\chapter{Technical reference guide for low-level functions}

In this chapter, we give a description of almost all low-level functions of
the PARI system. These essentially include functions for handling all the
PARI types. Higher level functions, such as arithmetic or transcendental
functions, are described fully in Chapter 3 of this manual.

\section{Level 0 kernel (operations on unsigned longs)}
The type ``{\tt ulong}'' is defined in the file {\tt gencom.h} as 
{\tt unsigned long}. The global {\tt ulong} variables {\tt overflow} 
(which will contain only 0 or 1) and {\tt hiremainder} are defined in the
file {\tt versionXXX.c}. 

For the non-68k versions, we need level 0 operations simulating basic
operations of the 68020 processor. For compatibility reasons, and also
for very rare cases, they are also included in the 68k version, although
they are not really needed. Note that in the given prototypes ulongs are 
sometimes implicitly typecasted as ints or longs.

{\tt int addll(int x, int y)}: add the ulongs x and y, output the lower 32
bits and put the overflow bit in {\tt overflow}.

{\tt int addllx(int x, int y)}: add {\tt overflow} to the sum of the ulongs x
and y, output the lower 32 bits and put the overflow bit in {\tt overflow}.

{\tt int subll(int x, int y)}: subtract the ulongs x and y, output the lower 
32 bits and put the overflow bit in {\tt overflow}.

{\tt int subllx(int x, int y)}: subtract {\tt overflow} from the difference
of the ulongs x and y, output the lower 32 bits and put the overflow bit in
{\tt overflow}.

{\tt int shiftl(ulong x, ulong y)}: left shift the ulong x by y bits, output
the lower 32 bits and put the higher 32 bits in {\tt hiremainder}. We must
have $\text{y}\le32$.

{\tt int shiftlr(ulong x, ulong y)}: right shift the ulong $\text{x}<<32$ by
y bits, output the higher 32 bits and put the lower 32 bits in 
{\tt hiremainder}. We must have $\text{y}\le32$.

{\tt int bfffo(ulong x)}: output the number of bits that the ulong x should
be shifted left so that its leftmost bit is equal to 1 (32 if x is equal to 0).

{\tt int mulll(ulong x, ulong y)}: multiply the ulong x by the ulong y, output
the lower 32 bits and put the higher 32 bits in {\tt hiremainder}.

{\tt int addmul(ulong x, ulong y)}: add {\tt hiremainder} to the product of
the ulong x by the ulong y, output the lower 32 bits and put the higher 32
bits in {\tt hiremainder}.

{\tt int divll(ulong x, ulong y)}: output the Euclidean quotient of the ulong
x by the ulong y and put the remainder in {\tt hiremainder}.

{\tt long mulmodll(ulong x, ulong y, ulong z)}: output x$*$y modulo z.

\section{Level 1 kernel (operations on longs, integers and reals)}
In this section as elsewhere, ``long'' denotes a 32-bit signed C-integer,
``integer'' denotes a PARI multiprecise integer (type 1), ``real'' denotes
a PARI multiprecise real (type 2). Refer to the users manual for general
background.

\subsec{Basic unit and subunit handling functions}

{\tt long typ(GEN x)}: gives the type number of x.

{\tt long lg(GEN x)}: gives the length in 32-bit words of x.

{\tt long lgef(GEN x)}: gives the effective length in 32-bit words of x. This
makes sense only if x is an integer (type 1) or a polynomial (type 10).

{\tt long mant(GEN x,long n)}: gives the n-th 32-bit mantissa word of the 
integer or real x (the most significant word is n=1).

{\tt long signe(GEN x)}: gives the sign ($-1$, 0 or 1) of x. Should be used
for integers, reals, polynomials and power series (for these last two only 0 
or 1 are possible). For all other types, use {\tt gsigne(GEN x)}.

{\tt long expo(GEN x)}: gives the unbiased 24-bit binary exponent of the real
number x. If x is not a real number, one can define such a quantity, but in
that case use the function {\tt gexpo(GEN x)}.

