fmpz_factor.h – integer factorisation

Types, macros and constants

type fmpz_factor_struct
type fmpz_factor_t

Factoring integers

An integer may be represented in factored form using the fmpz_factor_t data structure. This consists of two fmpz vectors representing bases and exponents, respectively. Canonically, the bases will be prime numbers sorted in ascending order and the exponents will be positive. A separate int field holds the sign, which may be \(-1\), \(0\) or \(1\).

void fmpz_factor_init(fmpz_factor_t factor)

Initialises an fmpz_factor_t structure.

void fmpz_factor_clear(fmpz_factor_t factor)

Clears an fmpz_factor_t structure.

void _fmpz_factor_append_ui(fmpz_factor_t factor, mp_limb_t p, ulong exp)

Append a factor \(p\) to the given exponent to the fmpz_factor_t structure factor.

void _fmpz_factor_append(fmpz_factor_t factor, const fmpz_t p, ulong exp)

Append a factor \(p\) to the given exponent to the fmpz_factor_t structure factor.

void fmpz_factor(fmpz_factor_t factor, const fmpz_t n)

Factors \(n\) into prime numbers. If \(n\) is zero or negative, the sign field of the factor object will be set accordingly.

int fmpz_factor_smooth(fmpz_factor_t factor, const fmpz_t n, slong bits, int proved)

Factors \(n\) into prime numbers up to approximately the given number of bits and possibly one additional cofactor, which may or may not be prime.

If the number is definitely factored fully, the return value is \(1\), otherwise the final factor (which may have exponent greater than \(1\)) is composite and needs to be factored further.

If the number has a factor of around the given number of bits, there is at least a two-thirds chance of finding it. Smaller factors should be found with a much higher probability.

The amount of time spent factoring can be controlled by lowering or increasing bits. However, the quadratic sieve may be faster if bits is set to more than one third of the number of bits of \(n\).

The function uses trial factoring up to bits = 15, followed by a primality test and a perfect power test to check if the factorisation is complete. If bits is at least 16, it proceeds to use the elliptic curve method to look for larger factors.

The behavior of primality testing is determined by the proved parameter:

If proved is set to \(1\) the function will prove all factors prime (other than the last factor, if the return value is \(0\)).

If proved is set to \(0\), the function will only check that factors are probable primes.

If proved is set to \(-1\), the function will not test primality after performing trial division. A perfect power test is still performed.

As an exception to the rules stated above, this function will call n_factor internally if \(n\) or the remainder after trial division is smaller than one word, guaranteeing a complete factorisation.

void fmpz_factor_si(fmpz_factor_t factor, slong n)

Like fmpz_factor, but takes a machine integer \(n\) as input.

int fmpz_factor_trial_range(fmpz_factor_t factor, const fmpz_t n, ulong start, ulong num_primes)

Factors \(n\) into prime factors using trial division. If \(n\) is zero or negative, the sign field of the factor object will be set accordingly.

The algorithm starts with the given start index in the flint_primes table and uses at most num_primes primes from that point.

The function returns 1 if \(n\) is completely factored, otherwise it returns \(0\).

int fmpz_factor_trial(fmpz_factor_t factor, const fmpz_t n, slong num_primes)

Factors \(n\) into prime factors using trial division. If \(n\) is zero or negative, the sign field of the factor object will be set accordingly.

The algorithm uses the given number of primes, which must be in the range \([0, 3512]\). An exception is raised if a number outside this range is passed.

The function returns 1 if \(n\) is completely factored, otherwise it returns \(0\).

The final entry in the factor struct is set to the cofactor after removing prime factors, if this is not \(1\).

void fmpz_factor_refine(fmpz_factor_t res, const fmpz_factor_t f)

Attempts to improve a partial factorization of an integer by “refining” the factorization f to a more complete factorization res whose bases are pairwise relatively prime.

This function does not require its input to be in canonical form, nor does it guarantee that the resulting factorization will be canonical.

void fmpz_factor_expand_iterative(fmpz_t n, const fmpz_factor_t factor)

Evaluates an integer in factored form back to an fmpz_t.

