misc.py

# -*- coding: utf-8 -*-
r"""
Miscellaneous arithmetic functions
"""

# ****************************************************************************
#       Copyright (C) 2006 William Stein <wstein@gmail.com>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#                  https://www.gnu.org/licenses/
# ****************************************************************************

import math
from collections.abc import Iterable

from sage.misc.misc import powerset
from sage.misc.misc_c import prod

from sage.libs.pari.all import pari
from sage.libs.flint.arith import (bernoulli_number as flint_bernoulli,
                                   dedekind_sum as flint_dedekind_sum)

from sage.structure.element import parent
from sage.structure.coerce import py_scalar_to_element

from sage.rings.rational_field import QQ
from sage.rings.integer_ring import ZZ
from sage.rings.integer import Integer, GCD_list
from sage.rings.rational import Rational
from sage.rings.real_mpfr import RealNumber
from sage.rings.complex_mpfr import ComplexNumber

from sage.rings.fast_arith import arith_int, arith_llong, prime_range


##################################################################
# Elementary Arithmetic
##################################################################


def algdep(z, degree, known_bits=None, use_bits=None, known_digits=None,
           use_digits=None, height_bound=None, proof=False):
    """
    Return an irreducible polynomial of degree at most `degree` which
    is approximately satisfied by the number `z`.

    You can specify the number of known bits or digits of `z` with
    ``known_bits=k`` or ``known_digits=k``. PARI is then told to
    compute the result using `0.8k` of these bits/digits. Or, you can
    specify the precision to use directly with ``use_bits=k`` or
    ``use_digits=k``. If none of these are specified, then the precision
    is taken from the input value.

    A height bound may be specified to indicate the maximum coefficient
    size of the returned polynomial; if a sufficiently small polynomial
    is not found, then ``None`` will be returned. If ``proof=True`` then
    the result is returned only if it can be proved correct (i.e. the
    only possible minimal polynomial satisfying the height bound, or no
    such polynomial exists). Otherwise a ``ValueError`` is raised
    indicating that higher precision is required.

    ALGORITHM: Uses LLL for real/complex inputs, PARI C-library
    ``algdep`` command otherwise.

    Note that ``algebraic_dependency`` is a synonym for ``algdep``.

    INPUT:


    -  ``z`` - real, complex, or `p`-adic number

    -  ``degree`` - an integer

    -  ``height_bound`` - an integer (default: ``None``) specifying the maximum
                          coefficient size for the returned polynomial

    -  ``proof`` - a boolean (default: ``False``), requires height_bound to be set


    EXAMPLES::

        sage: algdep(1.888888888888888, 1)
        9*x - 17
        sage: algdep(0.12121212121212,1)
        33*x - 4
        sage: algdep(sqrt(2),2)
        x^2 - 2

    This example involves a complex number::

        sage: z = (1/2)*(1 + RDF(sqrt(3)) *CC.0); z
        0.500000000000000 + 0.866025403784439*I
        sage: algdep(z, 6)
        x^2 - x + 1

    This example involves a `p`-adic number::

        sage: K = Qp(3, print_mode = 'series')
        sage: a = K(7/19); a
        1 + 2*3 + 3^2 + 3^3 + 2*3^4 + 2*3^5 + 3^8 + 2*3^9 + 3^11 + 3^12 + 2*3^15 + 2*3^16 + 3^17 + 2*3^19 + O(3^20)
        sage: algdep(a, 1)
        19*x - 7

    These examples show the importance of proper precision control. We
    compute a 200-bit approximation to `sqrt(2)` which is wrong in the
    33'rd bit::

        sage: z = sqrt(RealField(200)(2)) + (1/2)^33
        sage: p = algdep(z, 4); p
        227004321085*x^4 - 216947902586*x^3 - 99411220986*x^2 + 82234881648*x - 211871195088
        sage: factor(p)
        227004321085*x^4 - 216947902586*x^3 - 99411220986*x^2 + 82234881648*x - 211871195088
        sage: algdep(z, 4, known_bits=32)
        x^2 - 2
        sage: algdep(z, 4, known_digits=10)
        x^2 - 2
        sage: algdep(z, 4, use_bits=25)
        x^2 - 2
        sage: algdep(z, 4, use_digits=8)
        x^2 - 2

    Using the ``height_bound`` and ``proof`` parameters, we can see that
    `pi` is not the root of an integer polynomial of degree at most 5
    and coefficients bounded above by 10::

        sage: algdep(pi.n(), 5, height_bound=10, proof=True) is None
        True

    For stronger results, we need more precision::

        sage: algdep(pi.n(), 5, height_bound=100, proof=True) is None
        Traceback (most recent call last):
        ...
        ValueError: insufficient precision for non-existence proof
        sage: algdep(pi.n(200), 5, height_bound=100, proof=True) is None
        True

        sage: algdep(pi.n(), 10, height_bound=10, proof=True) is None
        Traceback (most recent call last):
        ...
        ValueError: insufficient precision for non-existence proof
        sage: algdep(pi.n(200), 10, height_bound=10, proof=True) is None
        True

    We can also use ``proof=True`` to get positive results::

        sage: a = sqrt(2) + sqrt(3) + sqrt(5)
        sage: algdep(a.n(), 8, height_bound=1000, proof=True)
        Traceback (most recent call last):
        ...
        ValueError: insufficient precision for uniqueness proof
        sage: f = algdep(a.n(1000), 8, height_bound=1000, proof=True); f
        x^8 - 40*x^6 + 352*x^4 - 960*x^2 + 576
        sage: f(a).expand()
        0

    TESTS::

        sage: algdep(complex("1+2j"), 4)
        x^2 - 2*x + 5

    We get an irreducible polynomial even if PARI returns a reducible
    one::

        sage: z = CDF(1, RR(3).sqrt())/2
        sage: pari(z).algdep(5)
        x^5 + x^2
        sage: algdep(z, 5)
        x^2 - x + 1

    Check that cases where a constant polynomial might look better
    get handled correctly::

        sage: z=CC(-1)**(1/3)
        sage: algdep(z,1)
        x

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8, float64
        sage: algdep(float64(1.888888888888888), int8(1))
        9*x - 17
        sage: from gmpy2 import mpz, mpfr
        sage: algdep(mpfr(1.888888888888888), mpz(1))
        9*x - 17
    """
    if proof and not height_bound:
        raise ValueError("height_bound must be given for proof=True")

    R = ZZ['x']
    x = R.gen()

    z = py_scalar_to_element(z)

    if isinstance(z, Integer):
        if height_bound and abs(z) >= height_bound:
            return None
        return x - z

    degree = ZZ(degree)

    if isinstance(z, Rational):
        if height_bound and max(abs(z.denominator()), abs(z.numerator())) >= height_bound:
            return None
        return z.denominator()*x - z.numerator()

    if isinstance(z, (RealNumber, ComplexNumber)):

        log2_10 = math.log(10,2)

        prec = z.prec() - 6
        if known_digits is not None:
            known_bits = known_digits * log2_10
        if known_bits is not None:
            use_bits = known_bits * 0.8
        if use_digits is not None:
            use_bits = use_digits * log2_10
        if use_bits is not None:
            prec = int(use_bits)

        is_complex = isinstance(z, ComplexNumber)
        n = degree+1
        from sage.matrix.all import matrix
        M = matrix(ZZ, n, n+1+int(is_complex))
        r = ZZ.one() << prec
        M[0, 0] = 1
        M[0, -1] = r
        for k in range(1, degree + 1):
            M[k, k] = 1
            r *= z
            if is_complex:
                M[k, -1] = r.real().round()
                M[k, -2] = r.imag().round()
            else:
                M[k, -1] = r.round()
        LLL = M.LLL(delta=.75)
        coeffs = LLL[0][:n]
        #we're supposed to find an irreducible polynomial, so we cannot
        #return a constant one. If the first LLL basis vector gives
        #a constant polynomial, use the next one.
        if all(c == 0 for c in coeffs[1:]):
            coeffs = LLL[1][:n]

        if height_bound:
            def norm(v):
                # norm on an integer vector invokes Integer.sqrt() which tries to factor...
                from sage.rings.real_mpfi import RIF
                return v.change_ring(RIF).norm()
            if max(abs(a) for a in coeffs) > height_bound:
                if proof:
                    # Given an LLL reduced basis $b_1, ..., b_n$, we only
                    # know that $|b_1| <= 2^((n-1)/2) |x|$ for non-zero $x \in L$.
                    if norm(LLL[0]) <= 2**((n-1)/2) * n.sqrt() * height_bound:
                        raise ValueError("insufficient precision for non-existence proof")
                return None
            elif proof and norm(LLL[1]) < 2**((n-1)/2) * max(norm(LLL[0]), n.sqrt()*height_bound):
                raise ValueError("insufficient precision for uniqueness proof")
        if coeffs[degree] < 0:
            coeffs = -coeffs
        f = list(coeffs)

    elif proof or height_bound:
        raise NotImplementedError("proof and height bound only implemented for real and complex numbers")

    else:
        y = pari(z)
        f = y.algdep(degree)

    # f might be reducible. Find the best fitting irreducible factor
    factors = [p for p, e in R(f).factor()]
    return min(factors, key=lambda f: abs(f(z)))


algebraic_dependency = algdep


def bernoulli(n, algorithm='default', num_threads=1):
    r"""
    Return the n-th Bernoulli number, as a rational number.

    INPUT:

    - ``n`` - an integer
    - ``algorithm``:

      - ``'default'`` -- use 'flint' for n <= 20000, then 'arb' for n <= 300000
        and 'bernmm' for larger values (this is just a heuristic, and not guaranteed
        to be optimal on all hardware)
      - ``'arb'`` -- use the arb library
      - ``'flint'`` -- use the FLINT library
      - ``'pari'`` -- use the PARI C library
      - ``'gap'`` -- use GAP
      - ``'gp'`` -- use PARI/GP interpreter
      - ``'magma'`` -- use MAGMA (optional)
      - ``'bernmm'`` -- use bernmm package (a multimodular algorithm)

    - ``num_threads`` - positive integer, number of
      threads to use (only used for bernmm algorithm)

    EXAMPLES::

        sage: bernoulli(12)
        -691/2730
        sage: bernoulli(50)
        495057205241079648212477525/66

    We demonstrate each of the alternative algorithms::

        sage: bernoulli(12, algorithm='arb')
        -691/2730
        sage: bernoulli(12, algorithm='flint')
        -691/2730
        sage: bernoulli(12, algorithm='gap')
        -691/2730
        sage: bernoulli(12, algorithm='gp')
        -691/2730
        sage: bernoulli(12, algorithm='magma')           # optional - magma
        -691/2730
        sage: bernoulli(12, algorithm='pari')
        -691/2730
        sage: bernoulli(12, algorithm='bernmm')
        -691/2730
        sage: bernoulli(12, algorithm='bernmm', num_threads=4)
        -691/2730

    TESTS::

        sage: algs = ['arb','gap','gp','pari','bernmm','flint']
        sage: test_list = [ZZ.random_element(2, 2255) for _ in range(500)]
        sage: vals = [[bernoulli(i,algorithm = j) for j in algs] for i in test_list]  # long time (up to 21s on sage.math, 2011)
        sage: union([len(union(x))==1 for x in vals])  # long time (depends on previous line)
        [True]
        sage: algs = ['gp','pari','bernmm']
        sage: test_list = [ZZ.random_element(2256, 5000) for _ in range(500)]
        sage: vals = [[bernoulli(i,algorithm = j) for j in algs] for i in test_list]  # long time (up to 30s on sage.math, 2011)
        sage: union([len(union(x))==1 for x in vals])  # long time (depends on previous line)
        [True]
        sage: from numpy import int8
        sage: bernoulli(int8(12))
        -691/2730
        sage: from gmpy2 import mpz
        sage: bernoulli(mpz(12))
        -691/2730

    AUTHOR:

    - David Joyner and William Stein
    """
    n = ZZ(n)

    if algorithm == 'default':
        if n <= 20000:
            algorithm = 'flint'
        elif n <= 300000:
            algorithm = 'arb'
        else:
            algorithm = 'bernmm'

    if algorithm == 'arb':
        import sage.libs.arb.arith as arb_arith
        return arb_arith.bernoulli(n)
    elif algorithm == 'flint':
        if n >= 100000:
            from warnings import warn
            warn("flint is known to not be accurate for large Bernoulli numbers")
        return flint_bernoulli(n)
    elif algorithm == 'pari':
        x = pari(n).bernfrac()         # Use the PARI C library
        return Rational(x)
    elif algorithm == 'gap':
        import sage.interfaces.gap
        x = sage.interfaces.gap.gap('Bernoulli(%s)'%n)
        return Rational(x)
    elif algorithm == 'magma':
        import sage.interfaces.magma
        x = sage.interfaces.magma.magma('Bernoulli(%s)'%n)
        return Rational(x)
    elif algorithm == 'gp':
        import sage.interfaces.gp
        x = sage.interfaces.gp.gp('bernfrac(%s)'%n)
        return Rational(x)
    elif algorithm == 'bernmm':
        import sage.rings.bernmm
        return sage.rings.bernmm.bernmm_bern_rat(n, num_threads)
    else:
        raise ValueError("invalid choice of algorithm")


def factorial(n, algorithm='gmp'):
    r"""
    Compute the factorial of `n`, which is the product
    `1\cdot 2\cdot 3 \cdots (n-1)\cdot n`.

    INPUT:

    -  ``n`` - an integer

    -  ``algorithm`` - string (default: 'gmp'):

       -  ``'gmp'`` - use the GMP C-library factorial function

       -  ``'pari'`` - use PARI's factorial function

    OUTPUT: an integer

    EXAMPLES::

        sage: from sage.arith.misc import factorial
        sage: factorial(0)
        1
        sage: factorial(4)
        24
        sage: factorial(10)
        3628800
        sage: factorial(1) == factorial(0)
        True
        sage: factorial(6) == 6*5*4*3*2
        True
        sage: factorial(1) == factorial(0)
        True
        sage: factorial(71) == 71* factorial(70)
        True
        sage: factorial(-32)
        Traceback (most recent call last):
        ...
        ValueError: factorial -- must be nonnegative

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: factorial(int8(4))
        24
        sage: from gmpy2 import mpz
        sage: factorial(mpz(4))
        24


    PERFORMANCE: This discussion is valid as of April 2006. All timings
    below are on a Pentium Core Duo 2Ghz MacBook Pro running Linux with
    a 2.6.16.1 kernel.


    -  It takes less than a minute to compute the factorial of
       `10^7` using the GMP algorithm, and the factorial of
       `10^6` takes less than 4 seconds.

    -  The GMP algorithm is faster and more memory efficient than the
       PARI algorithm. E.g., PARI computes `10^7` factorial in 100
       seconds on the core duo 2Ghz.

    -  For comparison, computation in Magma `\leq` 2.12-10 of
       `n!` is best done using ``*[1..n]``. It takes
       113 seconds to compute the factorial of `10^7` and 6
       seconds to compute the factorial of `10^6`. Mathematica
       V5.2 compute the factorial of `10^7` in 136 seconds and the
       factorial of `10^6` in 7 seconds. (Mathematica is notably
       very efficient at memory usage when doing factorial
       calculations.)
    """
    if n < 0:
        raise ValueError("factorial -- must be nonnegative")
    if algorithm == 'gmp':
        return ZZ(n).factorial()
    elif algorithm == 'pari':
        return pari.factorial(n)
    else:
        raise ValueError('unknown algorithm')


def is_prime(n):
    r"""
    Return ``True`` if `n` is a prime number, and ``False`` otherwise.

    Use a provable primality test or a strong pseudo-primality test depending
    on the global :mod:`arithmetic proof flag <sage.structure.proof.proof>`.

    INPUT:

    -  ``n`` - the object for which to determine primality

    .. SEEALSO::

        - :meth:`is_pseudoprime`
        - :meth:`sage.rings.integer.Integer.is_prime`

    AUTHORS:

    - Kevin Stueve kstueve@uw.edu (2010-01-17):
      delegated calculation to ``n.is_prime()``

    EXAMPLES::

        sage: is_prime(389)
        True
        sage: is_prime(2000)
        False
        sage: is_prime(2)
        True
        sage: is_prime(-1)
        False
        sage: is_prime(1)
        False
        sage: is_prime(-2)
        False

        sage: a = 2**2048 + 981
        sage: is_prime(a)    # not tested - takes ~ 1min
        sage: proof.arithmetic(False)
        sage: is_prime(a)    # instantaneous!
        True
        sage: proof.arithmetic(True)
    """
    try:
        return n.is_prime()
    except (AttributeError, NotImplementedError):
        return ZZ(n).is_prime()


def is_pseudoprime(n):
    r"""
    Test whether ``n`` is a pseudo-prime

    The result is *NOT* proven correct - *this is a pseudo-primality test!*.

    INPUT:

    - ``n`` -- an integer

    .. note::

       We do not consider negatives of prime numbers as prime.

    EXAMPLES::

        sage: is_pseudoprime(389)
        True
        sage: is_pseudoprime(2000)
        False
        sage: is_pseudoprime(2)
        True
        sage: is_pseudoprime(-1)
        False
        sage: factor(-6)
        -1 * 2 * 3
        sage: is_pseudoprime(1)
        False
        sage: is_pseudoprime(-2)
        False
    """
    return ZZ(n).is_pseudoprime()


def is_prime_power(n, get_data=False):
    r"""
    Test whether ``n`` is a positive power of a prime number

    This function simply calls the method :meth:`Integer.is_prime_power()
    <sage.rings.integer.Integer.is_prime_power>` of Integers.

    INPUT:

    - ``n`` -- an integer

    - ``get_data`` -- if set to ``True``, return a pair ``(p,k)`` such that
      this integer equals ``p^k`` instead of ``True`` or ``(self,0)`` instead of
      ``False``

    EXAMPLES::

        sage: is_prime_power(389)
        True
        sage: is_prime_power(2000)
        False
        sage: is_prime_power(2)
        True
        sage: is_prime_power(1024)
        True
        sage: is_prime_power(1024, get_data=True)
        (2, 10)

    The same results can be obtained with::

        sage: 389.is_prime_power()
        True
        sage: 2000.is_prime_power()
        False
        sage: 2.is_prime_power()
        True
        sage: 1024.is_prime_power()
        True
        sage: 1024.is_prime_power(get_data=True)
        (2, 10)

    TESTS::

        sage: is_prime_power(-1)
        False
        sage: is_prime_power(1)
        False
        sage: is_prime_power(QQ(997^100))
        True
        sage: is_prime_power(1/2197)
        Traceback (most recent call last):
        ...
        TypeError: no conversion of this rational to integer
        sage: is_prime_power("foo")
        Traceback (most recent call last):
        ...
        TypeError: unable to convert 'foo' to an integer
        sage: from gmpy2 import mpz
        sage: is_prime_power(mpz(389))
        True
        sage: from numpy import int16
        sage: is_prime_power(int16(389))
        True
    """
    return ZZ(n).is_prime_power(get_data=get_data)


def is_pseudoprime_power(n, get_data=False):
    r"""
    Test if ``n`` is a power of a pseudoprime.

    The result is *NOT* proven correct - *this IS a pseudo-primality test!*.
    Note that a prime power is a positive power of a prime number so that 1 is
    not a prime power.

    INPUT:

    -  ``n`` - an integer

    -  ``get_data`` - (boolean) instead of a boolean return a pair `(p,k)` so
       that ``n`` equals `p^k` and `p` is a pseudoprime or `(n,0)` otherwise.

    EXAMPLES::

        sage: is_pseudoprime_power(389)
        True
        sage: is_pseudoprime_power(2000)
        False
        sage: is_pseudoprime_power(2)
        True
        sage: is_pseudoprime_power(1024)
        True
        sage: is_pseudoprime_power(-1)
        False
        sage: is_pseudoprime_power(1)
        False
        sage: is_pseudoprime_power(997^100)
        True

    Use of the get_data keyword::

        sage: is_pseudoprime_power(3^1024, get_data=True)
        (3, 1024)
        sage: is_pseudoprime_power(2^256, get_data=True)
        (2, 256)
        sage: is_pseudoprime_power(31, get_data=True)
        (31, 1)
        sage: is_pseudoprime_power(15, get_data=True)
        (15, 0)

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: is_pseudoprime_power(int16(1024))
        True
        sage: from gmpy2 import mpz
        sage: is_pseudoprime_power(mpz(1024))
        True
    """
    return ZZ(n).is_prime_power(proof=False, get_data=get_data)


def valuation(m, *args, **kwds):
    """
    Return the valuation of ``m``.

    This function simply calls the m.valuation() method.
    See the documentation of m.valuation() for a more precise description.