{\tt long expi(GEN x)}: binary exponent of the real number equal to the integer
x. This is the special case of the function {\tt gexpo(GEN x)} corresponding
to the case where x is of type integer.

{\tt long pere(GEN x)}: gives the number of objects pointing at x, saturating
at 255. Now unused.

{\tt long valp(GEN x)}: gives the unbiased 16-bit p-adic valuation (for a
p-adic) or X-adic valuation (for a power series) of x.

{\tt long precp(GEN x)}: gives the precision of the p-adic x.

{\tt long varn(GEN x)}: gives the variable number of x (between 0 and 255).
Should be used only for polynomials and power series. For all other types,
use {\tt gvar(GEN x)}.

{\tt void settyp(GEN x, long s)}: set equal to s the type number of x. This
should be used with extreme care since usually the type is set otherwise.

{\tt void setlg(GEN x, long s)}: set equal to s the length of x. Again this
should be used with extreme care since usually the length is set otherwise.

{\tt void setlgef(GEN x, long s)}: set equal to s the effective length of x,
where x is an integer or a polynomial. The number s must be less than or equal
to the length of x.

{\tt void setmant(GEN x, long n, long s)}: set equal to s the n-th mantissa
word of the integer or real number x.

{\tt void setsigne(GEN x, long s)}: set equal to s the sign of x. If x is an
integer or real, s must be equal to $-1$, 0 or 1, and if x is a polynomial or
a power series, s must be equal to 0 or 1.

{\tt void setexpo(GEN x, long s)}: set equal to s the binary exponent of
the real number x. s must be a 24-bit signed number.

{\tt void setpere(GEN x, long s)}: set equal to s the number of objects 
pointing to x, where $0\le s\le255$. Now unused.

{\tt void incpere(GEN x)}: increase by 1 the number of objects pointing to x,
saturating at 255. Now unused.

{\tt void setvalp(GEN x, long s)}: set equal to s the p-adic valuation of x
(if x is a p-adic) or the X-adic valuation of x (if x is a power series).

{\tt void setprecp(GEN x, long s)}: set equal to s the p-adic precision of the
p-adic number x.

{\tt void setvarn(GEN x, long s)}: set equal to s ($0\le s\le255$) the variable
number of the polynomial or power series x.

\subsec{Memory allocation on the PARI stack}

{\tt GEN cgetg(long n, long t)}: allocate memory on the PARI stack for an
object of length n and type t, and initialize its first codeword (containing
{\tt typ} (1 byte), {\tt pere} (1 byte), {\tt lg} (2 bytes).

{\tt GEN cgeti(long n)}: allocate memory on the PARI stack for an integer
of length n, and initialize its first codeword. Identical to {\tt cgetg(n,1)}.

{\tt GEN cgetr(long n)}: allocate memory on the PARI stack for a real
of length n, and initialize its first codeword. Identical to {\tt cgetg(n,2)}.

{\tt void cgiv(GEN x)}: free object x if it is the last created on the PARI
stack (otherwise disaster occurs).

{\tt GEN gerepile(long p, long q, GEN x)}: general garbage collector for the
PARI stack. See chapter 4 of the manual for a detailed explanation.

\subsec{Assignments, conversions and integer parts}

{\tt void mpaff(GEN x, GEN z)}: assign x into z (x and z integers or reals).

{\tt void affsz(long s, GEN z)}: assign the long s into the integer or real z.

{\tt void affsi(long s, GEN z)}: assign the long s into the integer z.

{\tt void affsr(long s, GEN z)}: assign the long s into the real z.

{\tt void affii(GEN x, GEN z)}: assign the integer x into the integer z.

{\tt void affir(GEN x, GEN z)}: assign the integer x into the real z.

{\tt void affrs(GEN x, long s)}: assign the real x into the long s. This is a
forbidden assignment in PARI, so an error message is issued.