This currently exponentiates the bases separately and multiplies them together one by one, although much more efficient algorithms exist.

int fmpz_factor_pp1(fmpz_t factor, const fmpz_t n, ulong B1, ulong B2_sqrt, ulong c)

Use Williams’ \(p + 1\) method to factor \(n\), using a prime bound in stage 1 of B1 and a prime limit in stage 2 of at least the square of B2_sqrt. If a factor is found, the function returns \(1\) and factor is set to the factor that is found. Otherwise, the function returns \(0\).

The value \(c\) should be a random value greater than \(2\). Successive calls to the function with different values of \(c\) give additional chances to factor \(n\) with roughly exponentially decaying probability of finding a factor which has been missed (if \(p+1\) or \(p-1\) is not smooth for any prime factors \(p\) of \(n\) then the function will not ever succeed).

int fmpz_factor_pollard_brent_single(fmpz_t p_factor, fmpz_t n_in, fmpz_t yi, fmpz_t ai, mp_limb_t max_iters)

Pollard Rho algorithm for integer factorization. Assumes that the \(n\) is not prime. factor is set as the factor if found. Takes as input the initial value \(y\), to start polynomial evaluation, and \(a\), the constant of the polynomial used. It is not assured that the factor found will be prime. Does not compute the complete factorization, just one factor. Returns the number of limbs of factor if factorization is successful (non trivial factor is found), else returns 0.

max_iters is the number of iterations tried in process of finding the cycle. If the algorithm fails to find a non trivial factor in one call, it tries again (this time with a different set of random values).

int fmpz_factor_pollard_brent(fmpz_t factor, flint_rand_t state, fmpz_t n, mp_limb_t max_tries, mp_limb_t max_iters)

Pollard Rho algorithm for integer factorization. Assumes that the \(n\) is not prime. factor is set as the factor if found. It is not assured that the factor found will be prime. Does not compute the complete factorization, just one factor. Returns the number of limbs of factor if factorization is successful (non trivial factor is found), else returns 0.

max_iters is the number of iterations tried in process of finding the cycle. If the algorithm fails to find a non trivial factor in one call, it tries again (this time with a different set of random values). This process is repeated a maximum of max_tries times.

The algorithm used is a modification of the original Pollard Rho algorithm, suggested by Richard Brent. It can be found in the paper available at https://maths-people.anu.edu.au/~brent/pd/rpb051i.pdf

Input and output

int fmpz_factor_fprint(FILE *fs, const fmpz_factor_t factor)
int fmpz_factor_print(const fmpz_factor_t factor)

Prints the factorization factor into fs or stdout. If factor is zero, it prints 0. Else, it prints the factorization as f_{1}^e_{1} * ... * f_{n}^e_{n}, where f_{i} and e_{i} are the \(i\)-th factor and exponent, where ^e_{i} is omitted if \(e_{i} = 1\). In particular, if factor is \(1\) or \(-1\), it prints 1 or -1, respectively.

Returns the number of characters written to file stream.

Elliptic curve (ECM) method

Factoring of fmpz integers using ECM

void fmpz_factor_ecm_init(ecm_t ecm_inf, mp_limb_t sz)

Initializes the ecm_t struct. This is needed in some functions and carries data between subsequent calls.

void fmpz_factor_ecm_clear(ecm_t ecm_inf)

Clears the ecm_t struct.

void fmpz_factor_ecm_addmod(mp_ptr a, mp_ptr b, mp_ptr c, mp_ptr n, mp_limb_t n_size)

Sets \(a\) to \((b + c)\) % \(n\). This is not a normal add mod function, it assumes \(n\) is normalized (highest bit set) and \(b\) and \(c\) are reduced modulo \(n\).

Used for arithmetic operations in fmpz_factor_ecm.

void fmpz_factor_ecm_submod(mp_ptr x, mp_ptr a, mp_ptr b, mp_ptr n, mp_limb_t n_size)

Sets \(x\) to \((a - b)\) % \(n\). This is not a normal subtract mod function, it assumes \(n\) is normalized (highest bit set) and \(b\) and \(c\) are reduced modulo \(n\).

