used by **GAP** in former releases. You do not need to construct new random sources of this kind which would all use the same global data structure. Just use the existing random source `GlobalRandomSource`

. This uses the additive random number generator described in [Knu98] (Algorithm A in 3.2.2 with lag \(30\)).

Other kinds of random sources are implemented by **GAP** packages.

`‣ RandomSource` ( cat[, seed] ) | ( operation ) |

This operation is used to create new random sources. The first argument `cat` is the category describing the type of the random generator, for example one of the categories listed in Section 14.7-4.

An optional `seed` can be given to specify the initial state. For details, see Section 14.7-3.

gap> rs1 := RandomSource(IsMersenneTwister); <RandomSource in IsMersenneTwister> gap> l1 := List([1..10000], i-> Random(rs1, [1..6]));; gap> state1 := State(rs1);; gap> rs2 := RandomSource(IsMersenneTwister);; gap> l2 := List([1..10000], i-> Random(rs2, [1..6]));; gap> l1 = l2; true gap> l3 := List([1..10000], i-> Random(rs1, [1..6]));; gap> l1 = l3; false gap> rs3 := RandomSource(IsMersenneTwister, state1);; gap> l4 := List([1..10000], i-> Random(rs3, [1..6]));; gap> l3 = l4; true gap> n := Random(rs1, 1, 2^220); 1077726777923092117987668044202944212469136000816111066409337432400

If one wants to implement a new kind of random sources then the first step is the declaration of a new category `C`

, say, that implies `IsRandomSource`

(14.7-1), analogous to the categories listed in Section 14.7-4, as follows.

`DeclareCategory( "C", IsRandomSource );`

.

Then the following method installations are needed.

`InstallMethod( Init, [ C, IsObject ], function( prers, seed ) ... end );`

Here `prers`

is an empty component object (which has already the filter `C`

), and `seed`

is an integer or a state value as returned by `State`

(14.7-3) that describes the initial state of the random source. The function should fill in the actual data and then return the (now initialized) object `prers`

. The default used for `seed`

is the integer `1`

. A given state value need not be copied by the function.

`InstallMethod( Random, [ C, IsInt, IsInt ], function( rs, low, high ) ... end );`

Here `rs`

is an already initialized random source object in the category `C`

, and the function returns an integer between `low`

and `high`

(inclusive). It is not defined what happens when `low`

is larger than `high`

.

`InstallMethod( State, [ C ], function( rs ) ... end );`

If `rs`

supports resetting then the function must return an object that describes the current state of `rs`

. This object must be an independent copy, that is, calling `Random`

(14.7-2) for `rs`

must not change the object that was returned by `State`

(14.7-3); otherwise `ReturnFail`

(5.4-3) should be installed.

`InstallMethod( Reset, [ C, IsObject ], function( rs, seed ) ... end );`

If `rs`

supports resetting then the function must reinitialize `rs`

to the integer or `State`

(14.7-3) value `seed`

and must return the `State`

(14.7-3) value of `rs`

before these changes; if resetting is not supported then `ReturnNothing`

(5.4-4) should be installed. Reset need not copy a given state. Note that the generic unary `Reset`

(14.7-3) method uses the default seed `1`

.

Examples of implementations as described here are given by the random sources with defining filter `IsMersenneTwister`

(14.7-4) or `IsRealRandomSource`

. (For the latter, see `RandomSource`

(IO: RandomSource) in the **GAP** package **IO**.)

Bitfields are a low-level feature intended to support efficient subdivision of immediate integers into bitfields of various widths. This is typically useful in implementing space-efficient and/or cache-efficient data structures. This feature should be used with care because (*inter alia*) it has different limitations on 32-bit and 64-bit architectures.

`‣ MakeBitfields` ( width.... ) | ( function ) |

This function sets up the machinery for a set of bitfields of the given widths. All bitfield values are treated as unsigned. The total of the widths must not exceed 60 bits on 64-bit architecture or 28 bits on a 32-bit architecture. For performance reasons some checks that one might wish to do are omitted. In particular, the builder and setter functions do not check if the value[s] passed to them are negative or too large (unless **GAP** is specially compiled for debugging). Behaviour when such arguments are passed is undefined. You can tell which type of architecture you are running on by accessing `GAPInfo.BytesPerVariable`

which is 8 on 64-bits and 4 on 32. The return value when \(n\) widths are given is a record whose fields are

`widths`

a copy of the arguments, for convenience,

`getters`

a list of \(n\) functions of one argument each of which extracts one of the fields from an immediate integer

`setters`

a list of \(n\) functions each taking two arguments: a packed value and a new value for one of its fields and returning a new packed value. The \(i\)th function returned the new packed value in which the \(i\)th field has been replaced by the new value. Note that this does NOT modify the original packed value.

Two additional fields may be present if any of the field widths is one. Each is a list and only has entried bound in the positions corresponding to the width 1 fields.

`booleanGetters`

if the \(i\)th position of this list is set, it contains a function which extracts the \(i\)th field (which will have width one) and returns

`true`

if it contains 1 and`false`

if it contains 0`booleanSetters`

if the \(i\)th position of this list is set, it contains a function of two arguments. The first argument is a packed value, the second is

`true`

or`false`

. It returns a new packed value in which the \(i\)th field is set to 1 if the second argument was`true`

and 0 if it was`false`

. Behaviour for any other value is undefined.

`‣ BuildBitfields` ( widths, vals... ) | ( function ) |

This function takes one or more argument. Its first argument is a list of field widths, as found in the `widths`

entry of a record returned by `MakeBitfields`

. The remaining arguments are unsigned integer values, equal in number to the entries of the list of field widths. It returns a small integer in which those entries are packed into bitfields of the given widths. The first entry occupies the least significant bits. DeclareGlobalFunction("BuildBitfields");

gap> bf := MakeBitfields(1,2,3); rec( booleanGetters := [ function( data ) ... end ], booleanSetters := [ function( data, val ) ... end ], getters := [ function( data ) ... end, function( data ) ... end, function( data ) ... end ], setters := [ function( data, val ) ... end, function( data, val ) ... end, function( data, val ) ... end ], widths := [ 1, 2, 3 ] ) gap> x := BuildBitfields(bf.widths,0,3,5); 46 gap> bf.getters[3](x); 5 gap> y := bf.setters[1](x,1); 47 gap> x; 46 gap> bf.booleanGetters[1](x); false gap> bf.booleanGetters[1](y); true

Goto Chapter: Top 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 Bib Ind

generated by GAPDoc2HTML