    Note that the use of this functions is discouraged as it is better to use
    m.valuation() directly.

    .. NOTE::

        This is not always a valuation in the mathematical sense.
        For more information see:
        sage.rings.finite_rings.integer_mod.IntegerMod_int.valuation

    EXAMPLES::

        sage: valuation(512,2)
        9
        sage: valuation(1,2)
        0
        sage: valuation(5/9, 3)
        -2

    Valuation of 0 is defined, but valuation with respect to 0 is not::

        sage: valuation(0,7)
        +Infinity
        sage: valuation(3,0)
        Traceback (most recent call last):
        ...
        ValueError: You can only compute the valuation with respect to a integer larger than 1.

    Here are some other examples::

        sage: valuation(100,10)
        2
        sage: valuation(200,10)
        2
        sage: valuation(243,3)
        5
        sage: valuation(243*10007,3)
        5
        sage: valuation(243*10007,10007)
        1
        sage: y = QQ['y'].gen()
        sage: valuation(y^3, y)
        3
        sage: x = QQ[['x']].gen()
        sage: valuation((x^3-x^2)/(x-4))
        2
        sage: valuation(4r,2r)
        2
        sage: valuation(1r,1r)
        Traceback (most recent call last):
        ...
        ValueError: You can only compute the valuation with respect to a integer larger than 1.
        sage: from numpy import int16
        sage: valuation(int16(512), int16(2))
        9
        sage: from gmpy2 import mpz
        sage: valuation(mpz(512), mpz(2))
        9
    """
    try:
        return m.valuation(*args, **kwds)
    except AttributeError:
        return ZZ(m).valuation(*args, **kwds)


def prime_powers(start, stop=None):
    r"""
    List of all positive primes powers between ``start`` and
    ``stop``-1, inclusive. If the second argument is omitted, returns
    the prime powers up to the first argument.

    INPUT:

    - ``start`` - an integer. If two inputs are given, a lower bound
      for the returned set of prime powers. If this is the only input,
      then it is an upper bound.

    - ``stop`` - an integer (default: ``None``). An upper bound for the
      returned set of prime powers.

    OUTPUT:

    The set of all prime powers between ``start`` and ``stop`` or, if
    only one argument is passed, the set of all prime powers between 1
    and ``start``. The number `n` is a prime power if `n=p^k`, where
    `p` is a prime number and `k` is a positive integer. Thus, `1` is
    not a prime power.

    EXAMPLES::

        sage: prime_powers(20)
        [2, 3, 4, 5, 7, 8, 9, 11, 13, 16, 17, 19]
        sage: len(prime_powers(1000))
        193
        sage: len(prime_range(1000))
        168

        sage: a = [z for z in range(95,1234) if is_prime_power(z)]
        sage: b = prime_powers(95,1234)
        sage: len(b)
        194
        sage: len(a)
        194
        sage: a[:10]
        [97, 101, 103, 107, 109, 113, 121, 125, 127, 128]
        sage: b[:10]
        [97, 101, 103, 107, 109, 113, 121, 125, 127, 128]
        sage: a == b
        True

        sage: prime_powers(100) == [i for i in range(100) if is_prime_power(i)]
        True

        sage: prime_powers(10,7)
        []
        sage: prime_powers(-5)
        []
        sage: prime_powers(-1,3)
        [2]

    TESTS:

    Check that output are always Sage integers (:trac:`922`)::

        sage: v = prime_powers(10)
        sage: type(v[0])
        <type 'sage.rings.integer.Integer'>

        sage: prime_powers(0,1)
        []
        sage: prime_powers(2)
        []
        sage: prime_powers(3)
        [2]

        sage: prime_powers("foo")
        Traceback (most recent call last):
        ...
        TypeError: unable to convert 'foo' to an integer

        sage: prime_powers(6, "bar")
        Traceback (most recent call last):
        ...
        TypeError: unable to convert 'bar' to an integer

    Check that long input are accepted (:trac:`17852`)::

        sage: prime_powers(6l)
        [2, 3, 4, 5]
        sage: prime_powers(6l,10l)
        [7, 8, 9]

    Check numpy and gmpy2 support::

        sage: from numpy import int8
        sage: prime_powers(int8(20))
        [2, 3, 4, 5, 7, 8, 9, 11, 13, 16, 17, 19]
        sage: from gmpy2 import mpz
        sage: prime_powers(mpz(20))
        [2, 3, 4, 5, 7, 8, 9, 11, 13, 16, 17, 19]
    """
    start = ZZ(start)

    ZZ_2 = Integer(2)
    if stop is None:
        stop = start
        start = ZZ_2
    else:
        stop = ZZ(stop)

    if stop <= ZZ_2 or start >= stop:
        return []

    output = []
    for p in prime_range(stop):
        q = p
        while q < start:
            q *= p
        while q < stop:
            output.append(q)
            q *= p

    output.sort()
    return output


def primes_first_n(n, leave_pari=False):
    r"""
    Return the first `n` primes.

    INPUT:

    - `n` - a nonnegative integer

    OUTPUT:

    - a list of the first `n` prime numbers.

    EXAMPLES::

        sage: primes_first_n(10)
        [2, 3, 5, 7, 11, 13, 17, 19, 23, 29]
        sage: len(primes_first_n(1000))
        1000
        sage: primes_first_n(0)
        []
    """
    if n < 0:
        raise ValueError("n must be nonnegative")
    if n < 1:
        return []
    return prime_range(nth_prime(n) + 1)


# This is from
#    https://github.com/ActiveState/code
#    https://github.com/ActiveState/code/tree/master/recipes/Python/366178_fast_prime_number_list
# It's impressively fast given that it's in Pure Python.


def eratosthenes(n):
    r"""
    Return a list of the primes `\leq n`.

    This is extremely slow and is for educational purposes only.

    INPUT:

    -  ``n`` - a positive integer

    OUTPUT:

    - a list of primes less than or equal to n.

    EXAMPLES::

        sage: eratosthenes(3)
        [2, 3]
        sage: eratosthenes(50)
        [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47]
        sage: len(eratosthenes(100))
        25
        sage: eratosthenes(213) == prime_range(213)
        True

    TESTS::

        sage: from numpy import int8
        sage: eratosthenes(int8(3))
        [2, 3]
        sage: from gmpy2 import mpz
        sage: eratosthenes(mpz(3))
        [2, 3]
    """
    n = int(n)

    if n < 2:
        return []
    elif n == 2:
        return [ZZ(2)]

    s = list(range(3, n + 3, 2))
    mroot = int(n ** 0.5)
    half = (n+1) // 2
    i = 0
    m = 3
    while m <= mroot:
        if s[i]:
            j = (m*m-3) // 2
            s[j] = 0
            while j < half:
                s[j] = 0
                j += m
        i = i+1
        m = 2*i+3

    return [ZZ(2)] + [ZZ(x) for x in s if x and x <= n]


def primes(start, stop=None, proof=None):
    r"""
    Return an iterator over all primes between start and stop-1,
    inclusive. This is much slower than ``prime_range``, but
    potentially uses less memory.  As with :func:`next_prime`, the optional
    argument proof controls whether the numbers returned are
    guaranteed to be prime or not.

    This command is like the Python 2 ``xrange`` command, except it only iterates
    over primes. In some cases it is better to use primes than
    ``prime_range``, because primes does not build a list of all primes in
    the range in memory all at once. However, it is potentially much
    slower since it simply calls the :func:`next_prime` function
    repeatedly, and :func:`next_prime` is slow.

    INPUT:

    - ``start`` - an integer - lower bound for the primes

    - ``stop`` - an integer (or infinity) optional argument -
      giving upper (open) bound for the primes

    - ``proof`` - bool or None (default: None)  If True, the function
      yields only proven primes.  If False, the function uses a
      pseudo-primality test, which is much faster for really big
      numbers but does not provide a proof of primality. If None,
      uses the global default (see :mod:`sage.structure.proof.proof`)

    OUTPUT:

    -  an iterator over primes from start to stop-1, inclusive


    EXAMPLES::

        sage: for p in primes(5,10):
        ....:     print(p)
        5
        7
        sage: list(primes(13))
        [2, 3, 5, 7, 11]
        sage: list(primes(10000000000, 10000000100))
        [10000000019, 10000000033, 10000000061, 10000000069, 10000000097]
        sage: max(primes(10^100, 10^100+10^4, proof=False))
        10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000009631
        sage: next(p for p in primes(10^20, infinity) if is_prime(2*p+1))
        100000000000000001243


    TESTS::

        sage: for a in range(-10, 50):
        ....:     for b in range(-10, 50):
        ....:         assert list(primes(a,b)) == list(filter(is_prime, range(a,b)))
        sage: sum(primes(-10, 9973, proof=False)) == sum(filter(is_prime, range(-10, 9973)))
        True
        sage: for p in primes(10, infinity):
        ....:     if p > 20: break
        ....:     print(p)
        11
        13
        17
        19
        sage: next(p for p in primes(10,oo)) # checks alternate infinity notation
        11
        sage: from numpy import int8
        sage: list(primes(int8(13)))
        [2, 3, 5, 7, 11]
        sage: from gmpy2 import mpz
        sage: list(primes(mpz(13)))
        [2, 3, 5, 7, 11]
    """
    from sage.rings.infinity import infinity

    start = ZZ(start)
    if stop is None:
        stop = start
        start = ZZ(2)
    elif stop != infinity:
        stop = ZZ(stop)
    n = start - 1
    while True:
        n = n.next_prime(proof)
        if n < stop:
            yield n
        else:
            return


def next_prime_power(n):
    """
    Return the smallest prime power greater than ``n``.

    Note that if ``n`` is a prime power, then this function does not return
    ``n``, but the next prime power after ``n``.

    This function just calls the method
    :meth:`Integer.next_prime_power() <sage.rings.integer.Integer.next_prime_power>`
    of Integers.

    .. SEEALSO::

        - :func:`is_prime_power` (and
          :meth:`Integer.is_prime_power() <sage.rings.integer.Integer.is_prime_power()>`)
        - :func:`previous_prime_power` (and
          :meth:`Integer.previous_prime_power() <sage.rings.integer.Integer.next_prime_power>`)

    EXAMPLES::

        sage: next_prime_power(1)
        2
        sage: next_prime_power(2)
        3
        sage: next_prime_power(10)
        11
        sage: next_prime_power(7)
        8
        sage: next_prime_power(99)
        101

    The same results can be obtained with::

        sage: 1.next_prime_power()
        2
        sage: 2.next_prime_power()
        3
        sage: 10.next_prime_power()
        11

    Note that `2` is the smallest prime power::

        sage: next_prime_power(-10)
        2
        sage: next_prime_power(0)
        2

    TESTS::

        sage: from numpy import int8
        sage: next_prime_power(int8(10))
        11
        sage: from gmpy2 import mpz
        sage: next_prime_power(mpz(10))
        11
    """
    return ZZ(n).next_prime_power()


def next_probable_prime(n):
    """
    Return the next probable prime after self, as determined by PARI.

    INPUT:


    -  ``n`` - an integer


    EXAMPLES::

        sage: next_probable_prime(-100)
        2
        sage: next_probable_prime(19)
        23
        sage: next_probable_prime(int(999999999))
        1000000007
        sage: next_probable_prime(2^768)
        1552518092300708935148979488462502555256886017116696611139052038026050952686376886330878408828646477950487730697131073206171580044114814391444287275041181139204454976020849905550265285631598444825262999193716468750892846853816058039

    TESTS::

        sage: from numpy import int8
        sage: next_probable_prime(int8(19))
        23
        sage: from gmpy2 import mpz
        sage: next_probable_prime(mpz(19))
        23
    """
    return ZZ(n).next_probable_prime()


def next_prime(n, proof=None):
    """
    The next prime greater than the integer n. If n is prime, then this
    function does not return n, but the next prime after n. If the
    optional argument proof is False, this function only returns a
    pseudo-prime, as defined by the PARI nextprime function. If it is
    None, uses the global default (see :mod:`sage.structure.proof.proof`)

    INPUT:


    -  ``n`` - integer

    -  ``proof`` - bool or None (default: None)


    EXAMPLES::

        sage: next_prime(-100)
        2
        sage: next_prime(1)
        2
        sage: next_prime(2)
        3
        sage: next_prime(3)
        5
        sage: next_prime(4)
        5

    Notice that the next_prime(5) is not 5 but 7.

    ::

        sage: next_prime(5)
        7
        sage: next_prime(2004)
        2011

    TESTS::

        sage: from numpy import int8
        sage: next_prime(int8(3))
        5
        sage: from gmpy2 import mpz
        sage: next_probable_prime(mpz(3))
        5
    """
    return ZZ(n).next_prime(proof)


def previous_prime(n):
    """
    The largest prime < n. The result is provably correct. If n <= 1,
    this function raises a ValueError.

    EXAMPLES::

        sage: previous_prime(10)
        7
        sage: previous_prime(7)
        5
        sage: previous_prime(8)
        7
        sage: previous_prime(7)
        5
        sage: previous_prime(5)
        3
        sage: previous_prime(3)
        2
        sage: previous_prime(2)
        Traceback (most recent call last):
        ...
        ValueError: no previous prime
        sage: previous_prime(1)
        Traceback (most recent call last):
        ...
        ValueError: no previous prime
        sage: previous_prime(-20)
        Traceback (most recent call last):
        ...
        ValueError: no previous prime

    TESTS::

        sage: from numpy import int8
        sage: previous_prime(int8(7))
        5
        sage: from gmpy2 import mpz
        sage: previous_prime(mpz(7))
        5
    """
    n = ZZ(n)-1
    if n <= 1:
        raise ValueError("no previous prime")
    if n <= 3:
        return ZZ(n)
    if n % 2 == 0:
        n -= 1
    while not is_prime(n):
        n -= 2
    return ZZ(n)


def previous_prime_power(n):
    r"""
    Return the largest prime power smaller than ``n``.

    The result is provably correct. If ``n`` is smaller or equal than ``2`` this
    function raises an error.

    This function simply call the method
    :meth:`Integer.previous_prime_power() <sage.rings.integer.Integer.previous_prime_power>`
    of Integers.

    .. SEEALSO::

        - :func:`is_prime_power` (and :meth:`Integer.is_prime_power()
          <sage.rings.integer.Integer.is_prime_power>`)
        - :func:`next_prime_power` (and :meth:`Integer.next_prime_power()
          <sage.rings.integer.Integer.next_prime_power>`)

    EXAMPLES::

        sage: previous_prime_power(3)
        2
        sage: previous_prime_power(10)
        9
        sage: previous_prime_power(7)
        5
        sage: previous_prime_power(127)
        125

    The same results can be obtained with::

        sage: 3.previous_prime_power()
        2
        sage: 10.previous_prime_power()
        9
        sage: 7.previous_prime_power()
        5
        sage: 127.previous_prime_power()
        125

    Input less than or equal to `2` raises errors::

        sage: previous_prime_power(2)
        Traceback (most recent call last):
        ...
        ValueError: no prime power less than 2
        sage: previous_prime_power(-10)
        Traceback (most recent call last):
        ...
        ValueError: no prime power less than 2

    ::

        sage: n = previous_prime_power(2^16 - 1)
        sage: while is_prime(n):
        ....:     n = previous_prime_power(n)
        sage: factor(n)
        251^2

    TESTS::

        sage: from numpy import int8
        sage: previous_prime_power(int8(10))
        9
        sage: from gmpy2 import mpz
        sage: previous_prime_power(mpz(10))
        9
    """
    return ZZ(n).previous_prime_power()


def random_prime(n, proof=None, lbound=2):
    r"""
    Return a random prime `p` between ``lbound`` and `n`.

    The returned prime `p` satisfies ``lbound`` `\leq p \leq n`.

    The returned prime `p` is chosen uniformly at random from the set
    of prime numbers less than or equal to `n`.

    INPUT:

    -  ``n`` - an integer >= 2.

    -  ``proof`` - bool or None (default: None) If False, the function uses a
       pseudo-primality test, which is much faster for really big numbers but
       does not provide a proof of primality. If None, uses the global default
       (see :mod:`sage.structure.proof.proof`)

    - ``lbound`` - an integer >= 2, lower bound for the chosen primes

    EXAMPLES::

        sage: random_prime(100000)
        30029
        sage: random_prime(2)
        2

    Here we generate a random prime between 100 and 200::

        sage: random_prime(200, lbound=100)
        167

    If all we care about is finding a pseudo prime, then we can pass
    in ``proof=False`` ::

        sage: random_prime(200, proof=False, lbound=100)
        197

    TESTS::

        sage: type(random_prime(2))
        <type 'sage.rings.integer.Integer'>
        sage: type(random_prime(100))
        <type 'sage.rings.integer.Integer'>
        sage: random_prime(1, lbound=-2)   #caused Sage hang #10112
        Traceback (most recent call last):
        ...
        ValueError: n must be greater than or equal to 2
        sage: random_prime(126, lbound=114)
        Traceback (most recent call last):
        ...
        ValueError: There are no primes between 114 and 126 (inclusive)


    AUTHORS:

    - Jon Hanke (2006-08-08): with standard Stein cleanup

    - Jonathan Bober (2007-03-17)
    """
    # since we do not want current_randstate to get
    # pulled when you say "from sage.arith.misc import *".
    from sage.structure.proof.proof import get_flag
    proof = get_flag(proof, "arithmetic")
    n = ZZ(n)
    if n < 2:
        raise ValueError("n must be greater than or equal to 2")
    if n < lbound:
        raise ValueError("n must be at least lbound: %s"%(lbound))
    elif n == 2:
        return n
    lbound = max(2, lbound)
    if lbound > 2:
        if lbound == 3 or n <= 2*lbound - 2:
        # check for Betrand's postulate (proved by Chebyshev)
            if lbound < 25 or n <= 6*lbound/5:
            # see J. Nagura, Proc. Japan Acad. 28, (1952). 177-181.
                if lbound < 2010760 or n <= 16598*lbound/16597:
                # see L. Schoenfeld, Math. Comp. 30 (1976), no. 134, 337-360.
                    if proof:
                        smallest_prime = ZZ(lbound-1).next_prime()
                    else:
                        smallest_prime = ZZ(lbound-1).next_probable_prime()
                    if smallest_prime > n:
                        raise ValueError("There are no primes between %s and %s (inclusive)" % (lbound, n))

    if proof:
        prime_test = is_prime
    else:
        prime_test = is_pseudoprime
    randint = ZZ.random_element
    while True:
        # In order to ensure that the returned prime is chosen
        # uniformly from the set of primes it is necessary to
        # choose a random number and then test for primality.
        # The method of choosing a random number and then returning
        # the closest prime smaller than it would typically not,
        # for example, return the first of a pair of twin primes.
        p = randint(lbound, n)
        if prime_test(p):
            return p


def divisors(n):
    """
    Return the list of all divisors (up to units) of this element
    of a unique factorization domain.

    For an integer, the list of all positive integer divisors
    of this integer, sorted in increasing order, is returned.

    INPUT:

    -  ``n`` - the element

    EXAMPLES:

    Divisors of integers::

        sage: divisors(-3)
        [1, 3]
        sage: divisors(6)
        [1, 2, 3, 6]
        sage: divisors(28)
        [1, 2, 4, 7, 14, 28]
        sage: divisors(2^5)
        [1, 2, 4, 8, 16, 32]
        sage: divisors(100)
        [1, 2, 4, 5, 10, 20, 25, 50, 100]
        sage: divisors(1)
        [1]
        sage: divisors(0)
        Traceback (most recent call last):
        ...
        ValueError: n must be nonzero
        sage: divisors(2^3 * 3^2 * 17)
        [1, 2, 3, 4, 6, 8, 9, 12, 17, 18, 24, 34, 36, 51, 68, 72,
        102, 136, 153, 204, 306, 408, 612, 1224]

    This function works whenever one has unique factorization::

        sage: K.<a> = QuadraticField(7)
        sage: divisors(K.ideal(7))
        [Fractional ideal (1), Fractional ideal (a), Fractional ideal (7)]
        sage: divisors(K.ideal(3))
        [Fractional ideal (1), Fractional ideal (3),
        Fractional ideal (-a + 2), Fractional ideal (-a - 2)]
        sage: divisors(K.ideal(35))
        [Fractional ideal (1), Fractional ideal (5), Fractional ideal (a),
        Fractional ideal (7), Fractional ideal (5*a), Fractional ideal (35)]

    TESTS::

        sage: divisors(int(300))
        [1, 2, 3, 4, 5, 6, 10, 12, 15, 20, 25, 30, 50, 60, 75, 100, 150, 300]
        sage: import numpy
        sage: divisors(numpy.int8(100))
        [1, 2, 4, 5, 10, 20, 25, 50, 100]
        sage: import gmpy2
        sage: divisors(gmpy2.mpz(100))
        [1, 2, 4, 5, 10, 20, 25, 50, 100]
        sage: divisors([])
        Traceback (most recent call last):
        ...
        TypeError: unable to factor []
    """
    try:
        m = n.divisors
    except AttributeError:
        # Maybe n is not a Sage Element, try to convert it
        e = py_scalar_to_element(n)

        if e is n:
            f = factor(n)
            one = parent(n)(1)
            output = [one]
            for p, e in f:
                prev = output[:]
                pn = one
                for i in range(e):
                    pn *= p
                    output.extend(a * pn for a in prev)
            output.sort()
            return output
        n = e
        m = n.divisors

    if not n:
        raise ValueError("n must be nonzero")

    return m()


class Sigma:
    """
    Return the sum of the k-th powers of the divisors of n.