Used for arithmetic operations in fmpz_factor_ecm.

void fmpz_factor_ecm_double(mp_ptr x, mp_ptr z, mp_ptr x0, mp_ptr z0, mp_ptr n, ecm_t ecm_inf)

Sets the point \((x : z)\) to two times \((x_0 : z_0)\) modulo \(n\) according to the formula

\[x = (x_0 + z_0)^2 \cdot (x_0 - z_0)^2 \mod n,\]
\[z = 4 x_0 z_0 \left((x_0 - z_0)^2 + 4a_{24}x_0z_0\right) \mod n.\]

ecm_inf is used just to use temporary mp_ptr’s in the structure. This group doubling is valid only for points expressed in Montgomery projective coordinates.

void fmpz_factor_ecm_add(mp_ptr x, mp_ptr z, mp_ptr x1, mp_ptr z1, mp_ptr x2, mp_ptr z2, mp_ptr x0, mp_ptr z0, mp_ptr n, ecm_t ecm_inf)

Sets the point \((x : z)\) to the sum of \((x_1 : z_1)\) and \((x_2 : z_2)\) modulo \(n\), given the difference \((x_0 : z_0)\) according to the formula

\[\begin{split}x = 4z_0(x_1x_2 - z_1z_2)^2 \mod n, \\ z = 4x_0(x_2z_1 - x_1z_2)^2 \mod n.\end{split}\]

ecm_inf is used just to use temporary mp_ptr’s in the structure. This group addition is valid only for points expressed in Montgomery projective coordinates.

void fmpz_factor_ecm_mul_montgomery_ladder(mp_ptr x, mp_ptr z, mp_ptr x0, mp_ptr z0, mp_limb_t k, mp_ptr n, ecm_t ecm_inf)

Montgomery ladder algorithm for scalar multiplication of elliptic points.

Sets the point \((x : z)\) to \(k(x_0 : z_0)\) modulo \(n\).

ecm_inf is used just to use temporary mp_ptr’s in the structure. Valid only for points expressed in Montgomery projective coordinates.

int fmpz_factor_ecm_select_curve(mp_ptr f, mp_ptr sigma, mp_ptr n, ecm_t ecm_inf)

Selects a random elliptic curve given a random integer sigma, according to Suyama’s parameterization. If the factor is found while selecting the curve, the number of limbs required to store the factor is returned, otherwise \(0\).

It could be possible that the selected curve is unsuitable for further computations, in such a case, \(-1\) is returned.

Also selects the initial point \(x_0\), and the value of \((a + 2)/4\), where \(a\) is a curve parameter. Sets \(z_0\) as \(1\). All these are stored in the ecm_t struct.

The curve selected is of Montgomery form, the points selected satisfy the curve and are projective coordinates.

int fmpz_factor_ecm_stage_I(mp_ptr f, const mp_limb_t *prime_array, mp_limb_t num, mp_limb_t B1, mp_ptr n, ecm_t ecm_inf)

Stage I implementation of the ECM algorithm.

f is set as the factor if found. num is number of prime numbers \(\le\) the bound B1. prime_array is an array of first B1 primes. \(n\) is the number being factored.

If the factor is found, number of words required to store the factor is returned, otherwise \(0\).

int fmpz_factor_ecm_stage_II(mp_ptr f, mp_limb_t B1, mp_limb_t B2, mp_limb_t P, mp_ptr n, ecm_t ecm_inf)

Stage II implementation of the ECM algorithm.

f is set as the factor if found. B1, B2 are the two bounds. P is the primorial (approximately equal to \(\sqrt{B2}\)). \(n\) is the number being factored.

If the factor is found, number of words required to store the factor is returned, otherwise \(0\).

int fmpz_factor_ecm(fmpz_t f, mp_limb_t curves, mp_limb_t B1, mp_limb_t B2, flint_rand_t state, const fmpz_t n_in)

Outer wrapper function for the ECM algorithm. In case f can fit in a single unsigned word, a call to n_factor_ecm is made.

The function calls stage I and II, and the precomputations (builds prime_array for stage I, GCD_table and prime_table for stage II).

f is set as the factor if found. curves is the number of random curves being tried. B1, B2 are the two bounds or stage I and stage II. \(n\) is the number being factored.

If a factor is found in stage I, \(1\) is returned. If a factor is found in stage II, \(2\) is returned. If a factor is found while selecting the curve, \(-1\) is returned. Otherwise \(0\) is returned.