{\tt void affri(GEN x, GEN z)}: assign the real x into the integer z. This is a
forbidden assignment in PARI, so an error message is issued.

{\tt void affrr(GEN x, GEN z)}: assign the real x into the real z.
\smallskip
{\tt GEN stoi(long s)}: create the PARI integer corresponding to the long s.

{\tt long itos(GEN x)}: convert to long the PARI integer x (if possible,
otherwise an error message is issued).
\smallskip
{\tt GEN mptrunc(GEN x)}: truncation of the integer or real x (not the same
as the integer part if x is noninteger and negative).

{\tt void mptruncz(GEN x, GEN z)}: assign the truncation of the integer or
real x into z. 

{\tt GEN mpent(GEN x)}: true integer part of the integer or real x (i.e. the
{\tt floor} function).

{\tt void mpentz(GEN x, GEN z)}: assign the true integer part of the integer
or real x into z.

\subsec{Valuation and shift}

{\tt long vals(long s)}: 2-adic valuation of the long s. Returns $-1$ if
s is equal to 0, with no error.

{\tt long vali(GEN x)}: 2-adic valuation of the integer x. Returns $-1$ if
s is equal to 0, with no error.

{\tt GEN mpshift(GEN x, long n)}: shift by n the real or integer x. If n is
positive, this is a left shift, i.e. multiplication by $2^{\text{n}}$. If
n is negative, it is a right shift, hence the truncation of the division of x
by $2^{-\text{n}}$.

{\tt void mpshiftz(GEN x, long n, GEN z)}: assign into z the real or integer x
shifted by n.

{\tt GEN shifts(long s, long n)}: convert the long s into a PARI integer and
shift by n.

{\tt GEN shifti(GEN x, long n)}: shift by n the integer x.

{\tt GEN shiftr(GEN x, long n)}: shift by n the real x.

\subsec{Unary operations}

Let ``op'' be some unary operation of type GEN. The names and prototypes of
the low-level functions corresponding to op will be as follows.

{\tt GEN mpop(GEN x)}: create the result of op applied to the integer or real
x.

{\tt GEN ops(long s)}: create the result of op applied to the long s.

{\tt GEN opi(GEN x)}: create the result of op applied to the integer x.

{\tt GEN opr(GEN x)}: create the result of op applied to the real x.

{\tt void mpopz(GEN x, GEN z)}: assign into the integer or real z the result
of applying op to the integer or real x.

Remark: it has not been considered useful to include in PARI the functions
{\tt void opsz(long s, GEN z)}, {\tt void opiz(GEN x, GEN z)} and 
{\tt void oprz(GEN x, GEN z)}.
\smallskip
This can be applied to the following operators.

op=neg: negation ($-$x). The result is the same type as x.

op=abs: absolute value ($|\text{x}|$). The result is the same type as x.

In addition, there exist the following special unary functions with assignment.

{\tt void mpinvz(GEN x, GEN z)}: assign the inverse of the integer or real x
into the real z. The inverse is computed as a quotient of real numbers, not as
a Euclidean division.

{\tt void mpinvsr(long s, GEN z)}: assign the inverse of the long s into the
real z.

{\tt void mpinvir(GEN x, GEN z)}: assign the inverse of the integer x into the
real z.

{\tt void mpinvrr(GEN x, GEN z)}: assign the inverse of the real x into the
real z.

\subsec{Comparison operators}

{\tt long mpcmp(GEN x, GEN y)}: compare the integer or real x with the integer
or real y. The result is the sign of x$-$y.

{\tt long cmpss(long s, long t)}: sign of s$-$t.

{\tt long cmpsi(long s, GEN x)}: compare the long s with the integer x.

{\tt long cmpsr(long s, GEN x)}: compare the long s with the real x.

{\tt long cmpis(GEN x, long s)}: compare the integer x with the long s.

{\tt long cmpii(GEN x, GEN y)}: compare the integer x with the integer y.

{\tt long cmpir(GEN x, GEN y)}: compare the integer x with the real y.