    INPUT:


    -  ``n`` - integer

    -  ``k`` - integer (default: 1)


    OUTPUT: integer

    EXAMPLES::

        sage: sigma(5)
        6
        sage: sigma(5,2)
        26

    The sigma function also has a special plotting method.

    ::

        sage: P = plot(sigma, 1, 100)

    This method also works with k-th powers.

    ::

        sage: P = plot(sigma, 1, 100, k=2)

    AUTHORS:

    - William Stein: original implementation

    - Craig Citro (2007-06-01): rewrote for huge speedup

    TESTS::

        sage: sigma(100,4)
        106811523
        sage: sigma(factorial(100),3).mod(144169)
        3672
        sage: sigma(factorial(150),12).mod(691)
        176
        sage: RR(sigma(factorial(133),20))
        2.80414775675747e4523
        sage: sigma(factorial(100),0)
        39001250856960000
        sage: sigma(factorial(41),1)
        229199532273029988767733858700732906511758707916800
        sage: from numpy import int8
        sage: sigma(int8(100),int8(4))
        106811523
        sage: from gmpy2 import mpz
        sage: sigma(mpz(100),mpz(4))
        106811523
    """
    def __repr__(self):
        """
        A description of this class, which computes the sum of the
        k-th powers of the divisors of n.

        EXAMPLES::

            sage: from sage.arith.misc import Sigma
            sage: Sigma().__repr__()
            'Function that adds up (k-th powers of) the divisors of n'
        """
        return "Function that adds up (k-th powers of) the divisors of n"

    def __call__(self, n, k=1):
        """
        Computes the sum of (the k-th powers of) the divisors of n.

        EXAMPLES::

            sage: from sage.arith.misc import Sigma
            sage: q = Sigma()
            sage: q(10)
            18
            sage: q(10,2)
            130
        """
        n = ZZ(n)
        k = ZZ(k)
        one = ZZ(1)

        if (k == ZZ(0)):
            return prod(expt+one for p, expt in factor(n))
        elif (k == one):
            return prod((p**(expt+one) - one).divide_knowing_divisible_by(p - one)
                          for p, expt in factor(n))
        else:
            return prod((p**((expt+one)*k)-one).divide_knowing_divisible_by(p**k-one)
                          for p,expt in factor(n))

    def plot(self, xmin=1, xmax=50, k=1, pointsize=30, rgbcolor=(0,0,1), join=True,
             **kwds):
        """
        Plot the sigma (sum of k-th powers of divisors) function.

        INPUT:


        -  ``xmin`` - default: 1

        -  ``xmax`` - default: 50

        -  ``k`` - default: 1

        -  ``pointsize`` - default: 30

        -  ``rgbcolor`` - default: (0,0,1)

        -  ``join`` - default: True; whether to join the
           points.

        -  ``**kwds`` - passed on

        EXAMPLES::

            sage: from sage.arith.misc import Sigma
            sage: p = Sigma().plot()
            sage: p.ymax()
            124.0
        """
        v = [(n,sigma(n,k)) for n in range(xmin,xmax + 1)]
        from sage.plot.all import list_plot
        P = list_plot(v, pointsize=pointsize, rgbcolor=rgbcolor, **kwds)
        if join:
            P += list_plot(v, plotjoined=True, rgbcolor=(0.7,0.7,0.7), **kwds)
        return P


sigma = Sigma()


def gcd(a, b=None, **kwargs):
    r"""
    Return the greatest common divisor of ``a`` and ``b``.

    If ``a`` is a list and ``b`` is omitted, return instead the
    greatest common divisor of all elements of ``a``.

    INPUT:

    - ``a,b`` -- two elements of a ring with gcd or

    - ``a`` -- a list or tuple of elements of a ring with gcd

    Additional keyword arguments are passed to the respectively called
    methods.

    OUTPUT:

    The given elements are first coerced into a common parent. Then,
    their greatest common divisor *in that common parent* is returned.

    EXAMPLES::

        sage: GCD(97,100)
        1
        sage: GCD(97*10^15, 19^20*97^2)
        97
        sage: GCD(2/3, 4/5)
        2/15
        sage: GCD([2,4,6,8])
        2
        sage: GCD(srange(0,10000,10))  # fast  !!
        10

    Note that to take the gcd of `n` elements for `n \not= 2` you must
    put the elements into a list by enclosing them in ``[..]``.  Before
    :trac:`4988` the following wrongly returned 3 since the third parameter
    was just ignored::

        sage: gcd(3, 6, 2)
        Traceback (most recent call last):
        ...
        TypeError: gcd() takes ...
        sage: gcd([3, 6, 2])
        1

    Similarly, giving just one element (which is not a list) gives an error::

        sage: gcd(3)
        Traceback (most recent call last):
        ...
        TypeError: 'sage.rings.integer.Integer' object is not iterable

    By convention, the gcd of the empty list is (the integer) 0::

        sage: gcd([])
        0
        sage: type(gcd([]))
        <type 'sage.rings.integer.Integer'>

    TESTS:

    The following shows that indeed coercion takes place before computing
    the gcd. This behaviour was introduced in :trac:`10771`::

        sage: R.<x>=QQ[]
        sage: S.<x>=ZZ[]
        sage: p = S.random_element(degree=(0,10))
        sage: q = R.random_element(degree=(0,10))
        sage: parent(gcd(1/p,q))
        Fraction Field of Univariate Polynomial Ring in x over Rational Field
        sage: parent(gcd([1/p,q]))
        Fraction Field of Univariate Polynomial Ring in x over Rational Field

    Make sure we try QQ and not merely ZZ (:trac:`13014`)::

        sage: bool(gcd(2/5, 3/7) == gcd(SR(2/5), SR(3/7)))
        True

    Make sure that the gcd of Expressions stays symbolic::

        sage: parent(gcd(2, 4))
        Integer Ring
        sage: parent(gcd(SR(2), 4))
        Symbolic Ring
        sage: parent(gcd(2, SR(4)))
        Symbolic Ring
        sage: parent(gcd(SR(2), SR(4)))
        Symbolic Ring

    Verify that objects without gcd methods but which cannot be
    coerced to ZZ or QQ raise an error::

        sage: F.<a,b> = FreeMonoid(2)
        sage: gcd(a,b)
        Traceback (most recent call last):
        ...
        TypeError: unable to call gcd with a

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: GCD(int8(97),int8(100))
        1
        sage: from gmpy2 import mpq, mpz
        sage: GCD(mpq(2/3), mpq(4/5))
        2/15
        sage: GCD((mpz(2), mpz(4)))
        2
    """
    # Most common use case first:
    if b is not None:
        try:
            m = a.gcd
        except (AttributeError, TypeError):
            e = py_scalar_to_element(a)
            if e is a:
                raise TypeError("unable to call gcd with {!r}".format(a))
            m = e.gcd

        try:
            return m(b, **kwargs)
        except TypeError:
            return m(py_scalar_to_element(b), **kwargs)

    from sage.structure.sequence import Sequence
    seq = Sequence(py_scalar_to_element(el) for el in a)
    if seq.universe() is ZZ:
        return GCD_list(a)
    else:
        return __GCD_sequence(seq, **kwargs)


GCD = gcd


def __GCD_sequence(v, **kwargs):
    """
    Internal function returning the gcd of the elements of a sequence

    INPUT:


    -  ``v`` - A sequence (possibly empty)


    OUTPUT: The gcd of the elements of the sequence as an element of
    the sequence's universe, or the integer 0 if the sequence is
    empty.

    EXAMPLES::

        sage: from sage.arith.misc import __GCD_sequence
        sage: from sage.structure.sequence import Sequence
        sage: l = ()
        sage: __GCD_sequence(l)
        0
        sage: __GCD_sequence(Sequence(srange(10)))
        1
        sage: X=polygen(QQ)
        sage: __GCD_sequence(Sequence((2*X+4,2*X^2,2)))
        1
        sage: X=polygen(ZZ)
        sage: __GCD_sequence(Sequence((2*X+4,2*X^2,2)))
        2
        sage: __GCD_sequence(Sequence((1/1,1/2)))
        1/2

    TESTS::

        sage: __GCD_sequence(Sequence((1,1/2,1/5)))
        1/10
    """
    if len(v) == 0:
        return ZZ(0)
    if hasattr(v,'universe'):
        g = v.universe()(0)
    else:
        g = ZZ(0)
    for vi in v:
        g = vi.gcd(g, **kwargs)
    return g


def xlcm(m, n):
    r"""
    Extended lcm function: given two positive integers `m,n`, returns
    a triple `(l,m_1,n_1)` such that `l=\mathop{\mathrm{lcm}}(m,n)=m_1
    \cdot n_1` where `m_1|m`, `n_1|n` and `\gcd(m_1,n_1)=1`, all with no
    factorization.

    Used to construct an element of order `l` from elements of orders `m,n`
    in any group: see sage/groups/generic.py for examples.

    EXAMPLES::

        sage: xlcm(120,36)
        (360, 40, 9)

    TESTS::

        sage: from numpy import int16
        sage: xlcm(int16(120), int16(36))
        (360, 40, 9)
        sage: from gmpy2 import mpz
        sage: xlcm(mpz(120), mpz(36))
        (360, 40, 9)
    """
    m = py_scalar_to_element(m)
    n = py_scalar_to_element(n)

    g = gcd(m, n)
    l = m*n//g       # = lcm(m, n)
    g = gcd(m, n//g) # divisible by those primes which divide n to a
                     # higher power than m

    while g != 1:
        m //= g
        g = gcd(m, g)

    n = l//m
    return (l, m, n)


def xgcd(a, b):
    r"""
    Return a triple ``(g,s,t)`` such that `g = s\cdot a+t\cdot b = \gcd(a,b)`.

    .. NOTE::

       One exception is if `a` and `b` are not in a principal ideal domain (see
       :wikipedia:`Principal_ideal_domain`), e.g., they are both polynomials
       over the integers. Then this function can't in general return ``(g,s,t)``
       as above, since they need not exist.  Instead, over the integers, we
       first multiply `g` by a divisor of the resultant of `a/g` and `b/g`, up
       to sign.

    INPUT:

    -  ``a, b`` - integers or more generally, element of a ring for which the
       xgcd make sense (e.g. a field or univariate polynomials).

    OUTPUT:

    -  ``g, s, t`` - such that `g = s\cdot a + t\cdot b`

    .. NOTE::

       There is no guarantee that the returned cofactors (s and t) are
       minimal.

    EXAMPLES::

        sage: xgcd(56, 44)
        (4, 4, -5)
        sage: 4*56 + (-5)*44
        4

        sage: g, a, b = xgcd(5/1, 7/1); g, a, b
        (1, 3, -2)
        sage: a*(5/1) + b*(7/1) == g
        True

        sage: x = polygen(QQ)
        sage: xgcd(x^3 - 1, x^2 - 1)
        (x - 1, 1, -x)

        sage: K.<g> = NumberField(x^2-3)
        sage: g.xgcd(g+2)
        (1, 1/3*g, 0)

        sage: R.<a,b> = K[]
        sage: S.<y> = R.fraction_field()[]
        sage: xgcd(y^2, a*y+b)
        (1, a^2/b^2, ((-a)/b^2)*y + 1/b)
        sage: xgcd((b+g)*y^2, (a+g)*y+b)
        (1, (a^2 + (2*g)*a + 3)/(b^3 + (g)*b^2), ((-a + (-g))/b^2)*y + 1/b)

    Here is an example of a xgcd for two polynomials over the integers, where the linear
    combination is not the gcd but the gcd multiplied by the resultant::

        sage: R.<x> = ZZ[]
        sage: gcd(2*x*(x-1), x^2)
        x
        sage: xgcd(2*x*(x-1), x^2)
        (2*x, -1, 2)
        sage: (2*(x-1)).resultant(x)
        2

    Tests with numpy and gmpy2 types::

        sage: from numpy import int8
        sage: xgcd(4,int8(8))
        (4, 1, 0)
        sage: xgcd(int8(4),int8(8))
        (4, 1, 0)
        sage: from gmpy2 import mpz
        sage: xgcd(mpz(4), mpz(8))
        (4, 1, 0)
        sage: xgcd(4, mpz(8))
        (4, 1, 0)

    TESTS:

    We check that :trac:`3330` has been fixed::

        sage: R.<a,b> = NumberField(x^2-3,'g').extension(x^2-7,'h')[]
        sage: h = R.base_ring().gen()
        sage: S.<y> = R.fraction_field()[]
        sage: xgcd(y^2, a*h*y+b)
        (1, 7*a^2/b^2, (((-h)*a)/b^2)*y + 1/b)
    """
    try:
        return a.xgcd(b)
    except AttributeError:
        a = py_scalar_to_element(a)
        b = py_scalar_to_element(b)
    except TypeError:
        b = py_scalar_to_element(b)
    return a.xgcd(b)

XGCD = xgcd

## def XGCD_python(a, b):
##     """
##     Return triple (g,p,q) such that g = p*a+b*q = GCD(a,b).
##     This function should behave exactly the same as XGCD,
##     but is implemented in pure python.
##     """
##     if a == 0 and b == 0:
##         return (0,0,1)
##     if a == 0:
##         return (abs(b), 0, b/abs(b))
##     if b == 0:
##         return (abs(a), a/abs(a), 0)
##     psign = 1
##     qsign = 1
##     if a < 0:
##         a = -a
##         psign = -1
##     if b < 0:
##         b = -b
##         qsign = -1
##     p = 1; q = 0; r = 0; s = 1
##     while b != 0:
##         c = a % b
##         quot = a/b
##         a = b; b = c
##         new_r = p - quot*r
##         new_s = q - quot*s
##         p = r; q = s
##         r = new_r; s = new_s
##     return (a, p*psign, q*qsign)


def xkcd(n=""):
    r"""
    This function is similar to the xgcd function, but behaves
    in a completely different way.

    See https://xkcd.com/json.html for more details.

    INPUT:

    - ``n`` -- an integer (optional)

    OUTPUT: a fragment of HTML

    EXAMPLES::

        sage: xkcd(353)  # optional - internet
        <h1>Python</h1><img src="https://imgs.xkcd.com/comics/python.png" title="I wrote 20 short programs in Python yesterday.  It was wonderful.  Perl, I'm leaving you."><div>Source: <a href="http://xkcd.com/353" target="_blank">http://xkcd.com/353</a></div>
    """
    import contextlib
    import json
    from sage.misc.html import html

    # import compatible with py2 and py3
    from urllib.request import urlopen
    from urllib.error import HTTPError, URLError

    data = None
    if not n:
        # default to last comic
        url = "http://xkcd.com/info.0.json"
    else:
        url = "https://xkcd.com/{}/info.0.json".format(n)

    try:
        with contextlib.closing(urlopen(url)) as f:
            data = f.read()
    except HTTPError as error:
        if error.getcode() == 400: # this error occurs when asking for a non valid comic number
            raise RuntimeError("Could not obtain comic data from {}. Maybe you should enable time travel!".format(url))
    except URLError:
        pass

    if n == 1024:
        data = None

    if data:
        data = json.loads(data)
        img = data['img']
        alt = data['alt']
        title = data['safe_title']
        link = "http://xkcd.com/{}".format(data['num'])
        return html('<h1>{}</h1><img src="{}" title="{}">'.format(title, img, alt)
            + '<div>Source: <a href="{0}" target="_blank">{0}</a></div>'.format(link))

    # TODO: raise this error in such a way that it's not clear that
    # it is produced by sage, see http://xkcd.com/1024/
    return html('<script> alert("Error: -41"); </script>')


def inverse_mod(a, m):
    """
    The inverse of the ring element a modulo m.

    If no special inverse_mod is defined for the elements, it tries to
    coerce them into integers and perform the inversion there

    ::

        sage: inverse_mod(7,1)
        0
        sage: inverse_mod(5,14)
        3
        sage: inverse_mod(3,-5)
        2

    Tests with numpy and mpz numbers::

        sage: from numpy import int8
        sage: inverse_mod(int8(5),int8(14))
        3
        sage: from gmpy2 import mpz
        sage: inverse_mod(mpz(5),mpz(14))
        3
    """
    try:
        return a.inverse_mod(m)
    except AttributeError:
        return Integer(a).inverse_mod(m)

#######################################################
# Functions to find the fastest available commands
# for gcd and inverse_mod
#######################################################


def get_gcd(order):
    """
    Return the fastest gcd function for integers of size no larger than
    order.

    EXAMPLES::

        sage: sage.arith.misc.get_gcd(4000)
        <built-in method gcd_int of sage.rings.fast_arith.arith_int object at ...>
        sage: sage.arith.misc.get_gcd(400000)
        <built-in method gcd_longlong of sage.rings.fast_arith.arith_llong object at ...>
        sage: sage.arith.misc.get_gcd(4000000000)
        <function gcd at ...>
    """
    if order <= 46340:   # todo: don't hard code
        return arith_int().gcd_int
    elif order <= 2147483647:   # todo: don't hard code
        return arith_llong().gcd_longlong
    else:
        return gcd


def get_inverse_mod(order):
    """
    Return the fastest inverse_mod function for integers of size no
    larger than order.

    EXAMPLES::

        sage: sage.arith.misc.get_inverse_mod(6000)
        <built-in method inverse_mod_int of sage.rings.fast_arith.arith_int object at ...>
        sage: sage.arith.misc.get_inverse_mod(600000)
        <built-in method inverse_mod_longlong of sage.rings.fast_arith.arith_llong object at ...>
        sage: sage.arith.misc.get_inverse_mod(6000000000)
        <function inverse_mod at ...>
    """
    if order <= 46340:   # todo: don't hard code
        return arith_int().inverse_mod_int
    elif order <= 2147483647:   # todo: don't hard code
        return arith_llong().inverse_mod_longlong
    else:
        return inverse_mod

# def sqrt_mod(a, m):
#     """A square root of a modulo m."""

# def xxx_inverse_mod(a, m):
#     """The inverse of a modulo m."""
#     g,s,t = XGCD(a,m)
#     if g != 1:
#         raise "inverse_mod(a=%s,m=%s), error since GCD=%s"%(a,m,g)
#     return s


def power_mod(a, n, m):
    """
    Return the ``n``-th power of ``a`` modulo the integer ``m``.

    EXAMPLES::

        sage: power_mod(0,0,5)
        Traceback (most recent call last):
        ...
        ArithmeticError: 0^0 is undefined.
        sage: power_mod(2,390,391)
        285
        sage: power_mod(2,-1,7)
        4
        sage: power_mod(11,1,7)
        4
        sage: R.<x> = ZZ[]
        sage: power_mod(3*x, 10, 7)
        4*x^10

        sage: power_mod(11,1,0)
        Traceback (most recent call last):
        ...
        ZeroDivisionError: modulus must be nonzero.

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int32
        sage: power_mod(int32(2),int32(390),int32(391))
        285
        sage: from gmpy2 import mpz
        sage: power_mod(mpz(2),mpz(390),mpz(391))
        mpz(285)
    """
    if m == 0:
        raise ZeroDivisionError("modulus must be nonzero.")
    if m == 1:
        return 0
    if n < 0:
        ainv = inverse_mod(a, m)
        return power_mod(ainv, -n, m)
    if n == 0:
        if a == 0:
            raise ArithmeticError("0^0 is undefined.")
        return 1

    apow = a % m
    while n&1 == 0:
        apow = (apow*apow) % m
        n = n >> 1
    power = apow
    n = n >> 1
    while n != 0:
        apow = (apow*apow) % m
        if n&1 != 0:
            power = (power*apow) % m
        n = n >> 1

    return power


def rational_reconstruction(a, m, algorithm='fast'):
    r"""
    This function tries to compute `x/y`, where `x/y` is a rational number in
    lowest terms such that the reduction of `x/y` modulo `m` is equal to `a` and
    the absolute values of `x` and `y` are both `\le \sqrt{m/2}`. If such `x/y`
    exists, that pair is unique and this function returns it. If no
    such pair exists, this function raises ZeroDivisionError.

    An efficient algorithm for computing rational reconstruction is
    very similar to the extended Euclidean algorithm. For more details,
    see Knuth, Vol 2, 3rd ed, pages 656-657.

    INPUT:

    - ``a`` -- an integer

    - ``m`` -- a modulus

    - ``algorithm`` -- (default: 'fast')

      - ``'fast'`` - a fast implementation using direct MPIR calls
        in Cython.

    OUTPUT:

    Numerator and denominator `n`, `d` of the unique rational number
    `r=n/d`, if it exists, with `n` and `|d| \le \sqrt{N/2}`. Return
    `(0,0)` if no such number exists.

    The algorithm for rational reconstruction is described (with a
    complete nontrivial proof) on pages 656-657 of Knuth, Vol 2, 3rd
    ed. as the solution to exercise 51 on page 379. See in particular
    the conclusion paragraph right in the middle of page 657, which
    describes the algorithm thus:

        This discussion proves that the problem can be solved
        efficiently by applying Algorithm 4.5.2X with `u=m` and `v=a`,
        but with the following replacement for step X2: If
        `v3 \le \sqrt{m/2}`, the algorithm terminates. The pair
        `(x,y)=(|v2|,v3*\mathrm{sign}(v2))` is then the unique
        solution, provided that `x` and `y` are coprime and
        `x \le \sqrt{m/2}`; otherwise there is no solution. (Alg 4.5.2X is
        the extended Euclidean algorithm.)

    Knuth remarks that this algorithm is due to Wang, Kornerup, and
    Gregory from around 1983.

    EXAMPLES::

        sage: m = 100000
        sage: (119*inverse_mod(53,m))%m
        11323
        sage: rational_reconstruction(11323,m)
        119/53

    ::

        sage: rational_reconstruction(400,1000)
        Traceback (most recent call last):
        ...
        ArithmeticError: rational reconstruction of 400 (mod 1000) does not exist

    ::

        sage: rational_reconstruction(3, 292393)
        3
        sage: a = Integers(292393)(45/97); a
        204977
        sage: rational_reconstruction(a, 292393, algorithm='fast')
        45/97
        sage: rational_reconstruction(293048, 292393)
        Traceback (most recent call last):
        ...
        ArithmeticError: rational reconstruction of 655 (mod 292393) does not exist
        sage: rational_reconstruction(0, 0)
        Traceback (most recent call last):
        ...
        ZeroDivisionError: rational reconstruction with zero modulus
        sage: rational_reconstruction(0, 1, algorithm="foobar")
        Traceback (most recent call last):
        ...
        ValueError: unknown algorithm 'foobar'

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int32
        sage: rational_reconstruction(int32(3), int32(292393))
        3
        sage: from gmpy2 import mpz
        sage: rational_reconstruction(mpz(3), mpz(292393))
        3
    """
    if algorithm == 'fast':
        return ZZ(a).rational_reconstruction(ZZ(m))
    else:
        raise ValueError("unknown algorithm %r" % algorithm)


def mqrr_rational_reconstruction(u, m, T):
    r"""
    Maximal Quotient Rational Reconstruction.

    For research purposes only - this is pure Python, so slow.

    INPUT:

    - ``u, m, T`` -  integers such that `m > u \ge 0`, `T > 0`.

    OUTPUT:

    Either integers `n,d` such that `d>0`, `\mathop{\mathrm{gcd}}(n,d)=1`, `n/d=u \bmod m`, and
    `T \cdot d \cdot |n| < m`, or ``None``.

    Reference: Monagan, Maximal Quotient Rational Reconstruction: An
    Almost Optimal Algorithm for Rational Reconstruction (page 11)

    This algorithm is probabilistic.

    EXAMPLES::

        sage: mqrr_rational_reconstruction(21,3100,13)
        (21, 1)

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: mqrr_rational_reconstruction(int16(21),int16(3100),int16(13))
        (21, 1)
        sage: from gmpy2 import mpz
        sage: mqrr_rational_reconstruction(mpz(21),mpz(3100),mpz(13))
        (21, 1)
    """
    u = py_scalar_to_element(u)
    m = py_scalar_to_element(m)
    T = py_scalar_to_element(T)

    if u == 0:
        if m > T:
            return (0,1)
        else:
            return None
    n, d = 0, 0
    t0, r0 = 0, m
    t1, r1 = 1, u
    while r1 != 0 and r0 > T:
        q = r0/r1   # C division implicit floor
        if q > T:
            n, d, T = r1, t1, q
        r0, r1 = r1, r0 - q*r1
        t0, t1 = t1, t0 - q*t1
    if d != 0 and GCD(n,d) == 1:
        return (n,d)
    return None


######################


def trial_division(n, bound=None):
    """
    Return the smallest prime divisor <= bound of the positive integer
    n, or n if there is no such prime. If the optional argument bound
    is omitted, then bound <= n.

    INPUT:

    -  ``n`` - a positive integer

    - ``bound`` - (optional) a positive integer

    OUTPUT:

    -  ``int`` - a prime p=bound that divides n, or n if
       there is no such prime.


    EXAMPLES::

        sage: trial_division(15)
        3
        sage: trial_division(91)
        7
        sage: trial_division(11)
        11
        sage: trial_division(387833, 300)
        387833
        sage: # 300 is not big enough to split off a
        sage: # factor, but 400 is.
        sage: trial_division(387833, 400)
        389

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: trial_division(int8(91))
        7
        sage: from gmpy2 import mpz
        sage: trial_division(mpz(91))
        7
    """
    if bound is None:
        return ZZ(n).trial_division()
    else:
        return ZZ(n).trial_division(bound)


def factor(n, proof=None, int_=False, algorithm='pari', verbose=0, **kwds):
    """
    Return the factorization of ``n``.  The result depends on the
    type of ``n``.

    If ``n`` is an integer, returns the factorization as an object
    of type ``Factorization``.

    If n is not an integer, ``n.factor(proof=proof, **kwds)`` gets called.
    See ``n.factor??`` for more documentation in this case.

    .. warning::

       This means that applying ``factor`` to an integer result of
       a symbolic computation will not factor the integer, because it is
       considered as an element of a larger symbolic ring.

       EXAMPLES::

           sage: f(n)=n^2
           sage: is_prime(f(3))
           False
           sage: factor(f(3))
           9

    INPUT:

    -  ``n`` - an nonzero integer

    -  ``proof`` - bool or None (default: None)

    -  ``int_`` - bool (default: False) whether to return
       answers as Python ints

    -  ``algorithm`` - string

       - ``'pari'`` - (default) use the PARI c library

       - ``'kash'`` - use KASH computer algebra system (requires that
         kash be installed)

       - ``'magma'`` - use Magma (requires magma be installed)

    -  ``verbose`` - integer (default: 0); PARI's debug
       variable is set to this; e.g., set to 4 or 8 to see lots of output
       during factorization.

    OUTPUT:

    -  factorization of n

    The qsieve and ecm commands give access to highly optimized
    implementations of algorithms for doing certain integer
    factorization problems. These implementations are not used by the
    generic factor command, which currently just calls PARI (note that
    PARI also implements sieve and ecm algorithms, but they are not as
    optimized). Thus you might consider using them instead for certain
    numbers.

    The factorization returned is an element of the class
    :class:`~sage.structure.factorization.Factorization`; see Factorization??
    for more details, and examples below for usage. A Factorization contains
    both the unit factor (+1 or -1) and a sorted list of (prime, exponent)
    pairs.

    The factorization displays in pretty-print format but it is easy to
    obtain access to the (prime,exponent) pairs and the unit, to
    recover the number from its factorization, and even to multiply two
    factorizations. See examples below.

    EXAMPLES::

        sage: factor(500)
        2^2 * 5^3
        sage: factor(-20)
        -1 * 2^2 * 5
        sage: f=factor(-20)
        sage: list(f)
        [(2, 2), (5, 1)]
        sage: f.unit()
        -1
        sage: f.value()
        -20
        sage: factor( -next_prime(10^2) * next_prime(10^7) )
        -1 * 101 * 10000019

    ::

        sage: factor(-500, algorithm='kash')      # optional - kash
        -1 * 2^2 * 5^3

    ::

        sage: factor(-500, algorithm='magma')     # optional - magma
        -1 * 2^2 * 5^3

    ::

        sage: factor(0)
        Traceback (most recent call last):
        ...
        ArithmeticError: factorization of 0 is not defined
        sage: factor(1)
        1
        sage: factor(-1)
        -1
        sage: factor(2^(2^7)+1)
        59649589127497217 * 5704689200685129054721

    Sage calls PARI's factor, which has proof False by default.
    Sage has a global proof flag, set to True by default (see
    :mod:`sage.structure.proof.proof`, or proof.[tab]). To override
    the default, call this function with proof=False.

    ::

        sage: factor(3^89-1, proof=False)
        2 * 179 * 1611479891519807 * 5042939439565996049162197

    ::

        sage: factor(2^197 + 1)  # long time (2s)
        3 * 197002597249 * 1348959352853811313 * 251951573867253012259144010843

    Any object which has a factor method can be factored like this::

        sage: K.<i> = QuadraticField(-1)
        sage: factor(122 - 454*i)
        (-3*i - 2) * (-i - 2)^3 * (i + 1)^3 * (i + 4)

    To access the data in a factorization::

        sage: f = factor(420); f
        2^2 * 3 * 5 * 7
        sage: [x for x in f]
        [(2, 2), (3, 1), (5, 1), (7, 1)]
        sage: [p for p,e in f]
        [2, 3, 5, 7]
        sage: [e for p,e in f]
        [2, 1, 1, 1]
        sage: [p^e for p,e in f]
        [4, 3, 5, 7]

    We can factor Python, numpy and gmpy2 numbers::

        sage: factor(math.pi)
        3.141592653589793
        sage: import numpy
        sage: factor(numpy.int8(30))
        2 * 3 * 5
        sage: import gmpy2
        sage: factor(gmpy2.mpz(30))
        2 * 3 * 5

    TESTS::

        sage: factor(Mod(4, 100))
        Traceback (most recent call last):
        ...
        TypeError: unable to factor 4
        sage: factor("xyz")
        Traceback (most recent call last):
        ...
        TypeError: unable to factor 'xyz'
    """
    try:
        m = n.factor
    except AttributeError:
        # Maybe n is not a Sage Element, try to convert it
        e = py_scalar_to_element(n)
        if e is n:
            # Either n was a Sage Element without a factor() method
            # or we cannot it convert it to Sage
            raise TypeError("unable to factor {!r}".format(n))
        n = e
        m = n.factor

    if isinstance(n, Integer):
        return m(proof=proof, algorithm=algorithm, int_=int_,
                 verbose=verbose, **kwds)

    # Polynomial or other factorable object
    try:
        return m(proof=proof, **kwds)
    except TypeError:
        # Maybe the factor() method does not have a proof option
        return m(**kwds)


def radical(n, *args, **kwds):
    """
    Return the product of the prime divisors of n.

    This calls ``n.radical(*args, **kwds)``.

    EXAMPLES::

        sage: radical(2 * 3^2 * 5^5)
        30
        sage: radical(0)
        Traceback (most recent call last):
        ...
        ArithmeticError: Radical of 0 not defined.
        sage: K.<i> = QuadraticField(-1)
        sage: radical(K(2))
        i + 1

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: radical(int8(50))
        10
        sage: from gmpy2 import mpz
        sage: radical(mpz(50))
        10
    """
    try:
        m = n.radical
    except AttributeError:
        # Maybe n is not a Sage Element, try to convert it
        e = py_scalar_to_element(n)
        if e is n:
            m = factor(n, *args, **kwds).radical_value
        else:
            m = e.radical

    return m(*args, **kwds)


def prime_divisors(n):
    """
    Return the list of prime divisors (up to units) of this element
    of a unique factorization domain.

    INPUT:

    - ``n`` -- any object which can be decomposed into prime factors

    OUTPUT:

    A list of prime factors of ``n``. For integers, this list is sorted
    in increasing order.

    EXAMPLES:

    Prime divisors of positive integers::

        sage: prime_divisors(1)
        []
        sage: prime_divisors(100)
        [2, 5]
        sage: prime_divisors(2004)
        [2, 3, 167]

    If ``n`` is negative, we do *not* include -1 among the prime
    divisors, since -1 is not a prime number::

        sage: prime_divisors(-100)
        [2, 5]

    For polynomials we get all irreducible factors::

        sage: R.<x> = PolynomialRing(QQ)
        sage: prime_divisors(x^12 - 1)
        [x - 1, x + 1, x^2 - x + 1, x^2 + 1, x^2 + x + 1, x^4 - x^2 + 1]

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: prime_divisors(int8(-100))
        [2, 5]
        sage: from gmpy2 import mpz
        sage: prime_divisors(mpz(-100))
        [2, 5]
    """
    try:
        return n.prime_divisors()
    except AttributeError:
        pass
    return [p for p, _ in factor(n)]


prime_factors = prime_divisors


def odd_part(n):
    r"""
    The odd part of the integer `n`. This is `n / 2^v`,
    where `v = \mathrm{valuation}(n,2)`.

    EXAMPLES::

        sage: odd_part(5)
        5
        sage: odd_part(4)
        1
        sage: odd_part(factorial(31))
        122529844256906551386796875

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: odd_part(int8(5))
        5
        sage: from gmpy2 import mpz
        sage: odd_part(mpz(5))
        5
    """
    if not isinstance(n, Integer):
        n = ZZ(n)
    return n.odd_part()


def prime_to_m_part(n, m):
    """
    Return the prime-to-``m`` part of ``n``.

    This is the largest divisor of ``n`` that is coprime to ``m``.

    INPUT:

    - ``n`` -- Integer (nonzero)

    - ``m`` -- Integer

    OUTPUT: Integer

    EXAMPLES::

        sage: prime_to_m_part(240,2)
        15
        sage: prime_to_m_part(240,3)
        80
        sage: prime_to_m_part(240,5)
        48
        sage: prime_to_m_part(43434,20)
        21717

    Note that integers also have a method with the same name::

        sage: 240.prime_to_m_part(2)
        15

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: prime_to_m_part(int16(240), int16(2))
        15
        sage: from gmpy2 import mpz
        sage: prime_to_m_part(mpz(240), mpz(2))
        15
    """
    return ZZ(n).prime_to_m_part(m)


def is_square(n, root=False):
    """
    Return whether or not ``n`` is square.

    If ``n`` is a square also return the square root.
    If ``n`` is not square, also return ``None``.

    INPUT:

    - ``n`` -- an integer

    - ``root`` -- whether or not to also return a square
      root (default: ``False``)

    OUTPUT:

    - ``bool`` -- whether or not a square

    - ``object`` -- (optional) an actual square if found,
      and ``None`` otherwise.

    EXAMPLES::

        sage: is_square(2)
        False
        sage: is_square(4)
        True
        sage: is_square(2.2)
        True
        sage: is_square(-2.2)
        False
        sage: is_square(CDF(-2.2))
        True
        sage: is_square((x-1)^2)
        Traceback (most recent call last):
        ...
        NotImplementedError: is_square() not implemented for non numeric elements of Symbolic Ring

    ::

        sage: is_square(4, True)
        (True, 2)

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: is_square(int8(4))
        True
        sage: from gmpy2 import mpz
        sage: is_square(mpz(4))
        True

    Tests with Polynomial::

        sage: R.<v> = LaurentPolynomialRing(QQ, 'v')
        sage: H = IwahoriHeckeAlgebra('A3', v**2)
        sage: R.<a,b,c,d> = QQ[]
        sage: p = a*b + c*d*a*d*a + 5
        sage: is_square(p**2)
        True
    """
    try:
        m = n.is_square
    except (AttributeError, NotImplementedError):
        n = py_scalar_to_element(n)
        m = n.is_square

    if root:
        try:
            return m(root)
        except TypeError:
            if m():
                return True, n.sqrt()
            else:
                return False, None
    return m()


def is_squarefree(n):
    """
    Test whether ``n`` is square free.

    EXAMPLES::

        sage: is_squarefree(100)
        False
        sage: is_squarefree(101)
        True

        sage: R = ZZ['x']
        sage: x = R.gen()
        sage: is_squarefree((x^2+x+1) * (x-2))
        True
        sage: is_squarefree((x-1)**2 * (x-3))
        False

        sage: O = ZZ[sqrt(-1)]
        sage: I = O.gen(1)
        sage: is_squarefree(I+1)
        True
        sage: is_squarefree(O(2))
        False
        sage: O(2).factor()
        (-I) * (I + 1)^2

    This method fails on domains which are not Unique Factorization Domains::

        sage: O = ZZ[sqrt(-5)]
        sage: a = O.gen(1)
        sage: is_squarefree(a - 3)
        Traceback (most recent call last):
        ...
        ArithmeticError: non-principal ideal in factorization

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: is_squarefree(int8(100))
        False
        sage: is_squarefree(int8(101))
        True
        sage: from gmpy2 import mpz
        sage: is_squarefree(mpz(100))
        False
        sage: is_squarefree(mpz(101))
        True
    """
    e = py_scalar_to_element(n)

    try:
        return e.is_squarefree()
    except AttributeError:
        pass

    if e == 0:
        return False
    return all(r[1] == 1 for r in factor(e))


#################################################################
# Euler phi function
#################################################################
class Euler_Phi:
    r"""
    Return the value of the Euler phi function on the integer n. We
    defined this to be the number of positive integers <= n that are
    relatively prime to n. Thus if n<=0 then
    ``euler_phi(n)`` is defined and equals 0.

    INPUT:


    -  ``n`` - an integer


    EXAMPLES::

        sage: euler_phi(1)
        1
        sage: euler_phi(2)
        1
        sage: euler_phi(3)
        2
        sage: euler_phi(12)
        4
        sage: euler_phi(37)
        36

    Notice that euler_phi is defined to be 0 on negative numbers and
    0.

    ::

        sage: euler_phi(-1)
        0
        sage: euler_phi(0)
        0
        sage: type(euler_phi(0))
        <type 'sage.rings.integer.Integer'>

    We verify directly that the phi function is correct for 21.

    ::

        sage: euler_phi(21)
        12
        sage: [i for i in range(21) if gcd(21,i) == 1]
        [1, 2, 4, 5, 8, 10, 11, 13, 16, 17, 19, 20]

    The length of the list of integers 'i' in range(n) such that the
    gcd(i,n) == 1 equals euler_phi(n).

    ::

        sage: len([i for i in range(21) if gcd(21,i) == 1]) == euler_phi(21)
        True

    The phi function also has a special plotting method.

    ::

        sage: P = plot(euler_phi, -3, 71)

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: euler_phi(int8(37))
        36
        sage: from gmpy2 import mpz
        sage: euler_phi(mpz(37))
        36

    AUTHORS:

    - William Stein

    - Alex Clemesha (2006-01-10): some examples
    """
    def __repr__(self):
        """
        Return a string describing this class.

        EXAMPLES::

            sage: from sage.arith.misc import Euler_Phi
            sage: Euler_Phi().__repr__()
            'Number of positive integers <=n but relatively prime to n'
        """
        return "Number of positive integers <=n but relatively prime to n"

    def __call__(self, n):
        """
        Calls the euler_phi function.

        EXAMPLES::

            sage: from sage.arith.misc import Euler_Phi
            sage: Euler_Phi()(10)
            4
            sage: Euler_Phi()(720)
            192
        """
        if n<=0:
            return ZZ(0)
        if n<=2:
            return ZZ(1)
        return ZZ(pari(n).eulerphi())

    def plot(self, xmin=1, xmax=50, pointsize=30, rgbcolor=(0,0,1), join=True,
             **kwds):
        """
        Plot the Euler phi function.

        INPUT:


        -  ``xmin`` - default: 1

        -  ``xmax`` - default: 50

        -  ``pointsize`` - default: 30

        -  ``rgbcolor`` - default: (0,0,1)

        -  ``join`` - default: True; whether to join the
           points.

        -  ``**kwds`` - passed on

        EXAMPLES::

            sage: from sage.arith.misc import Euler_Phi
            sage: p = Euler_Phi().plot()
            sage: p.ymax()
            46.0
        """
        v = [(n, euler_phi(n)) for n in range(xmin, xmax + 1)]
        from sage.plot.all import list_plot
        P = list_plot(v, pointsize=pointsize, rgbcolor=rgbcolor, **kwds)
        if join:
            P += list_plot(v, plotjoined=True, rgbcolor=(0.7,0.7,0.7), **kwds)
        return P


euler_phi = Euler_Phi()


def crt(a, b, m=None, n=None):
    r"""
    Return a solution to a Chinese Remainder Theorem problem.