{\tt long cmprs(GEN x, long s)}: compare the real x with the long s.

{\tt long cmpri(GEN x, GEN y)}: compare the real x with the integer y.

{\tt long cmprr(GEN x, GEN y)}: compare the real x with the real y.

\subsec{Binary operations}

Let ``op'' be some operation of type GEN. The names and prototypes of the 
low-level functions corresponding to op will be as follows.

{\tt GEN mpop(GEN x, GEN y)}: create the result of op applied to the integer
or reals x and y.

{\tt GEN opss(long s, long t)}: create the result of op applied to the longs
s and t.

{\tt GEN opsi(long s, GEN x)}: create the result of op applied to the long s
and the integer x.

{\tt GEN opsr(long s, GEN x)}: create the result of op applied to the long s
and the real x.

{\tt GEN opis(GEN x, long s)}: create the result of op applied to the integer x
and the long s.

{\tt GEN opii(GEN x, GEN y)}: create the result of op applied to the integers x
and y.

{\tt GEN opir(GEN x, GEN y)}: create the result of op applied to the integer x
and the real y.

{\tt GEN oprs(GEN x, long s)}: create the result of op applied to the real x
and the long s.

{\tt GEN opri(GEN x, GEN y)}: create the result of op applied to the real x
and the integer y.

{\tt GEN oprr(GEN x, GEN y)}: create the result of op applied to the reals x
and y.
\smallskip
{\tt GEN mpopz(GEN x, GEN y, GEN z)}: assign into the integer or real z
the result of op applied to the integer or reals x and y.

{\tt GEN opssz(long s, long t, GEN z)}: assign into the integer or real z
the result of op applied to the longs s and t.

{\tt GEN opsiz(long s, GEN x, GEN z)}: assign into the integer or real z
the result of op applied to the long s and the integer x.

{\tt GEN opsrz(long s, GEN x, GEN z)}: assign into the integer or real z
the result of op applied to the long s and the real x.

{\tt GEN opisz(GEN x, long s, GEN z)}: assign into the integer or real z
the result of op applied to the integer x and the long s.

{\tt GEN opiiz(GEN x, GEN y, GEN z)}: assign into the integer or real z
the result of op applied to the integers x and y.

{\tt GEN opirz(GEN x, GEN y, GEN z)}: assign into the integer or real z
the result of op applied to the integer x and the real y.

{\tt GEN oprsz(GEN x, long s, GEN z)}: assign into the integer or real z
the result of op applied to the real x and the long s.

{\tt GEN opriz(GEN x, GEN y, GEN z)}: assign into the integer or real z
the result of op applied to the real x and the integer y.

{\tt GEN oprrz(GEN x, GEN y, GEN z)}: assign into the integer or real z
the result of op applied to the reals x and y.
\smallskip
This can be applied to the following operators.

op=add: addition (x$+$y). The result is real except if both x and y are
integers.

op=sub: subtraction (x$-$y). The result is real except if both x and y are
integers.

op=mul: multiplication (x$*$y). The result is real except if both x and y are
integers OR if x or y is the integer or long zero.

op=div: division (x$/$y). In the case where x and y are both integers or longs,
the result is the Euclidean quotient, the remainder being of the sign of the
dividend x. If one of x or y is real, the result is real except if x is the
integer or long zero. Division by zero error if y is equal to zero.

op=res: remainder (x$\%$y). This operation exists only when x and y are longs
or integers. The result is the Euclidean remainder corresponding to div, i.e.
its sign is that of the dividend x. The result is always an integer.

op=mod: remainder (x$\%$y). This operation exists only when x and y are longs
or integers. The result is the true Euclidean remainder, i.e. nonnegative
and less than the absolute value of y.

\subsec{Division with remainder}

{\tt GEN dvmdss(long s, long t, GEN$*$ r}: create the Euclidean quotient and
remainder of the longs s and t. If r is not 0 or $-1$. put the address of the
remainder in r, and output the quotient. If r is equal to 0, only the
quotient is output. If r is equal to $-1$, the remainder is output instead
of the quotient. The remainder is always of the sign of the dividend s.