    INPUT:

    - ``a``, ``b`` - two residues (elements of some ring for which
      extended gcd is available), or two lists, one of residues and
      one of moduli.

    - ``m``, ``n`` - (default: ``None``) two moduli, or ``None``.

    OUTPUT:

    If ``m``, ``n`` are not ``None``, returns a solution `x` to the
    simultaneous congruences `x\equiv a \bmod m` and `x\equiv b \bmod
    n`, if one exists. By the Chinese Remainder Theorem, a solution to the
    simultaneous congruences exists if and only if
    `a\equiv b\pmod{\gcd(m,n)}`. The solution `x` is only well-defined modulo
    `\text{lcm}(m,n)`.

    If ``a`` and ``b`` are lists, returns a simultaneous solution to
    the congruences `x\equiv a_i\pmod{b_i}`, if one exists.

    .. SEEALSO::

        - :func:`CRT_list`

    EXAMPLES:

    Using ``crt`` by giving it pairs of residues and moduli::

        sage: crt(2, 1, 3, 5)
        11
        sage: crt(13, 20, 100, 301)
        28013
        sage: crt([2, 1], [3, 5])
        11
        sage: crt([13, 20], [100, 301])
        28013

    You can also use upper case::

        sage: c = CRT(2,3, 3, 5); c
        8
        sage: c % 3 == 2
        True
        sage: c % 5 == 3
        True

    Note that this also works for polynomial rings::

        sage: K.<a> = NumberField(x^3 - 7)
        sage: R.<y> = K[]
        sage: f = y^2 + 3
        sage: g = y^3 - 5
        sage: CRT(1,3,f,g)
        -3/26*y^4 + 5/26*y^3 + 15/26*y + 53/26
        sage: CRT(1,a,f,g)
        (-3/52*a + 3/52)*y^4 + (5/52*a - 5/52)*y^3 + (15/52*a - 15/52)*y + 27/52*a + 25/52

    You can also do this for any number of moduli::

        sage: K.<a> = NumberField(x^3 - 7)
        sage: R.<x> = K[]
        sage: CRT([], [])
        0
        sage: CRT([a], [x])
        a
        sage: f = x^2 + 3
        sage: g = x^3 - 5
        sage: h = x^5 + x^2 - 9
        sage: k = CRT([1, a, 3], [f, g, h]); k
        (127/26988*a - 5807/386828)*x^9 + (45/8996*a - 33677/1160484)*x^8 + (2/173*a - 6/173)*x^7 + (133/6747*a - 5373/96707)*x^6 + (-6/2249*a + 18584/290121)*x^5 + (-277/8996*a + 38847/386828)*x^4 + (-135/4498*a + 42673/193414)*x^3 + (-1005/8996*a + 470245/1160484)*x^2 + (-1215/8996*a + 141165/386828)*x + 621/8996*a + 836445/386828
        sage: k.mod(f)
        1
        sage: k.mod(g)
        a
        sage: k.mod(h)
        3

    If the moduli are not coprime, a solution may not exist::

        sage: crt(4,8,8,12)
        20
        sage: crt(4,6,8,12)
        Traceback (most recent call last):
        ...
        ValueError: No solution to crt problem since gcd(8,12) does not divide 4-6

        sage: x = polygen(QQ)
        sage: crt(2,3,x-1,x+1)
        -1/2*x + 5/2
        sage: crt(2,x,x^2-1,x^2+1)
        -1/2*x^3 + x^2 + 1/2*x + 1
        sage: crt(2,x,x^2-1,x^3-1)
        Traceback (most recent call last):
        ...
        ValueError: No solution to crt problem since gcd(x^2 - 1,x^3 - 1) does not divide 2-x

        sage: crt(int(2), int(3), int(7), int(11))
        58

    crt also work with numpy and gmpy2 numbers::

        sage: import numpy
        sage: crt(numpy.int8(2), numpy.int8(3), numpy.int8(7), numpy.int8(11))
        58
        sage: from gmpy2 import mpz
        sage: crt(mpz(2), mpz(3), mpz(7), mpz(11))
        58
        sage: crt(mpz(2), 3, mpz(7), numpy.int8(11))
        58
    """
    if isinstance(a, list):
        return CRT_list(a, b)

    try:
        f = (b-a).quo_rem
    except (TypeError, AttributeError):
        # Maybe there is no coercion between a and b.
        # Maybe (b-a) does not have a quo_rem attribute
        a = py_scalar_to_element(a)
        b = py_scalar_to_element(b)
        f = (b-a).quo_rem

    g, alpha, beta = XGCD(m, n)
    q, r = f(g)
    if r != 0:
        raise ValueError("No solution to crt problem since gcd(%s,%s) does not divide %s-%s" % (m, n, a, b))
    from sage.arith.functions import lcm

    x = a + q*alpha*py_scalar_to_element(m)
    return x % lcm(m, n)


CRT = crt


def CRT_list(v, moduli):
    r""" Given a list ``v`` of elements and a list of corresponding
    ``moduli``, find a single element that reduces to each element of
    ``v`` modulo the corresponding moduli.

    .. SEEALSO::

        - :func:`crt`

    EXAMPLES::

        sage: CRT_list([2,3,2], [3,5,7])
        23
        sage: x = polygen(QQ)
        sage: c = CRT_list([3], [x]); c
        3
        sage: c.parent()
        Univariate Polynomial Ring in x over Rational Field

    It also works if the moduli are not coprime::

        sage: CRT_list([32,2,2],[60,90,150])
        452

    But with non coprime moduli there is not always a solution::

        sage: CRT_list([32,2,1],[60,90,150])
        Traceback (most recent call last):
        ...
        ValueError: No solution to crt problem since gcd(180,150) does not divide 92-1

    The arguments must be lists::

        sage: CRT_list([1,2,3],"not a list")
        Traceback (most recent call last):
        ...
        ValueError: Arguments to CRT_list should be lists
        sage: CRT_list("not a list",[2,3])
        Traceback (most recent call last):
        ...
        ValueError: Arguments to CRT_list should be lists

    The list of moduli must have the same length as the list of elements::

        sage: CRT_list([1,2,3],[2,3,5])
        23
        sage: CRT_list([1,2,3],[2,3])
        Traceback (most recent call last):
        ...
        ValueError: Arguments to CRT_list should be lists of the same length
        sage: CRT_list([1,2,3],[2,3,5,7])
        Traceback (most recent call last):
        ...
        ValueError: Arguments to CRT_list should be lists of the same length

    TESTS::

        sage: CRT([32r,2r,2r],[60r,90r,150r])
        452
        sage: from numpy import int8
        sage: CRT_list([int8(2),int8(3),int8(2)], [int8(3),int8(5),int8(7)])
        23
        sage: from gmpy2 import mpz
        sage: CRT_list([mpz(2),mpz(3),mpz(2)], [mpz(3),mpz(5),mpz(7)])
        23
    """
    if not isinstance(v,list) or not isinstance(moduli,list):
        raise ValueError("Arguments to CRT_list should be lists")
    if len(v) != len(moduli):
        raise ValueError("Arguments to CRT_list should be lists of the same length")
    if len(v) == 0:
        return ZZ(0)
    if len(v) == 1:
        return moduli[0].parent()(v[0])
    x = v[0]
    m = moduli[0]
    from sage.arith.functions import lcm
    for i in range(1, len(v)):
        x = CRT(x,v[i],m,moduli[i])
        m = lcm(m,moduli[i])
    return x % m


def CRT_basis(moduli):
    r"""
    Return a CRT basis for the given moduli.

    INPUT:

    - ``moduli`` - list of pairwise coprime moduli `m` which admit an
       extended Euclidean algorithm

    OUTPUT:

    - a list of elements `a_i` of the same length as `m` such that
      `a_i` is congruent to 1 modulo `m_i` and to 0 modulo `m_j` for
      `j\not=i`.

    .. note::

       The pairwise coprimality of the input is not checked.

    EXAMPLES::

        sage: a1 = ZZ(mod(42,5))
        sage: a2 = ZZ(mod(42,13))
        sage: c1,c2 = CRT_basis([5,13])
        sage: mod(a1*c1+a2*c2,5*13)
        42

    A polynomial example::

        sage: x=polygen(QQ)
        sage: mods = [x,x^2+1,2*x-3]
        sage: b = CRT_basis(mods)
        sage: b
        [-2/3*x^3 + x^2 - 2/3*x + 1, 6/13*x^3 - x^2 + 6/13*x, 8/39*x^3 + 8/39*x]
        sage: [[bi % mj for mj in mods] for bi in b]
        [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
    """
    n = len(moduli)
    if n == 0:
        return []
    M = prod(moduli)
    return [((xgcd(m,M//m)[2])*(M//m))%M for m in moduli]


def CRT_vectors(X, moduli):
    r"""
    Vector form of the Chinese Remainder Theorem: given a list of integer
    vectors `v_i` and a list of coprime moduli `m_i`, find a vector `w` such
    that `w = v_i \pmod m_i` for all `i`. This is more efficient than applying
    :func:`CRT` to each entry.

    INPUT:

    -  ``X`` - list or tuple, consisting of lists/tuples/vectors/etc of
       integers of the same length
    -  ``moduli`` - list of len(X) moduli

    OUTPUT:

    -  ``list`` - application of CRT componentwise.

    EXAMPLES::

        sage: CRT_vectors([[3,5,7],[3,5,11]], [2,3])
        [3, 5, 5]

        sage: CRT_vectors([vector(ZZ, [2,3,1]), Sequence([1,7,8],ZZ)], [8,9])
        [10, 43, 17]
    """
    # First find the CRT basis:
    if len(X) == 0 or len(X[0]) == 0:
        return []
    n = len(X)
    if n != len(moduli):
        raise ValueError("number of moduli must equal length of X")
    a = CRT_basis(moduli)
    modulus = prod(moduli)
    return [sum(a[i]*X[i][j] for i in range(n)) % modulus for j in range(len(X[0]))]


def binomial(x, m, **kwds):
    r"""
    Return the binomial coefficient

    .. MATH::

        \binom{x}{m} = x (x-1) \cdots (x-m+1) / m!

    which is defined for `m \in \ZZ` and any
    `x`. We extend this definition to include cases when
    `x-m` is an integer but `m` is not by

    .. MATH::

        \binom{x}{m} = \binom{x}{x-m}

    If `m < 0`, return `0`.

    INPUT:

    -  ``x``, ``m`` - numbers or symbolic expressions. Either ``m``
       or ``x-m`` must be an integer.

    OUTPUT: number or symbolic expression (if input is symbolic)

    EXAMPLES::

        sage: from sage.arith.misc import binomial
        sage: binomial(5,2)
        10
        sage: binomial(2,0)
        1
        sage: binomial(1/2, 0)
        1
        sage: binomial(3,-1)
        0
        sage: binomial(20,10)
        184756
        sage: binomial(-2, 5)
        -6
        sage: binomial(-5, -2)
        0
        sage: binomial(RealField()('2.5'), 2)
        1.87500000000000
        sage: n=var('n'); binomial(n,2)
        1/2*(n - 1)*n
        sage: n=var('n'); binomial(n,n)
        1
        sage: n=var('n'); binomial(n,n-1)
        n
        sage: binomial(2^100, 2^100)
        1

        sage: x = polygen(ZZ)
        sage: binomial(x, 3)
        1/6*x^3 - 1/2*x^2 + 1/3*x
        sage: binomial(x, x-3)
        1/6*x^3 - 1/2*x^2 + 1/3*x

    If `x \in \ZZ`, there is an optional 'algorithm' parameter, which
    can be 'mpir' (faster for small values) or 'pari' (faster for
    large values)::

        sage: a = binomial(100, 45, algorithm='mpir')
        sage: b = binomial(100, 45, algorithm='pari')
        sage: a == b
        True

    TESTS:

    We test that certain binomials are very fast (this should be
    instant) -- see :trac:`3309`::

        sage: a = binomial(RR(1140000.78), 23310000)

    We test conversion of arguments to Integers -- see :trac:`6870`::

        sage: binomial(1/2,1/1)
        1/2
        sage: binomial(10^20+1/1,10^20)
        100000000000000000001
        sage: binomial(SR(10**7),10**7)
        1
        sage: binomial(3/2,SR(1/1))
        3/2

    Some floating point cases -- see :trac:`7562`, :trac:`9633`, and
    :trac:`12448`::

        sage: binomial(1.,3)
        0.000000000000000
        sage: binomial(-2.,3)
        -4.00000000000000
        sage: binomial(0.5r, 5)
        0.02734375
        sage: a = binomial(float(1001), float(1)); a
        1001.0
        sage: type(a)
        <... 'float'>
        sage: binomial(float(1000), 1001)
        0.0

    Test more output types::

        sage: type(binomial(5r, 2))
        <... 'int'>
        sage: type(binomial(5, 2r))
        <type 'sage.rings.integer.Integer'>

        sage: type(binomial(5.0r, 2))
        <... 'float'>

        sage: type(binomial(5/1, 2))
        <type 'sage.rings.rational.Rational'>

        sage: R = Integers(11)
        sage: b = binomial(R(7), R(3))
        sage: b
        2
        sage: b.parent()
        Ring of integers modulo 11

    Test symbolic and uni/multivariate polynomials::

        sage: x = polygen(ZZ)
        sage: binomial(x, 3)
        1/6*x^3 - 1/2*x^2 + 1/3*x
        sage: binomial(x, 3).parent()
        Univariate Polynomial Ring in x over Rational Field

        sage: K.<x,y> = Integers(7)[]
        sage: binomial(y,3)
        6*y^3 + 3*y^2 + 5*y
        sage: binomial(y,3).parent()
        Multivariate Polynomial Ring in x, y over Ring of integers modulo 7

        sage: n = var('n')
        sage: binomial(n,2)
        1/2*(n - 1)*n

    Invalid inputs::

        sage: x = polygen(ZZ)
        sage: binomial(x, x^2)
        Traceback (most recent call last):
        ...
        TypeError: either m or x-m must be an integer

        sage: k, i = var('k,i')
        sage: binomial(k,i)
        Traceback (most recent call last):
        ...
        TypeError: either m or x-m must be an integer

        sage: R6 = Zmod(6)
        sage: binomial(R6(5), 2)
        Traceback (most recent call last):
        ...
        ZeroDivisionError: factorial(2) not invertible in Ring of integers modulo 6

        sage: R7 = Zmod(7)
        sage: binomial(R7(10), 7)
        Traceback (most recent call last):
        ...
        ZeroDivisionError: factorial(7) not invertible in Ring of integers modulo 7

    The last two examples failed to execute since `2!` and `7!` are respectively
    not invertible in `\ZZ/6\ZZ` and `\ZZ/7\ZZ`. One can check that there
    is no well defined value for that binomial coefficient in the quotient::

        sage: R6(binomial(5,2))
        4
        sage: R6(binomial(5+6,2))
        1

        sage: R7(binomial(3, 7))
        0
        sage: R7(binomial(10, 7))
        1
        sage: R7(binomial(17, 7))
        2

    For symbolic manipulation, you should use the function
    :func:`~sage.functions.other.binomial` from the module
    :mod:`sage.functions.other`::

        sage: from sage.functions.other import binomial
        sage: binomial(k, i)
        binomial(k, i)

    binomial support numpy and gmpy2 parameters::

        sage: from sage.arith.misc import binomial
        sage: import numpy
        sage: binomial(numpy.int32(20), numpy.int32(10))
        184756
        sage: import gmpy2
        sage: binomial(gmpy2.mpz(20), gmpy2.mpz(10))
        mpz(184756)
    """
    try:
        m = ZZ(m)
    except TypeError:
        try:
            m = ZZ(x-m)
        except TypeError:
            raise TypeError("either m or x-m must be an integer")

    P = parent(x)
    x = py_scalar_to_element(x)

    # case 1: native binomial implemented on x
    try:
        return P(x.binomial(m, **kwds))
    except (AttributeError,TypeError):
        pass

    # case 2: conversion to integers
    try:
        x = ZZ(x)
    except TypeError:
        pass
    else:
        # Check invertibility of factorial(m) in P
        try:
            c = P.characteristic()
        except AttributeError:
            # Assume that P has characteristic zero (can be int, float, ...)
            pass
        else:
            if c > 0 and any(c.gcd(k) > 1 for k in range(2, m+1)):
                raise ZeroDivisionError("factorial({}) not invertible in {}".format(m, P))
        return P(x.binomial(m, **kwds))

    # case 3: rational, real numbers, complex numbers -> use pari
    if isinstance(x, (Rational, RealNumber, ComplexNumber)):
        return P(x.__pari__().binomial(m))

    # case 4: naive method
    if m < ZZ.zero():
        return P(0)
    return P(prod(x - i for i in range(m))) / m.factorial()


def multinomial(*ks):
    r"""
    Return the multinomial coefficient

    INPUT:

    - An arbitrary number of integer arguments `k_1,\dots,k_n`
    - An iterable (e.g. a list) of integers `[k_1,\dots,k_n]`

    OUTPUT:

    Return the integer:

    .. MATH::

           \binom{k_1 + \cdots + k_n}{k_1, \cdots, k_n}
           =\frac{\left(\sum_{i=1}^n k_i\right)!}{\prod_{i=1}^n k_i!}
           = \prod_{i=1}^n \binom{\sum_{j=1}^i k_j}{k_i}

    EXAMPLES::

        sage: multinomial(0, 0, 2, 1, 0, 0)
        3
        sage: multinomial([0, 0, 2, 1, 0, 0])
        3
        sage: multinomial(3, 2)
        10
        sage: multinomial(2^30, 2, 1)
        618970023101454657175683075
        sage: multinomial([2^30, 2, 1])
        618970023101454657175683075
        sage: multinomial(Composition([1, 3]))
        4
        sage: multinomial(Partition([4, 2]))
        15

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: multinomial(int8(3), int8(2))
        10
        sage: from gmpy2 import mpz
        sage: multinomial(mpz(3), mpz(2))
        mpz(10)

    AUTHORS:

    - Gabriel Ebner
    """
    if isinstance(ks[0], Iterable):
        if len(ks) > 1:
            raise ValueError("multinomial takes only one iterable argument")
        ks = ks[0]

    s, c = 0, 1
    for k in ks:
        s += k
        c *= binomial(s, k)
    return c


def binomial_coefficients(n):
    r"""
    Return a dictionary containing pairs
    `\{(k_1,k_2) : C_{k,n}\}` where `C_{k_n}` are
    binomial coefficients and `n = k_1 + k_2`.

    INPUT:


    -  ``n`` - an integer


    OUTPUT: dict

    EXAMPLES::

        sage: sorted(binomial_coefficients(3).items())
        [((0, 3), 1), ((1, 2), 3), ((2, 1), 3), ((3, 0), 1)]

    Notice the coefficients above are the same as below::

        sage: R.<x,y> = QQ[]
        sage: (x+y)^3
        x^3 + 3*x^2*y + 3*x*y^2 + y^3

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: sorted(binomial_coefficients(int8(3)).items())
        [((0, 3), 1), ((1, 2), 3), ((2, 1), 3), ((3, 0), 1)]
        sage: from gmpy2 import mpz
        sage: sorted(binomial_coefficients(mpz(3)).items())
        [((0, 3), 1), ((1, 2), 3), ((2, 1), 3), ((3, 0), 1)]

    AUTHORS:

    - Fredrik Johansson
    """
    n = py_scalar_to_element(n)
    d = {(0, n):1, (n, 0):1}
    a = 1
    for k in range(1, n // 2 + 1):
        a = (a * (n-k+1))//k
        d[k, n-k] = d[n-k, k] = a
    return d


def multinomial_coefficients(m, n):
    r"""
    Return a dictionary containing pairs
    `\{(k_1, k_2, ..., k_m) : C_{k, n}\}` where
    `C_{k, n}` are multinomial coefficients such that
    `n = k_1 + k_2 + ...+ k_m`.

    INPUT:

    -  ``m`` - integer
    -  ``n`` - integer

    OUTPUT: dict

    EXAMPLES::

        sage: sorted(multinomial_coefficients(2, 5).items())
        [((0, 5), 1), ((1, 4), 5), ((2, 3), 10), ((3, 2), 10), ((4, 1), 5), ((5, 0), 1)]

    Notice that these are the coefficients of `(x+y)^5`::

        sage: R.<x,y> = QQ[]
        sage: (x+y)^5
        x^5 + 5*x^4*y + 10*x^3*y^2 + 10*x^2*y^3 + 5*x*y^4 + y^5

    ::

        sage: sorted(multinomial_coefficients(3, 2).items())
        [((0, 0, 2), 1), ((0, 1, 1), 2), ((0, 2, 0), 1), ((1, 0, 1), 2), ((1, 1, 0), 2), ((2, 0, 0), 1)]

    ALGORITHM: The algorithm we implement for computing the multinomial
    coefficients is based on the following result:

    .. MATH::

        \binom{n}{k_1, \cdots, k_m} =
        \frac{k_1+1}{n-k_1}\sum_{i=2}^m \binom{n}{k_1+1, \cdots, k_i-1, \cdots}

    e.g.::

        sage: k = (2, 4, 1, 0, 2, 6, 0, 0, 3, 5, 7, 1) # random value
        sage: n = sum(k)
        sage: s = 0
        sage: for i in range(1, len(k)):
        ....:     ki = list(k)
        ....:     ki[0] += 1
        ....:     ki[i] -= 1
        ....:     s += multinomial(n, *ki)
        sage: multinomial(n, *k) == (k[0] + 1) / (n - k[0]) * s
        True

    TESTS::

        sage: multinomial_coefficients(0, 0)
        {(): 1}
        sage: multinomial_coefficients(0, 3)
        {}
        sage: from numpy import int8
        sage: sorted(multinomial_coefficients(int8(2), int8(5)).items())
        [((0, 5), 1), ((1, 4), 5), ((2, 3), 10), ((3, 2), 10), ((4, 1), 5), ((5, 0), 1)]
        sage: from gmpy2 import mpz
        sage: sorted(multinomial_coefficients(mpz(2), mpz(5)).items())
        [((0, 5), 1), ((1, 4), 5), ((2, 3), 10), ((3, 2), 10), ((4, 1), 5), ((5, 0), 1)]
    """
    if not m:
        if n:
            return {}
        else:
            return {(): 1}

    m = py_scalar_to_element(m)
    n = py_scalar_to_element(n)

    if m == 2:
        return binomial_coefficients(n)
    t = [n] + [0] * (m - 1)
    r = {tuple(t): 1}
    if n:
        j = 0 # j will be the leftmost nonzero position
    else:
        j = m
    # enumerate tuples in co-lex order
    while j < m - 1:
        # compute next tuple
        tj = t[j]
        if j:
            t[j] = 0
            t[0] = tj
        if tj > 1:
            t[j + 1] += 1
            j = 0
            start = 1
            v = 0
        else:
            j += 1
            start = j + 1
            v = r[tuple(t)]
            t[j] += 1
        # compute the value
        # NB: the initialization of v was done above
        for k in range(start, m):
            if t[k]:
                t[k] -= 1
                v += r[tuple(t)]
                t[k] += 1
        t[0] -= 1
        r[tuple(t)] = (v * tj) // (n - t[0])
    return r


def kronecker_symbol(x,y):
    """
    The Kronecker symbol `(x|y)`.

    INPUT:

    - ``x`` -- integer

    - ``y`` -- integer

    OUTPUT:

    - an integer

    EXAMPLES::

        sage: kronecker_symbol(13,21)
        -1
        sage: kronecker_symbol(101,4)
        1

    This is also available as :func:`kronecker`::

        sage: kronecker(3,5)
        -1
        sage: kronecker(3,15)
        0
        sage: kronecker(2,15)
        1
        sage: kronecker(-2,15)
        -1
        sage: kronecker(2/3,5)
        1

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: kronecker_symbol(int8(13),int8(21))
        -1
        sage: from gmpy2 import mpz
        sage: kronecker_symbol(mpz(13),mpz(21))
        -1
    """
    x = QQ(x).numerator() * QQ(x).denominator()
    return ZZ(x.kronecker(y))

kronecker = kronecker_symbol


def legendre_symbol(x,p):
    r"""
    The Legendre symbol `(x|p)`, for `p` prime.

    .. note::

       The :func:`kronecker_symbol` command extends the Legendre
       symbol to composite moduli and `p=2`.

    INPUT:


    -  ``x`` - integer

    -  ``p`` - an odd prime number


    EXAMPLES::

        sage: legendre_symbol(2,3)
        -1
        sage: legendre_symbol(1,3)
        1
        sage: legendre_symbol(1,2)
        Traceback (most recent call last):
        ...
        ValueError: p must be odd
        sage: legendre_symbol(2,15)
        Traceback (most recent call last):
        ...
        ValueError: p must be a prime
        sage: kronecker_symbol(2,15)
        1
        sage: legendre_symbol(2/3,7)
        -1

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: legendre_symbol(int8(2),int8(3))
        -1
        sage: from gmpy2 import mpz
        sage: legendre_symbol(mpz(2),mpz(3))
        -1
    """
    x = QQ(x).numerator() * QQ(x).denominator()
    p = ZZ(p)
    if not p.is_prime():
        raise ValueError("p must be a prime")
    if p == 2:
        raise ValueError("p must be odd")
    return x.kronecker(p)


def jacobi_symbol(a, b):
    r"""
    The Jacobi symbol of integers a and b, where b is odd.

    .. note::

       The :func:`kronecker_symbol` command extends the Jacobi
       symbol to all integers b.

    If

    `b = p_1^{e_1} * ... * p_r^{e_r}`

    then

    `(a|b) = (a|p_1)^{e_1} ... (a|p_r)^{e_r}`

    where `(a|p_j)` are Legendre Symbols.



    INPUT:

    -  ``a`` - an integer

    -  ``b`` - an odd integer

    EXAMPLES::

        sage: jacobi_symbol(10,777)
        -1
        sage: jacobi_symbol(10,5)
        0
        sage: jacobi_symbol(10,2)
        Traceback (most recent call last):
        ...
        ValueError: second input must be odd, 2 is not odd

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: jacobi_symbol(int16(10),int16(777))
        -1
        sage: from gmpy2 import mpz
        sage: jacobi_symbol(mpz(10),mpz(777))
        -1
    """
    if b % 2 == 0:
        raise ValueError("second input must be odd, %s is not odd" % b)

    return kronecker_symbol(a, b)


def primitive_root(n, check=True):
    """
    Return a positive integer that generates the multiplicative group
    of integers modulo `n`, if one exists; otherwise, raise a
    ``ValueError``.

    A primitive root exists if `n=4` or `n=p^k` or `n=2p^k`, where `p`
    is an odd prime and `k` is a nonnegative number.

    INPUT:

    - ``n`` -- a non-zero integer
    - ``check`` -- bool (default: True); if False, then `n` is assumed
      to be a positive integer possessing a primitive root, and behavior
      is undefined otherwise.

    OUTPUT:

    A primitive root of `n`. If `n` is prime, this is the smallest
    primitive root.

    EXAMPLES::

        sage: primitive_root(23)
        5
        sage: primitive_root(-46)
        5
        sage: primitive_root(25)
        2
        sage: print([primitive_root(p) for p in primes(100)])
        [1, 2, 2, 3, 2, 2, 3, 2, 5, 2, 3, 2, 6, 3, 5, 2, 2, 2, 2, 7, 5, 3, 2, 3, 5]
        sage: primitive_root(8)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root

    .. NOTE::

        It takes extra work to check if `n` has a primitive root; to
        avoid this, use ``check=False``, which may slightly speed things
        up (but could also result in undefined behavior).  For example,
        the second call below is an order of magnitude faster than the
        first:

    ::

        sage: n = 10^50 + 151   # a prime
        sage: primitive_root(n)
        11
        sage: primitive_root(n, check=False)
        11

    TESTS:

    Various special cases::

        sage: primitive_root(-1)
        0
        sage: primitive_root(0)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root
        sage: primitive_root(1)
        0
        sage: primitive_root(2)
        1
        sage: primitive_root(3)
        2
        sage: primitive_root(4)
        3

    We test that various numbers without primitive roots give
    an error - see :trac:`10836`::

        sage: primitive_root(15)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root
        sage: primitive_root(16)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root
        sage: primitive_root(1729)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root
        sage: primitive_root(4*7^8)
        Traceback (most recent call last):
        ...
        ValueError: no primitive root

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: primitive_root(int8(-46))
        5
        sage: from gmpy2 import mpz
        sage: primitive_root(mpz(-46))
        5
    """
    if not check:
        return ZZ(pari(n).znprimroot())
    n = ZZ(n).abs()
    if n <= 4:
        if n:
            # n-1 is a primitive root for n in {1,2,3,4}
            return n-1
    elif n%2: # n odd
        if n.is_prime_power():
            return ZZ(pari(n).znprimroot())
    else:   # n even
        m = n // 2
        if m%2 and m.is_prime_power():
            return ZZ(pari(n).znprimroot())
    raise ValueError("no primitive root")


def nth_prime(n):
    """

    Return the n-th prime number (1-indexed, so that 2 is the 1st prime.)

    INPUT:

    - ``n`` -- a positive integer

    OUTPUT:

    -  the n-th prime number

    EXAMPLES::

        sage: nth_prime(3)
        5
        sage: nth_prime(10)
        29
        sage: nth_prime(10^7)
        179424673

    ::

        sage: nth_prime(0)
        Traceback (most recent call last):
        ...
        ValueError: nth prime meaningless for non-positive n (=0)

    TESTS::

        sage: all(prime_pi(nth_prime(j)) == j for j in range(1, 1000, 10))
        True
        sage: from numpy import int8
        sage: nth_prime(int8(10))
        29
        sage: from gmpy2 import mpz
        sage: nth_prime(mpz(10))
        29
    """
    if n <= 0:
        raise ValueError("nth prime meaningless for non-positive n (=%s)" % n)
    return ZZ(pari.prime(n))


def quadratic_residues(n):
    r"""
    Return a sorted list of all squares modulo the integer `n`
    in the range `0\leq x < |n|`.

    EXAMPLES::

        sage: quadratic_residues(11)
        [0, 1, 3, 4, 5, 9]
        sage: quadratic_residues(1)
        [0]
        sage: quadratic_residues(2)
        [0, 1]
        sage: quadratic_residues(8)
        [0, 1, 4]
        sage: quadratic_residues(-10)
        [0, 1, 4, 5, 6, 9]
        sage: v = quadratic_residues(1000); len(v)
        159

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: quadratic_residues(int8(11))
        [0, 1, 3, 4, 5, 9]
        sage: from gmpy2 import mpz
        sage: quadratic_residues(mpz(11))
        [0, 1, 3, 4, 5, 9]
    """
    n = abs(int(n))
    return sorted(set(ZZ((a*a) % n) for a in range(n // 2 + 1)))


class Moebius:
    r"""
    Return the value of the Möbius function of abs(n), where n is an
    integer.

    DEFINITION: `\mu(n)` is 0 if `n` is not square
    free, and otherwise equals `(-1)^r`, where `n` has
    `r` distinct prime factors.

    For simplicity, if `n=0` we define `\mu(n) = 0`.

    IMPLEMENTATION: Factors or - for integers - uses the PARI C
    library.

    INPUT:


    -  ``n`` - anything that can be factored.


    OUTPUT: 0, 1, or -1

    EXAMPLES::

        sage: moebius(-5)
        -1
        sage: moebius(9)
        0
        sage: moebius(12)
        0
        sage: moebius(-35)
        1
        sage: moebius(-1)
        1
        sage: moebius(7)
        -1

    ::

        sage: moebius(0)   # potentially nonstandard!
        0

    The moebius function even makes sense for non-integer inputs.

    ::

        sage: x = GF(7)['x'].0
        sage: moebius(x+2)
        -1

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: moebius(int8(-5))
        -1
        sage: from gmpy2 import mpz
        sage: moebius(mpz(-5))
        -1
    """
    def __call__(self, n):
        """
        EXAMPLES::

            sage: from sage.arith.misc import Moebius
            sage: Moebius().__call__(7)
            -1
        """
        n = py_scalar_to_element(n)

        if not isinstance(n, Integer):
            # Use a generic algorithm.
            if n < 0:
                n = -n
            F = factor(n)
            for _, e in F:
                if e >= 2:
                    return 0
            return (-1)**len(F)

        # Use fast PARI algorithm
        if n == 0:
            return ZZ.zero()
        return ZZ(pari(n).moebius())


    def __repr__(self):
        """
        Return a description of this function.

        EXAMPLES::

            sage: from sage.arith.misc import Moebius
            sage: q = Moebius()
            sage: q.__repr__()
            'The Moebius function'
        """
        return "The Moebius function"

    def plot(self, xmin=0, xmax=50, pointsize=30, rgbcolor=(0,0,1), join=True,
             **kwds):
        """
        Plot the Möbius function.

        INPUT:


        -  ``xmin`` - default: 0

        -  ``xmax`` - default: 50

        -  ``pointsize`` - default: 30

        -  ``rgbcolor`` - default: (0,0,1)

        -  ``join`` - default: True; whether to join the points
           (very helpful in seeing their order).

        -  ``**kwds`` - passed on

        EXAMPLES::

            sage: from sage.arith.misc import Moebius
            sage: p = Moebius().plot()
            sage: p.ymax()
            1.0
        """
        values = self.range(xmin, xmax + 1)
        v = [(n,values[n-xmin]) for n in range(xmin,xmax + 1)]
        from sage.plot.all import list_plot
        P = list_plot(v, pointsize=pointsize, rgbcolor=rgbcolor, **kwds)
        if join:
            P += list_plot(v, plotjoined=True, rgbcolor=(0.7,0.7,0.7), **kwds)
        return P

    def range(self, start, stop=None, step=None):
        """
        Return the Möbius function evaluated at the given range of values,
        i.e., the image of the list range(start, stop, step) under the
        Möbius function.

        This is much faster than directly computing all these values with a
        list comprehension.

        EXAMPLES::

            sage: v = moebius.range(-10,10); v
            [1, 0, 0, -1, 1, -1, 0, -1, -1, 1, 0, 1, -1, -1, 0, -1, 1, -1, 0, 0]
            sage: v == [moebius(n) for n in range(-10,10)]
            True
            sage: v = moebius.range(-1000, 2000, 4)
            sage: v == [moebius(n) for n in range(-1000,2000, 4)]
            True
        """
        if stop is None:
            start, stop = 1, int(start)
        else:
            start = int(start)
            stop = int(stop)
        if step is None:
            step = 1
        else:
            step = int(step)

        if start <= 0 and 0 < stop and start % step == 0:
            return self.range(start, 0, step) + [ZZ.zero()] +\
                   self.range(step, stop, step)

        if step == 1:
            v = pari('vector(%s, i, moebius(i-1+%s))'%(
                stop-start, start))
        else:
            n = len(range(start, stop, step)) # stupid
            v = pari('vector(%s, i, moebius(%s*(i-1) + %s))'%(
                n, step, start))
        return [Integer(x) for x in v]

moebius = Moebius()


## Note: farey, convergent, continued_fraction_list and convergents have been moved to
## sage.rings.continued_fraction

def continuant(v, n=None):
    r"""
    Function returns the continuant of the sequence `v` (list
    or tuple).

    Definition: see Graham, Knuth and Patashnik, *Concrete Mathematics*,
    section 6.7: Continuants. The continuant is defined by

    - `K_0() = 1`
    - `K_1(x_1) = x_1`
    - `K_n(x_1, \cdots, x_n) = K_{n-1}(x_n, \cdots x_{n-1})x_n + K_{n-2}(x_1,  \cdots, x_{n-2})`

    If ``n = None`` or ``n > len(v)`` the default
    ``n = len(v)`` is used.

    INPUT:

    -  ``v`` - list or tuple of elements of a ring
    -  ``n`` - optional integer

    OUTPUT: element of ring (integer, polynomial, etcetera).

    EXAMPLES::

        sage: continuant([1,2,3])
        10
        sage: p = continuant([2, 1, 2, 1, 1, 4, 1, 1, 6, 1, 1, 8, 1, 1, 10])
        sage: q = continuant([1, 2, 1, 1, 4, 1, 1, 6, 1, 1, 8, 1, 1, 10])
        sage: p/q
        517656/190435
        sage: continued_fraction([2, 1, 2, 1, 1, 4, 1, 1, 6, 1, 1, 8, 1, 1, 10]).convergent(14)
        517656/190435
        sage: x = PolynomialRing(RationalField(),'x',5).gens()
        sage: continuant(x)
        x0*x1*x2*x3*x4 + x0*x1*x2 + x0*x1*x4 + x0*x3*x4 + x2*x3*x4 + x0 + x2 + x4
        sage: continuant(x, 3)
        x0*x1*x2 + x0 + x2
        sage: continuant(x,2)
        x0*x1 + 1

    We verify the identity

    .. MATH::

        K_n(z,z,\cdots,z) = \sum_{k=0}^n \binom{n-k}{k} z^{n-2k}

    for `n = 6` using polynomial arithmetic::

        sage: z = QQ['z'].0
        sage: continuant((z,z,z,z,z,z,z,z,z,z,z,z,z,z,z),6)
        z^6 + 5*z^4 + 6*z^2 + 1

        sage: continuant(9)
        Traceback (most recent call last):
        ...
        TypeError: object of type 'sage.rings.integer.Integer' has no len()

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: continuant([int8(1),int8(2),int8(3)])
        10
        sage: from gmpy2 import mpz
        sage: continuant([mpz(1),mpz(2),mpz(3)])
        mpz(10)

    AUTHORS:

    - Jaap Spies (2007-02-06)
    """
    m = len(v)
    if n is None or m < n:
        n = m
    if n == 0:
        return 1
    if n == 1:
        return v[0]
    a, b = 1, v[0]
    for k in range(1,n):
        a, b = b, a + b*v[k]
    return b


def number_of_divisors(n):
    """
    Return the number of divisors of the integer n.

    INPUT:

    - ``n`` - a nonzero integer

    OUTPUT:

    - an integer, the number of divisors of n

    EXAMPLES::

        sage: number_of_divisors(100)
        9
        sage: number_of_divisors(-720)
        30

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: number_of_divisors(int8(100))
        9
        sage: from gmpy2 import mpz
        sage: number_of_divisors(mpz(100))
        9
    """
    m = ZZ(n)
    if m.is_zero():
        raise ValueError("input must be nonzero")
    return ZZ(pari(m).numdiv())



def hilbert_symbol(a, b, p, algorithm="pari"):
    """
    Return 1 if `ax^2 + by^2` `p`-adically represents
    a nonzero square, otherwise returns `-1`. If either a or b
    is 0, returns 0.

    INPUT:


    -  ``a, b`` - integers

    -  ``p`` - integer; either prime or -1 (which
       represents the archimedean place)

    -  ``algorithm`` - string

       -  ``'pari'`` - (default) use the PARI C library

       -  ``'direct'`` - use a Python implementation

       -  ``'all'`` - use both PARI and direct and check that
          the results agree, then return the common answer


    OUTPUT: integer (0, -1, or 1)

    EXAMPLES::

        sage: hilbert_symbol (-1, -1, -1, algorithm='all')
        -1
        sage: hilbert_symbol (2,3, 5, algorithm='all')
        1
        sage: hilbert_symbol (4, 3, 5, algorithm='all')
        1
        sage: hilbert_symbol (0, 3, 5, algorithm='all')
        0
        sage: hilbert_symbol (-1, -1, 2, algorithm='all')
        -1
        sage: hilbert_symbol (1, -1, 2, algorithm='all')
        1
        sage: hilbert_symbol (3, -1, 2, algorithm='all')
        -1

        sage: hilbert_symbol(QQ(-1)/QQ(4), -1, 2) == -1
        True
        sage: hilbert_symbol(QQ(-1)/QQ(4), -1, 3) == 1
        True

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: hilbert_symbol(int8(2),int8(3),int8(5),algorithm='all')
        1
        sage: from gmpy2 import mpz
        sage: hilbert_symbol(mpz(2),mpz(3),mpz(5),algorithm='all')
        1

    AUTHORS:

    - William Stein and David Kohel (2006-01-05)
    """
    p = ZZ(p)
    if p != -1 and not p.is_prime():
        raise ValueError("p must be prime or -1")
    a = QQ(a).numerator() * QQ(a).denominator()
    b = QQ(b).numerator() * QQ(b).denominator()

    if algorithm == "pari":
        if p == -1:
            p = 0
        return ZZ(pari(a).hilbert(b, p))

    elif algorithm == 'direct':
        if a == 0 or b == 0:
            return ZZ(0)

        p = ZZ(p)
        one = ZZ(1)

        if p != -1:
            p_sqr = p**2
            while a % p_sqr == 0:
                a //= p_sqr
            while b % p_sqr == 0:
                b //= p_sqr

        if p != 2 and True in (kronecker(x, p) == 1 for x in (a, b, a + b)):
            return one
        if a % p == 0:
            if b % p == 0:
                return hilbert_symbol(p,-(b//p),p)*hilbert_symbol(a//p,b,p)
            elif p == 2 and (b%4) == 3:
                if kronecker(a+b,p) == -1:
                    return -one
            elif kronecker(b,p) == -1:
                return -one
        elif b%p == 0:
            if p == 2 and (a%4) == 3:
                if kronecker(a+b,p) == -1:
                    return -one
            elif kronecker(a,p) == -1:
                return -one
        elif p == 2 and (a%4) == 3 and (b%4) == 3:
            return -one
        return one
    elif algorithm == 'all':
        ans_pari = hilbert_symbol(a,b,p,algorithm='pari')
        ans_direct = hilbert_symbol(a,b,p,algorithm='direct')
        if ans_pari != ans_direct:
            raise RuntimeError("There is a bug in hilbert_symbol; two ways of computing the Hilbert symbol (%s,%s)_%s disagree"%(a,b,p))
        return ans_pari
    else:
        raise ValueError("Algorithm %s not defined"%algorithm)

def hilbert_conductor(a, b):
    r"""
    This is the product of all (finite) primes where the Hilbert symbol is -1.