{\tt GEN dvmdsi(long s, GEN x, GEN$*$ r)}: create the Euclidean quotient and
remainder of the long s by the integer x.

{\tt GEN dvmdis(GEN x, long s, GEN$*$ r)}: create the Euclidean quotient and
remainder of the integer x by the long s.

{\tt GEN dvmdii(GEN x, GEN y, GEN$*$ r)}: create the Euclidean quotient and
remainder of the integer x by the integer y.

{\tt void mpdvmdz(GEN x, GEN y, GEN z, GEN$*$ r)}: assign into the integer or
real z the Euclidean quotient of the integers x by y, putting the address of 
the remainder in r (except if r is equal to 0 or $-1$ as above).

{\tt void dvmdssz(long s, long t, GEN z, GEN$*$ r)}: assign into the integer or
real z the Euclidean quotient of the longs s by t, putting the address of 
the remainder in r (except if r is equal to 0 or $-1$ as above).

{\tt void dvmdsiz(long s, GEN x, GEN z, GEN$*$ r}: assign into the integer or
real z the Euclidean quotient of the long s by the integer x, putting the
address of the remainder in r (except if r is equal to 0 or $-1$ as above).

{\tt void dvmdisz(GEN x, long s, GEN z, GEN$*$ r)}: assign into the integer or
real z the Euclidean quotient of the integer x by the long s, putting the
address of the remainder in r (except if r is equal to 0 or $-1$ as above).

{\tt void dvmdiiz(GEN x, GEN y, GEN z, GEN$*$ r)}: assign into the integer or
real z the Euclidean quotient of the integers x by y, putting the address of 
the remainder in r (except if r is equal to 0 or $-1$ as above).

\subsec{Miscellaneous functions}

{\tt long mpdivis(GEN x, GEN y, GEN z)}: if the integer y divides the integer
x, assign the quotient in the integer z and return 1 (true), otherwise return
0 (false).

{\tt long divise(GEN x, GEN y)}: if the integer y divides the integer x, return
1 (true), otherwise return 0 (false).

{\tt GEN convi(GEN x)}: converts the absolute value of the integer x in base
$10^9$. The result is an integer z such that z[1] is set artificially to 
$-1$, z[2], \dots, z[m] contain the base $10^9$ digits of $|\text{x}|$, and
the output is the address of z[m$+$1].

{\tt GEN confrac(GEN x)}: converts the fractional part of the real x in base
$10^9$. The result is an array z on the PARI stack such that z[0] contains
the number of significant decimal digits, and z[1],\dots the base $10^9$
expansion of the fractional part of x.

Warning: since z is not a standard PARI object, the stack must be cleaned by
a simple avma$=$oldavma statement, and not by the {\tt gerepile} garbage
collector after execution of this function.

{\tt void addsii(long s, GEN x, GEN z)}: assign into the integer z the sum of
the long s and the integer x (essentially identical to {\tt addsiz} except
that z is specifically an integer).

{\tt void mulsii(long s, GEN x, GEN z)}: assign into the integer z the product
of the long s and the integer x (essentially identical to {\tt mulsiz} except
that z is specifically an integer).

{\tt long divisii(GEN x, long s, GEN z)}: assign into the integer z the 
Euclidean quotient of the integer x by the long s, and return the remainder
as a long.

\section{Level 2 kernel (operations on general PARI objects)}

In this section, functions followed by an assignment statement will be 
mentioned as in the following example:

{\tt GEN gadd[z](GEN x, GEN y[, GEN z])}: the explanation is given for
the function {\tt gadd}, and the brackets are there to say
that the {\tt void} function {\tt void gaddz(GEN x, GEN y, GEN z)} also
exists, where the result is assigned into z.

All these [z] functions are obtained using macros (see the file {\tt gencom.h})
hence if an extension of the list is desired, it is trivial to add the
suitable macro in the file {\tt gencom.h}.