    What is the same, this is the (reduced) discriminant of the quaternion
    algebra `(a,b)` over `\QQ`.

    INPUT:

    - ``a``, ``b`` -- integers

    OUTPUT:

    - squarefree positive integer

    EXAMPLES::

        sage: hilbert_conductor(-1, -1)
        2
        sage: hilbert_conductor(-1, -11)
        11
        sage: hilbert_conductor(-2, -5)
        5
        sage: hilbert_conductor(-3, -17)
        17

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: hilbert_conductor(int8(-3), int8(-17))
        17
        sage: from gmpy2 import mpz
        sage: hilbert_conductor(mpz(-3), mpz(-17))
        17

    AUTHOR:

    - Gonzalo Tornaria (2009-03-02)
    """
    a, b = ZZ(a), ZZ(b)
    d = ZZ(1)
    for p in set().union([2], prime_divisors(a), prime_divisors(b)):
        if hilbert_symbol(a, b, p) == -1:
            d *= p
    return d


def hilbert_conductor_inverse(d):
    r"""
    Finds a pair of integers `(a,b)` such that ``hilbert_conductor(a,b) == d``.

    The quaternion algebra `(a,b)` over `\QQ` will then have (reduced)
    discriminant `d`.

    INPUT:

    - ``d`` -- square-free positive integer

    OUTPUT: pair of integers

    EXAMPLES::

        sage: hilbert_conductor_inverse(2)
        (-1, -1)
        sage: hilbert_conductor_inverse(3)
        (-1, -3)
        sage: hilbert_conductor_inverse(6)
        (-1, 3)
        sage: hilbert_conductor_inverse(30)
        (-3, -10)
        sage: hilbert_conductor_inverse(4)
        Traceback (most recent call last):
        ...
        ValueError: d needs to be squarefree
        sage: hilbert_conductor_inverse(-1)
        Traceback (most recent call last):
        ...
        ValueError: d needs to be positive

    AUTHOR:

    - Gonzalo Tornaria (2009-03-02)

    TESTS::

        sage: for i in range(100):
        ....:     d = ZZ.random_element(2**32).squarefree_part()
        ....:     if hilbert_conductor(*hilbert_conductor_inverse(d)) != d:
        ....:         print("hilbert_conductor_inverse failed for d = {}".format(d))

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: hilbert_conductor_inverse(int8(30))
        (-3, -10)
        sage: from gmpy2 import mpz
        sage: hilbert_conductor_inverse(mpz(30))
        (-3, -10)
    """
    Z = ZZ
    d = Z(d)
    if d <= 0:
        raise ValueError("d needs to be positive")
    if d == 1:
        return (Z(-1), Z(1))
    if d == 2:
        return (Z(-1), Z(-1))
    if d.is_prime():
        if d%4 == 3:
            return (Z(-1), -d)
        if d%8 == 5:
            return (Z(-2), -d)
        q = 3
        while q%4 != 3 or kronecker_symbol(d,q) != -1:
            q = next_prime(q)
        return (Z(-q), -d)
    else:
        mo = moebius(d)
        if mo == 0:
            raise ValueError("d needs to be squarefree")
        if d % 2 == 0 and mo*d % 16 != 2:
            dd = mo * d / 2
        else:
            dd = mo * d
        q = 1
        while hilbert_conductor(-q, dd) != d:
            q += 1
        if dd%q == 0:
            dd /= q
        return (Z(-q), Z(dd))


##############################################################################
##  falling and rising factorials
##  By Jaap Spies
##
##       Copyright (C) 2006 Jaap Spies <j.spies@hccnet.nl>
##      Copyright (C) 2006 William Stein <wstein@gmail.com>
##
## Distributed under the terms of the GNU General Public License (GPL)
##                  http://www.gnu.org/licenses/
##############################################################################


def falling_factorial(x, a):
    r"""
    Return the falling factorial `(x)_a`.

    The notation in the literature is a mess: often `(x)_a`,
    but there are many other notations: GKP: Concrete Mathematics uses
    `x^{\underline{a}}`.

    Definition: for integer `a \ge 0` we have
    `x(x-1) \cdots (x-a+1)`. In all other cases we use the
    GAMMA-function: `\frac {\Gamma(x+1)} {\Gamma(x-a+1)}`.

    INPUT:

    -  ``x`` -- element of a ring

    -  ``a`` -- a non-negative integer or

    -  ``x and a`` -- any numbers

    OUTPUT: the falling factorial

    .. SEEALSO:: :func:`rising_factorial`

    EXAMPLES::

        sage: falling_factorial(10, 3)
        720
        sage: falling_factorial(10, RR('3.0'))
        720.000000000000
        sage: falling_factorial(10, RR('3.3'))
        1310.11633396601
        sage: falling_factorial(10, 10)
        3628800
        sage: factorial(10)
        3628800
        sage: a = falling_factorial(1+I, I); a
        gamma(I + 2)
        sage: CC(a)
        0.652965496420167 + 0.343065839816545*I
        sage: falling_factorial(1+I, 4)
        4*I + 2
        sage: falling_factorial(I, 4)
        -10

        sage: M = MatrixSpace(ZZ, 4, 4)
        sage: A = M([1,0,1,0,1,0,1,0,1,0,10,10,1,0,1,1])
        sage: falling_factorial(A, 2) # A(A - I)
        [  1   0  10  10]
        [  1   0  10  10]
        [ 20   0 101 100]
        [  2   0  11  10]

        sage: x = ZZ['x'].0
        sage: falling_factorial(x, 4)
        x^4 - 6*x^3 + 11*x^2 - 6*x

    TESTS:

    Check that :trac:`14858` is fixed::

        sage: falling_factorial(-4, SR(2))
        20

    Check that :trac:`16770` is fixed::

        sage: d = var('d')
        sage: parent(falling_factorial(d, 0))
        Symbolic Ring

    Check that :trac:`20075` is fixed::

        sage: bool(falling_factorial(int(4), int(2)) == falling_factorial(4,2))
        True

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: falling_factorial(int8(10), int8(3))
        720
        sage: from gmpy2 import mpz
        sage: falling_factorial(mpz(10), mpz(3))
        720

    AUTHORS:

    - Jaap Spies (2006-03-05)
    """
    from sage.symbolic.expression import Expression
    x = py_scalar_to_element(x)
    a = py_scalar_to_element(a)
    if (isinstance(a, Integer) or
        (isinstance(a, Expression) and
         a.is_integer())) and a >= 0:
        return prod(((x - i) for i in range(a)), z=x.parent().one())
    from sage.functions.all import gamma
    return gamma(x + 1) / gamma(x - a + 1)


def rising_factorial(x, a):
    r"""
    Return the rising factorial `(x)^a`.

    The notation in the literature is a mess: often `(x)^a`,
    but there are many other notations: GKP: Concrete Mathematics uses
    `x^{\overline{a}}`.

    The rising factorial is also known as the Pochhammer symbol, see
    Maple and Mathematica.

    Definition: for integer `a \ge 0` we have
    `x(x+1) \cdots (x+a-1)`. In all other cases we use the
    GAMMA-function: `\frac {\Gamma(x+a)} {\Gamma(x)}`.

    INPUT:

    -  ``x`` -- element of a ring

    -  ``a`` -- a non-negative integer or

    -  ``x and a`` -- any numbers

    OUTPUT: the rising factorial

    .. SEEALSO:: :func:`falling_factorial`

    EXAMPLES::

        sage: rising_factorial(10,3)
        1320

        sage: rising_factorial(10,RR('3.0'))
        1320.00000000000

        sage: rising_factorial(10,RR('3.3'))
        2826.38895824964

        sage: a = rising_factorial(1+I, I); a
        gamma(2*I + 1)/gamma(I + 1)
        sage: CC(a)
        0.266816390637832 + 0.122783354006372*I

        sage: a = rising_factorial(I, 4); a
        -10

        sage: x = polygen(ZZ)
        sage: rising_factorial(x, 4)
        x^4 + 6*x^3 + 11*x^2 + 6*x

    TESTS:

    Check that :trac:`14858` is fixed::

        sage: bool(rising_factorial(-4, 2) ==
        ....:      rising_factorial(-4, SR(2)) ==
        ....:      rising_factorial(SR(-4), SR(2)))
        True

    Check that :trac:`16770` is fixed::

        sage: d = var('d')
        sage: parent(rising_factorial(d, 0))
        Symbolic Ring

    Check that :trac:`20075` is fixed::

        sage: bool(rising_factorial(int(4), int(2)) == rising_factorial(4,2))
        True

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: rising_factorial(int8(10), int8(3))
        1320
        sage: from gmpy2 import mpz
        sage: rising_factorial(mpz(10), mpz(3))
        1320

    AUTHORS:

    - Jaap Spies (2006-03-05)
    """
    from sage.symbolic.expression import Expression
    x = py_scalar_to_element(x)
    a = py_scalar_to_element(a)
    if (isinstance(a, Integer) or
        (isinstance(a, Expression) and
         a.is_integer())) and a >= 0:
        return prod(((x + i) for i in range(a)), z=x.parent().one())
    from sage.functions.all import gamma
    return gamma(x + a) / gamma(x)


def integer_ceil(x):
    """
    Return the ceiling of x.

    EXAMPLES::

        sage: integer_ceil(5.4)
        6
        sage: integer_ceil(x)
        Traceback (most recent call last):
        ...
        NotImplementedError: computation of ceil of x not implemented

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import float32
        sage: integer_ceil(float32(5.4))
        6
        sage: from gmpy2 import mpfr
        sage: integer_ceil(mpfr(5.4))
        6
    """
    try:
        return ZZ(x.ceil())
    except AttributeError:
        try:
            return ZZ(math.ceil(float(x)))
        except TypeError:
            pass
    raise NotImplementedError("computation of ceil of %s not implemented"%x)


def integer_floor(x):
    r"""
    Return the largest integer `\leq x`.

    INPUT:

    -  ``x`` - an object that has a floor method or is
       coercible to int

    OUTPUT: an Integer

    EXAMPLES::

        sage: integer_floor(5.4)
        5
        sage: integer_floor(float(5.4))
        5
        sage: integer_floor(-5/2)
        -3
        sage: integer_floor(RDF(-5/2))
        -3

        sage: integer_floor(x)
        Traceback (most recent call last):
        ...
        NotImplementedError: computation of floor of x not implemented

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import float32
        sage: integer_floor(float32(5.4))
        5
        sage: from gmpy2 import mpfr
        sage: integer_floor(mpfr(5.4))
        5
    """
    try:
        return ZZ(x.floor())
    except AttributeError:
        try:
            return ZZ(math.floor(float(x)))
        except TypeError:
            pass
    raise NotImplementedError("computation of floor of %s not implemented"%x)


def two_squares(n):
    """
    Write the integer `n` as a sum of two integer squares if possible;
    otherwise raise a ``ValueError``.

    INPUT:

    - ``n`` -- an integer

    OUTPUT: a tuple `(a,b)` of non-negative integers such that
    `n = a^2 + b^2` with `a <= b`.

    EXAMPLES::

        sage: two_squares(389)
        (10, 17)
        sage: two_squares(21)
        Traceback (most recent call last):
        ...
        ValueError: 21 is not a sum of 2 squares
        sage: two_squares(21^2)
        (0, 21)
        sage: a,b = two_squares(100000000000000000129); a,b
        (4418521500, 8970878873)
        sage: a^2 + b^2
        100000000000000000129
        sage: two_squares(2^222+1)
        (253801659504708621991421712450521, 2583712713213354898490304645018692)
        sage: two_squares(0)
        (0, 0)
        sage: two_squares(-1)
        Traceback (most recent call last):
        ...
        ValueError: -1 is not a sum of 2 squares

    TESTS::

        sage: for _ in range(100):
        ....:     a = ZZ.random_element(2**16, 2**20)
        ....:     b = ZZ.random_element(2**16, 2**20)
        ....:     n = a**2 + b**2
        ....:     aa,bb = two_squares(n)
        ....:     assert aa**2 + bb**2 == n

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: two_squares(int16(389))
        (10, 17)
        sage: from gmpy2 import mpz
        sage: two_squares(mpz(389))
        (10, 17)

    ALGORITHM:

    See http://www.schorn.ch/howto.html
    """
    n = ZZ(n)

    if n <= 0:
        if n == 0:
            z = ZZ.zero()
            return (z, z)
        raise ValueError("%s is not a sum of 2 squares"%n)

    if n.nbits() <= 32:
        from sage.rings import sum_of_squares
        return sum_of_squares.two_squares_pyx(n)

    # Start by factoring n (which seems to be unavoidable)
    F = n.factor(proof=False)

    # First check whether it is possible to write n as a sum of two
    # squares: all prime powers p^e must have p = 2 or p = 1 mod 4
    # or e even.
    for (p,e) in F:
        if e % 2 == 1 and p % 4 == 3:
            raise ValueError("%s is not a sum of 2 squares"%n)

    # We run over all factors of n, write each factor p^e as
    # a sum of 2 squares and accumulate the product
    # (using multiplication in Z[I]) in a^2 + b^2.
    from sage.rings.finite_rings.integer_mod import Mod
    a = ZZ.one()
    b = ZZ.zero()
    for (p,e) in F:
        if e >= 2:
            m = p ** (e//2)
            a *= m
            b *= m
        if e % 2 == 1:
            if p == 2:
                # (a + bi) *= (1 + I)
                a,b = a - b, a + b
            else:  # p = 1 mod 4
                # Find a square root of -1 mod p.
                # If y is a non-square, then y^((p-1)/4) is a square root of -1.
                y = Mod(2,p)
                while True:
                    s = y**((p-1)/4)
                    if not s*s + 1:
                        s = s.lift()
                        break
                    y += 1
                # Apply Cornacchia's algorithm to write p as r^2 + s^2.
                r = p
                while s*s > p:
                    r,s = s, r % s
                r %= s

                # Multiply (a + bI) by (r + sI)
                a,b = a*r - b*s, b*r + a*s

    a = a.abs()
    b = b.abs()
    assert a*a + b*b == n
    if a <= b:
        return (a,b)
    else:
        return (b,a)


def three_squares(n):
    """
    Write the integer `n` as a sum of three integer squares if possible;
    otherwise raise a ``ValueError``.

    INPUT:

    - ``n`` -- an integer

    OUTPUT: a tuple `(a,b,c)` of non-negative integers such that
    `n = a^2 + b^2 + c^2` with `a <= b <= c`.

    EXAMPLES::

        sage: three_squares(389)
        (1, 8, 18)
        sage: three_squares(946)
        (9, 9, 28)
        sage: three_squares(2986)
        (3, 24, 49)
        sage: three_squares(7^100)
        (0, 0, 1798465042647412146620280340569649349251249)
        sage: three_squares(11^111-1)
        (616274160655975340150706442680, 901582938385735143295060746161, 6270382387635744140394001363065311967964099981788593947233)
        sage: three_squares(7 * 2^41)
        (1048576, 2097152, 3145728)
        sage: three_squares(7 * 2^42)
        Traceback (most recent call last):
        ...
        ValueError: 30786325577728 is not a sum of 3 squares
        sage: three_squares(0)
        (0, 0, 0)
        sage: three_squares(-1)
        Traceback (most recent call last):
        ...
        ValueError: -1 is not a sum of 3 squares

    TESTS::

        sage: for _ in range(100):
        ....:     a = ZZ.random_element(2**16, 2**20)
        ....:     b = ZZ.random_element(2**16, 2**20)
        ....:     c = ZZ.random_element(2**16, 2**20)
        ....:     n = a**2 + b**2 + c**2
        ....:     aa,bb,cc = three_squares(n)
        ....:     assert aa**2 + bb**2 + cc**2 == n

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: three_squares(int16(389))
        (1, 8, 18)
        sage: from gmpy2 import mpz
        sage: three_squares(mpz(389))
        (1, 8, 18)

    ALGORITHM:

    See http://www.schorn.ch/howto.html
    """
    n = ZZ(n)

    if n <= 0:
        if n == 0:
            z = ZZ.zero()
            return (z, z, z)
        raise ValueError("%s is not a sum of 3 squares"%n)

    if n.nbits() <= 32:
        from sage.rings import sum_of_squares
        return sum_of_squares.three_squares_pyx(n)

    # First, remove all factors 4 from n
    e = n.valuation(2)//2
    m = ZZ.one() << e
    N = n >> (2*e)

    # Let x be the largest integer at most sqrt(N)
    x, r = N.sqrtrem()
    # We need to check for this special case,
    # otherwise N - x^2 will always factor.
    if not r:
        z = ZZ.zero()
        return (z, z, x*m)

    # Consider different cases to find an x such that N - x^2 is easily
    # written as the sum of 2 squares, because it is either p or 2p,
    # with p a prime which is 1 mod 4.
    if N % 4 == 1:
        # Write N = x^2 + p with x even, p = 1 mod 4 prime
        if x % 2 == 1:
            x -= 1
        while x >= 0:
            p = N - x*x
            if p.is_pseudoprime():
                break
            x -= 2
    elif N % 4 == 2:
        # Write N = x^2 + p with x odd, p = 1 mod 4 prime
        if x % 2 == 0:
            x -= 1
        while x >= 0:
            p = N - x*x
            if p.is_pseudoprime():
                break
            x -= 2
    elif N % 8 == 3:
        # Write N = x^2 + 2p with x odd, p = 1 mod 4 prime
        if x % 2 == 0:
            x -= 1
        while x >= 0:
            p = (N - x*x) >> 1
            if p.is_pseudoprime():
                break
            x -= 2
    else:  # 7 mod 8
        raise ValueError("%s is not a sum of 3 squares" % n)

    if x < 0:
        # We found no good x, brute force instead.
        # Normally, this should only happen for small values of N.
        if N > 10000:
            from warnings import warn
            warn("Brute forcing sum of 3 squares for large N = %s"%N, RuntimeWarning)
        x = N.isqrt()

    # In the usual case, this loop will only be executed once, since
    # we already know the "right" value of x.
    # This will only really loop if we hit the "x < 0" case above.
    while True:
        try:
            a,b = two_squares(N - x*x)
            break
        except ValueError:
            x -= 1
            assert x >= 0

    if x >= b:
        return (a*m, b*m, x*m)
    elif x >= a:
        return (a*m, x*m, b*m)
    else:
        return (x*m, a*m, b*m)


def four_squares(n):
    """
    Write the integer `n` as a sum of four integer squares.

    INPUT:

    - ``n`` -- an integer

    OUTPUT: a tuple `(a,b,c,d)` of non-negative integers such that
    `n = a^2 + b^2 + c^2 + d^2` with `a <= b <= c <= d`.

    EXAMPLES::

        sage: four_squares(3)
        (0, 1, 1, 1)
        sage: four_squares(13)
        (0, 0, 2, 3)
        sage: four_squares(130)
        (0, 0, 3, 11)
        sage: four_squares(1101011011004)
        (90, 102, 1220, 1049290)
        sage: four_squares(10^100-1)
        (155024616290, 2612183768627, 14142135623730950488016887, 99999999999999999999999999999999999999999999999999)
        sage: for i in range(2^129, 2^129+10000):  # long time
        ....:     S = four_squares(i)
        ....:     assert sum(x^2 for x in S) == i

    TESTS::

        sage: for _ in range(100):
        ....:     n = ZZ.random_element(2**32,2**34)
        ....:     aa,bb,cc,dd = four_squares(n)
        ....:     assert aa**2 + bb**2 + cc**2 + dd**2 == n

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: four_squares(int16(389))
        (0, 1, 8, 18)
        sage: from gmpy2 import mpz
        sage: four_squares(mpz(389))
        (0, 1, 8, 18)
    """
    n = ZZ(n)

    if n <= 0:
        if n == 0:
            z = ZZ.zero()
            return (z, z, z, z)
        raise ValueError("%s is not a sum of 4 squares"%n)

    if n.nbits() <= 32:
        from sage.rings import sum_of_squares
        return sum_of_squares.four_squares_pyx(n)

    # First, remove all factors 4 from n
    e = n.valuation(2) // 2
    m = ZZ.one() << e
    N = n >> (2*e)

    # Subtract a suitable x^2 such that N - x^2 is 1,2,3,5,6 mod 8,
    # which can then be written as a sum of 3 squares.
    x = N.isqrt()
    y = N - x*x
    if y >= 7 and (y % 4 == 0 or y % 8 == 7):
        x -= 1
        y += 2*x + 1

    a,b,c = three_squares(y)

    # Correct sorting is guaranteed by construction
    return (a*m, b*m, c*m, x*m)


def sum_of_k_squares(k, n):
    """
    Write the integer `n` as a sum of `k` integer squares if possible;
    otherwise raise a ``ValueError``.

    INPUT:

    - ``k`` -- a non-negative integer

    - ``n`` -- an integer

    OUTPUT: a tuple `(x_1, ..., x_k)` of non-negative integers such that
    their squares sum to `n`.

    EXAMPLES::

        sage: sum_of_k_squares(2, 9634)
        (15, 97)
        sage: sum_of_k_squares(3, 9634)
        (0, 15, 97)
        sage: sum_of_k_squares(4, 9634)
        (1, 2, 5, 98)
        sage: sum_of_k_squares(5, 9634)
        (0, 1, 2, 5, 98)
        sage: sum_of_k_squares(6, 11^1111-1)
        (19215400822645944253860920437586326284, 37204645194585992174252915693267578306, 3473654819477394665857484221256136567800161086815834297092488779216863122, 5860191799617673633547572610351797996721850737768032876360978911074629287841061578270832330322236796556721252602860754789786937515870682024273948, 20457423294558182494001919812379023992538802203730791019728543439765347851316366537094696896669915675685581905102118246887673397020172285247862426612188418787649371716686651256443143210952163970564228423098202682066311189439731080552623884051737264415984619097656479060977602722566383385989, 311628095411678159849237738619458396497534696043580912225334269371611836910345930320700816649653412141574887113710604828156159177769285115652741014638785285820578943010943846225597311231847997461959204894255074229895666356909071243390280307709880906261008237873840245959883405303580405277298513108957483306488193844321589356441983980532251051786704380984788999660195252373574924026139168936921591652831237741973242604363696352878914129671292072201700073286987126265965322808664802662993006926302359371379531571194266134916767573373504566621665949840469229781956838744551367172353)
        sage: sum_of_k_squares(7, 0)
        (0, 0, 0, 0, 0, 0, 0)
        sage: sum_of_k_squares(30,999999)
        (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 3, 7, 44, 999)
        sage: sum_of_k_squares(1, 9)
        (3,)
        sage: sum_of_k_squares(1, 10)
        Traceback (most recent call last):
        ...
        ValueError: 10 is not a sum of 1 square
        sage: sum_of_k_squares(1, -10)
        Traceback (most recent call last):
        ...
        ValueError: -10 is not a sum of 1 square
        sage: sum_of_k_squares(0, 9)
        Traceback (most recent call last):
        ...
        ValueError: 9 is not a sum of 0 squares
        sage: sum_of_k_squares(0, 0)
        ()
        sage: sum_of_k_squares(7, -1)
        Traceback (most recent call last):
        ...
        ValueError: -1 is not a sum of 7 squares
        sage: sum_of_k_squares(-1, 0)
        Traceback (most recent call last):
        ...
        ValueError: k = -1 must be non-negative

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int16
        sage: sum_of_k_squares(int16(2), int16(9634))
        (15, 97)
        sage: from gmpy2 import mpz
        sage: sum_of_k_squares(mpz(2), mpz(9634))
        (15, 97)
    """
    n = ZZ(n)
    k = int(k)

    if k <= 4:
        if k == 4:
            return four_squares(n)
        if k == 3:
            return three_squares(n)
        if k == 2:
            return two_squares(n)
        if k == 1:
            if n >= 0:
                x, r = n.sqrtrem()
                if not r:
                    return (x,)
            raise ValueError("%s is not a sum of 1 square"%n)
        if k == 0:
            if n == 0:
                return tuple()
            raise ValueError("%s is not a sum of 0 squares"%n)
        raise ValueError("k = %s must be non-negative"%k)

    if n < 0:
        raise ValueError("%s is not a sum of %s squares"%(n,k))

    # Recursively subtract the largest square
    t = []
    while k > 4:
        x = n.isqrt()
        t.insert(0, x)
        n -= x*x
        k -= 1

    t = list(four_squares(n)) + t
    return tuple(t)


def subfactorial(n):
    r"""
    Subfactorial or rencontres numbers, or derangements: number of
    permutations of `n` elements with no fixed points.

    INPUT:


    -  ``n`` - non negative integer


    OUTPUT:


    -  ``integer`` - function value


    EXAMPLES::

        sage: subfactorial(0)
        1
        sage: subfactorial(1)
        0
        sage: subfactorial(8)
        14833

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: subfactorial(int8(8))
        14833
        sage: from gmpy2 import mpz
        sage: subfactorial(mpz(8))
        14833

    AUTHORS:

    - Jaap Spies (2007-01-23)
    """
    return factorial(n)*sum(((-1)**k)/factorial(k) for k in range(n+1))


def is_power_of_two(n):
    r"""
    Return whether ``n`` is a power of 2.

    INPUT:

    -  ``n`` -- integer

    OUTPUT:

    boolean

    EXAMPLES::

        sage: is_power_of_two(1024)
        True
        sage: is_power_of_two(1)
        True
        sage: is_power_of_two(24)
        False
        sage: is_power_of_two(0)
        False
        sage: is_power_of_two(-4)
        False

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: is_power_of_two(int8(16))
        True
        sage: is_power_of_two(int8(24))
        False
        sage: from gmpy2 import mpz
        sage: is_power_of_two(mpz(16))
        True
        sage: is_power_of_two(mpz(24))
        False
    """
    return ZZ(n).popcount() == 1


def differences(lis, n=1):
    """
    Return the `n` successive differences of the elements in
    `lis`.

    EXAMPLES::

        sage: differences(prime_range(50))
        [1, 2, 2, 4, 2, 4, 2, 4, 6, 2, 6, 4, 2, 4]
        sage: differences([i^2 for i in range(1,11)])
        [3, 5, 7, 9, 11, 13, 15, 17, 19]
        sage: differences([i^3 + 3*i for i in range(1,21)])
        [10, 22, 40, 64, 94, 130, 172, 220, 274, 334, 400, 472, 550, 634, 724, 820, 922, 1030, 1144]
        sage: differences([i^3 - i^2 for i in range(1,21)], 2)
        [10, 16, 22, 28, 34, 40, 46, 52, 58, 64, 70, 76, 82, 88, 94, 100, 106, 112]
        sage: differences([p - i^2 for i, p in enumerate(prime_range(50))], 3)
        [-1, 2, -4, 4, -4, 4, 0, -6, 8, -6, 0, 4]

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: differences([int8(1),int8(4),int8(6),int8(19)])
        [3, 2, 13]
        sage: from gmpy2 import mpz
        sage: differences([mpz(1),mpz(4),mpz(6),mpz(19)])
        [mpz(3), mpz(2), mpz(13)]

    AUTHORS:

    - Timothy Clemans (2008-03-09)
    """
    n = ZZ(n)
    if n < 1:
        raise ValueError('n must be greater than 0')
    lis = [lis[i + 1] - num for i, num in enumerate(lis[:-1])]
    if n == 1:
        return lis
    return differences(lis, n - 1)


def _key_complex_for_display(a):
    """
    Key function to sort complex numbers for display only.

    Real numbers (with a zero imaginary part) come before complex numbers,
    and are sorted. Complex numbers are sorted by their real part
    unless their real parts are quite close, in which case they are
    sorted by their imaginary part.

    EXAMPLES::

        sage: import sage.arith.misc
        sage: key_c = sage.arith.misc._key_complex_for_display

        sage: key_c(CC(5))
        (0, 5.00000000000000)
        sage: key_c(CC(5, 5))
        (1, 5.00000000, 5.00000000000000)

        sage: CIF200 = ComplexIntervalField(200)
        sage: key_c(CIF200(5))
        (0, 5)
        sage: key_c(CIF200(5, 5))
        (1, 5.00000000, 5)
    """
    ar = a.real()
    ai = a.imag()
    if not ai:
        return (0, ar)
    epsilon = 1e-10
    if ar.abs() < epsilon:
        ar_truncated = 0
    elif ar.prec() < 34:
        ar_truncated = ar
    else:
        ar_truncated = ar.n(digits=9)
    return (1, ar_truncated, ai)


def sort_complex_numbers_for_display(nums):
    r"""
    Given a list of complex numbers (or a list of tuples, where the
    first element of each tuple is a complex number), we sort the list
    in a "pretty" order.  First come the real numbers (with zero
    imaginary part), then the complex numbers sorted according to
    their real part.  If two complex numbers have the same real part,
    then they are sorted according to their
    imaginary part.

    This is not a useful function mathematically (not least because
    there is no principled way to determine whether the real components
    should be treated as equal or not).  It is called by various
    polynomial root-finders; its purpose is to make doctest printing
    more reproducible.

    We deliberately choose a cumbersome name for this function to
    discourage use, since it is mathematically meaningless.

    EXAMPLES::

        sage: import sage.arith.misc
        sage: sort_c = sort_complex_numbers_for_display
        sage: nums = [CDF(i) for i in range(3)]
        sage: for i in range(3):
        ....:     nums.append(CDF(i + RDF.random_element(-3e-11, 3e-11),
        ....:                     RDF.random_element()))
        ....:     nums.append(CDF(i + RDF.random_element(-3e-11, 3e-11),
        ....:                     RDF.random_element()))
        sage: shuffle(nums)
        sage: sort_c(nums)
        [0.0, 1.0, 2.0, -2.862406201002009e-11 - 0.7088740263015161*I, 2.2108362706985576e-11 - 0.43681052967509904*I, 1.0000000000138833 - 0.7587654737635712*I, 0.9999999999760288 - 0.7238965893336062*I, 1.9999999999874383 - 0.4560801012073723*I, 1.9999999999869107 + 0.6090836283134269*I]
    """
    if not nums:
        return nums

    if isinstance(nums[0], tuple):
        return sorted(nums, key=lambda t: _key_complex_for_display(t[0]))
    else:
        return sorted(nums, key=_key_complex_for_display)


def fundamental_discriminant(D):
    r"""
    Return the discriminant of the quadratic extension
    `K=Q(\sqrt{D})`, i.e. an integer d congruent to either 0 or
    1, mod 4, and such that, at most, the only square dividing it is
    4.

    INPUT:

    - ``D`` - an integer

    OUTPUT:

    - an integer, the fundamental discriminant

    EXAMPLES::

        sage: fundamental_discriminant(102)
        408
        sage: fundamental_discriminant(720)
        5
        sage: fundamental_discriminant(2)
        8

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: fundamental_discriminant(int8(102))
        408
        sage: from gmpy2 import mpz
        sage: fundamental_discriminant(mpz(102))
        408
    """
    D = ZZ(D)
    D = D.squarefree_part()
    if D % 4 == 1:
        return D
    return 4 * D


def squarefree_divisors(x):
    """
    Return an iterator over the squarefree divisors (up to units)
    of this ring element.

    Depends on the output of the prime_divisors function.

    Squarefree divisors of an integer are not necessarily
    yielded in increasing order.

    INPUT:

    - x -- an element of any ring for which the prime_divisors
      function works.

    EXAMPLES:

    Integers with few prime divisors::

        sage: list(squarefree_divisors(7))
        [1, 7]
        sage: list(squarefree_divisors(6))
        [1, 2, 3, 6]
        sage: list(squarefree_divisors(12))
        [1, 2, 3, 6]

    Squarefree divisors are not yielded in increasing order::

        sage: list(squarefree_divisors(30))
        [1, 2, 3, 6, 5, 10, 15, 30]

    TESTS:

    Check that the first divisor (i.e. `1`) is a Sage integer (see
    :trac:`17852`)::

        sage: a = next(squarefree_divisors(14))
        sage: a
        1
        sage: type(a)
        <type 'sage.rings.integer.Integer'>

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: list(squarefree_divisors(int8(12)))
        [1, 2, 3, 6]
        sage: from gmpy2 import mpz
        sage: list(squarefree_divisors(mpz(12)))
        [1, 2, 3, 6]
    """
    for a in powerset(prime_divisors(x)):
        yield prod(a, ZZ.one())


def dedekind_sum(p, q, algorithm='default'):
    r"""
    Return the Dedekind sum `s(p,q)` defined for integers `p`, `q` as

    .. MATH::

        s(p,q) = \sum_{i=0}^{q-1} \left(\!\left(\frac{i}{q}\right)\!\right)
                                  \left(\!\left(\frac{pi}{q}\right)\!\right)

    where

    .. MATH::

        ((x))=\begin{cases}
            x-\lfloor x \rfloor - \frac{1}{2} &\mbox{if }
                x \in \QQ \setminus \ZZ \\
            0 & \mbox{if } x \in \ZZ.
            \end{cases}

    .. WARNING::

        Caution is required as the Dedekind sum sometimes depends on the
        algorithm or is left undefined when `p` and `q` are not coprime.

    INPUT:

    -  ``p``, ``q`` -- integers
    -  ``algorithm`` -- must be one of the following

       -  ``'default'`` - (default) use FLINT
       -  ``'flint'`` - use FLINT
       -  ``'pari'`` - use PARI (gives different results if `p` and `q`
          are not coprime)

    OUTPUT: a rational number

    EXAMPLES:

    Several small values::

        sage: for q in range(10): print([dedekind_sum(p,q) for p in range(q+1)])
        [0]
        [0, 0]
        [0, 0, 0]
        [0, 1/18, -1/18, 0]
        [0, 1/8, 0, -1/8, 0]
        [0, 1/5, 0, 0, -1/5, 0]
        [0, 5/18, 1/18, 0, -1/18, -5/18, 0]
        [0, 5/14, 1/14, -1/14, 1/14, -1/14, -5/14, 0]
        [0, 7/16, 1/8, 1/16, 0, -1/16, -1/8, -7/16, 0]
        [0, 14/27, 4/27, 1/18, -4/27, 4/27, -1/18, -4/27, -14/27, 0]

    Check relations for restricted arguments::

        sage: q = 23; dedekind_sum(1, q); (q-1)*(q-2)/(12*q)
        77/46
        77/46
        sage: p, q = 100, 723    # must be coprime
        sage: dedekind_sum(p, q) + dedekind_sum(q, p)
        31583/86760
        sage: -1/4 + (p/q + q/p + 1/(p*q))/12
        31583/86760

    We check that evaluation works with large input::

        sage: dedekind_sum(3^54 - 1, 2^93 + 1)
        459340694971839990630374299870/29710560942849126597578981379
        sage: dedekind_sum(3^54 - 1, 2^93 + 1, algorithm='pari')
        459340694971839990630374299870/29710560942849126597578981379

    We check consistency of the results::

        sage: dedekind_sum(5, 7, algorithm='default')
        -1/14
        sage: dedekind_sum(5, 7, algorithm='flint')
        -1/14
        sage: dedekind_sum(5, 7, algorithm='pari')
        -1/14
        sage: dedekind_sum(6, 8, algorithm='default')
        -1/8
        sage: dedekind_sum(6, 8, algorithm='flint')
        -1/8
        sage: dedekind_sum(6, 8, algorithm='pari')
        -1/8

    Tests with numpy and gmpy2 numbers::

        sage: from numpy import int8
        sage: dedekind_sum(int8(5), int8(7), algorithm='default')
        -1/14
        sage: from gmpy2 import mpz
        sage: dedekind_sum(mpz(5), mpz(7), algorithm='default')
        -1/14

    REFERENCES:

    - [Ap1997]_

    - :wikipedia:`Dedekind\_sum`
    """
    if algorithm == 'default' or algorithm == 'flint':
        return flint_dedekind_sum(p, q)

    if algorithm == 'pari':
        import sage.interfaces.gp
        x = sage.interfaces.gp.gp('sumdedekind(%s,%s)' % (p, q))
        return Rational(x)

    raise ValueError('unknown algorithm')


def gauss_sum(char_value, finite_field):
    r"""
    Return the Gauss sums for a general finite field.

    INPUT:

    - ``char_value`` -- choice of multiplicative character, given by
      its value on the ``finite_field.multiplicative_generator()``

    - ``finite_field`` -- a finite field

    OUTPUT:

    an element of the parent ring of ``char_value``, that can be any
    field containing enough roots of unity, for example the
    ``UniversalCyclotomicField``, ``QQbar`` or ``ComplexField``

    For a finite field `F` of characteristic `p`, the Gauss sum
    associated to a multiplicative character `\chi` (with values in a
    ring `K`) is defined as

    .. MATH::

        \sum_{x \in F^{\times}} \chi(x) \zeta_p^{\operatorname{Tr} x},

    where `\zeta_p \in K` is a primitive root of unity of order `p` and
    Tr is the trace map from `F` to its prime field `\GF{p}`.

    For more info on Gauss sums, see :wikipedia:`Gauss_sum`.

    .. TODO::

        Implement general Gauss sums for an arbitrary pair
        ``(multiplicative_character, additive_character)``

    EXAMPLES::

        sage: from sage.arith.misc import gauss_sum
        sage: F = GF(5); q = 5
        sage: zq = UniversalCyclotomicField().zeta(q-1)
        sage: L = [gauss_sum(zq**i,F) for i in range(5)]; L
        [-1,
         E(20)^4 + E(20)^13 - E(20)^16 - E(20)^17,
         E(5) - E(5)^2 - E(5)^3 + E(5)^4,
         E(20)^4 - E(20)^13 - E(20)^16 + E(20)^17,
         -1]
        sage: [g*g.conjugate() for g in L]
        [1, 5, 5, 5, 1]

        sage: F = GF(11**2); q = 11**2
        sage: zq = UniversalCyclotomicField().zeta(q-1)
        sage: g = gauss_sum(zq**4,F)
        sage: g*g.conjugate()
        121

    TESTS::

        sage: F = GF(11); q = 11
        sage: zq = UniversalCyclotomicField().zeta(q-1)
        sage: gauss_sum(zq**2,F).n(60)
        2.6361055643248352 + 2.0126965627574471*I

        sage: zq = QQbar.zeta(q-1)
        sage: gauss_sum(zq**2,F)
        2.636105564324836? + 2.012696562757447?*I

        sage: zq = ComplexField(60).zeta(q-1)
        sage: gauss_sum(zq**2,F)
        2.6361055643248352 + 2.0126965627574471*I

        sage: F = GF(7); q = 7
        sage: zq = QQbar.zeta(q-1)
        sage: D = DirichletGroup(7, QQbar)
        sage: all(D[i].gauss_sum()==gauss_sum(zq**i,F) for i in range(6))
        True

        sage: gauss_sum(1,QQ)
        Traceback (most recent call last):
        ...
        ValueError: second input must be a finite field

    .. SEEALSO::

        - :func:`sage.rings.padics.misc.gauss_sum` for a `p`-adic version
        - :meth:`sage.modular.dirichlet.DirichletCharacter.gauss_sum`
          for prime finite fields
        - :meth:`sage.modular.dirichlet.DirichletCharacter.gauss_sum_numerical`
          for prime finite fields
    """
    from sage.categories.fields import Fields
    if finite_field not in Fields().Finite():
        raise ValueError('second input must be a finite field')

    ring = char_value.parent()
    q = finite_field.cardinality()
    p = finite_field.characteristic()
    gen = finite_field.multiplicative_generator()
    zeta_p_powers = ring.zeta(p).powers(p)
    zeta_q = char_value

    resu = ring.zero()
    gen_power = finite_field.one()
    zq_power = ring.one()
    for k in range(q - 1):
        resu += zq_power * zeta_p_powers[gen_power.trace().lift()]
        gen_power *= gen
        zq_power *= zeta_q
    return resu


def dedekind_psi(N):
    r"""
    Return the value of the Dedekind psi function at ``N``.

    INPUT:

    - ``N`` -- a positive integer

    OUTPUT:

    an integer

    The Dedekind psi function is the multiplicative function defined by

    .. MATH::

        \psi(n) = n \prod_{p|n, p prime} (1 + 1/p).

    See :wikipedia:`Dedekind_psi_function` and :oeis:`A001615`.

    EXAMPLES::

        sage: from sage.arith.misc import dedekind_psi
        sage: [dedekind_psi(d) for d in range(1, 12)]
        [1, 3, 4, 6, 6, 12, 8, 12, 12, 18, 12]
    """
    N = Integer(N)
    return Integer(N * prod(1 + 1 / p for p in N.prime_divisors()))