The functions available to handle subunits are the following.

{\tt GEN compo(GEN x, long n)}: create a copy of the n-th true component (i.e.
not counting the codewords) of the object x.

{\tt GEN truecoeff(GEN x, long n)}: create a copy of the coefficient of degree
n of x if x is a scalar, polynomial or power series, otherwise of the n-th
component of x.

{\tt long coeff(GEN x, long i, long j)}: x being a matrix (type 19), this gives
the address of the coefficient of row i and column j of the matrix x. This is
NOT a function, simply a macro, and can be used on the left or right hand side
of an assignment statement.

\subsec{Copying and conversion}

{\tt GEN cgetp(GEN x)}: create p-adic space sufficient to put x into, and
set the prime p and the p-adic precision to that of x, but do not copy x.

{\tt GEN gcopy(GEN x)}: create a new copy of the object x on the PARI stack.

{\tt GEN forcecopy(GEN x)}: same as copy except that even permanent pointers
are copied onto the stack.

{\tt long taille(GEN x)}: total number of 32-bit words occupied by the tree
representing x.

{\tt GEN gclone(GEN x)}: create a new copy of the object x on the heap.

{\tt GEN greffe(GEN x, long l)}: x being a polynomial (type 10), create a
power series (type 11) of length l starting with x, but without actually
copying the coefficients, just the pointers. For internal use.

{\tt double rtodbl(GEN x)}: x being a real (type 2), convert x into a
C double if possible.

{\tt GEN dbltor(double x)}: convert the C double x into a PARI real.

{\tt double gtodouble(GEN x)}: x being in the real numbers (but not 
necessarily of type real), convert x into a C double if possible. 

{\tt long gtolong(GEN x)}: x being in the integers (but not necessarily of type
integer), convert x into a C long if possible.

{\tt GEN gtopoly(GEN x, long v)}: convert or truncate the object x into a 
polynomial with main variable number v.

{\tt GEN gtopolyrev(GEN x, long v)}: convert or truncate the object x into a 
polynomial with main variable number v, but vectors are converted in reverse
order.

{\tt GEN gtoser(GEN x, long v)}: convert the object x into a power series
with main variable number v.

{\tt GEN gtovec(GEN x)}: convert the object x into a (row) vector.

{\tt GEN co8(GEN x, long l)}: x being a quadratic number (type 8), convert x
into a real number or complex number depending on the sign of the discriminant
of x, to precision l 32-bit words.

{\tt GEN gcvtop(GEN x, GEN p, long l)}: convert x into a p-adic number of
precision l.

{\tt GEN gmodulcp(GEN x, GEN y)}: create the object mod(x,y) on the PARI stack,
where x and y are either both integers, and the result is an integermod 
(type 3) or x is a scalar or a polynomial and y a polynomial, and the result is
a polymod (type 9).

{\tt GEN gmodulo(GEN x, GEN y)}: same as {\tt gmodulcp} except that the modulus
y is copied on the heap and not in the PARI stack.

{\tt long gexpo(GEN x)}: return the binary exponent of x or the maximal
binary exponent of the coefficients of x; return $-(2^{15}-1)$ if x has no
components. Error if x is exactly zero.

{\tt long gsize(GEN x)}: return 0 if x is exactly 0, otherwise return
{\tt gexpo(x)} multiplied by $\log(2)/\log(10)$. This gives a crude estimate
for the maximal number of decimal digits of the components of x.

{\tt long gsigne(GEN x)}: return the sign of x ($-1$, 0 or 1) when x is an 
integer, real or (irreducible or reducible) fraction. Error for all other
types.

{\tt long gvar(GEN x)}: return the main variable of x. If no component of
x is a polynomial or power series, return $2^{15}-1$.

{\tt int precision(GEN x)}: If x is of type real, return the precision of
x (the length of x in 32-bit words if x is not zero, and a reasonable
quantity obtained from the exponent of x if x is equal to zero). If x
is of type complex, return the minimum of the precisions of the real and
imaginary part. Otherwise, return 0 (coding in fact for $\infty$).

\subsec{Comparison operators and valuations}

{\tt int gcmp0(GEN x)}: 1 (true) if x is equal to 0, 0 (false) otherwise.

{\tt int isexactzero(GEN x)}: 1 (true) if x is exactly equal to 0, 0 (false)
otherwise.

{\tt int gcmp1(GEN x)}: 1 (true) if x is equal to 1, 0 (false) otherwise.

{\tt int gcmp\_1(GEN x)}: 1 (true) if x is equal to $-1$, 0 (false) otherwise.

{\tt long gcmp(GEN x, GEN y)}: comparison of x with y (sign of x$-$y).

{\tt long gcmpsg(long s, GEN x)}: comparison of the long s with x.

{\tt long gcmpgs(GEN x, long s)}: comparison of x with the long s.

{\tt long lexcmp(GEN x, GEN y)}: comparison of x with y for the lexicographic
ordering.

{\tt long gegal(GEN x, GEN y)}: 1 (true) if x is equal to y, 0 otherwise.

{\tt long gegalsg(long s, GEN x)}: 1 (true) if the long s is equal to x, 0
otherwise.

{\tt long gegalgs(GEN x, long s)}: 1 (true) if x is equal to the long s, 0
otherwise (it should not be very different from {\tt gegalsg}!).

{\tt long iscomplex(GEN x)}: 1 (true) if x is in the complex numbers but not
in the reals, 0 if x is in the reals, error if x is not in the complex numbers.

{\tt long ismonome(GEN x)}: 1 (true) if x is a nonzero monomial in its main
variable, 0 otherwise.

{\tt long ggval(GEN x, GEN p)}: greatest exponent of p dividing x, when this
makes sense.

{\tt long gval(GEN x, long v)}: highest power of the variable number v dividing
the polynomial x.

{\tt int pvaluation(GEN x, GEN p, GEN$*$r)}: x and p being integers, returns 
the highest power of p dividing x, creates the coprime to p part of x and 
returns its address in r.

\subsec{Assignment statements}

{\tt void gaffsg(long s, GEN x)}: assign the long s into the object x.

{\tt void gaffect(GEN x, GEN y)}: assign the object x into the object y.

\subsec{Unary operators}

{\tt GEN gneg[z](GEN x[, GEN z])}: create $-$x.

{\tt GEN gabs[z](GEN x[, GEN z])}: create $|\text{x}|$.

{\tt GEN gsqr(GEN x)}: create the square of x,

{\tt GEN ginv(GEN x)}: create the inverse of x.

{\tt GEN gfloor(GEN x)}: create the floor of x, i.e. the (true) integral part.

{\tt GEN gfrac(GEN x)}: create the fractional part of x, i.e. x minus the floor
of x.

{\tt GEN gceil(GEN x)}: create the ceiling of x.

{\tt GEN ground(GEN x)}: round the components of x to the nearest integer.

{\tt GEN grndtoi(GEN x, long$*$e)}: same as round, but in addition put in
$*$e minus the number of significant binary bits left after rounding. If
$*$e is positive, all significant bits are lost. This gives an error message
in {\tt ground} but not in {\tt grndtoi}.

{\tt GEN gtrunc(GEN x)}: truncate x. This is the (false) integer part if x is 
an integer, converts a series into a polynomial or rational function, takes
the polynomial part of a rational function.

{\tt GEN gcvtoi(GEN x, long$*$e)}: same as {\tt grndtoi} except that rounding
is replaced by truncation.

{\tt GEN gred[z](GEN x[, GEN z])}: reduce x to lowest terms if x is a fraction
or rational function (types 4, 5, 13 and 14), otherwise create a copy of x.

{\tt void gredsp(GEN$*$ x)}: reduce $*$x to lowest terms and put the result
in $*$x again. The type of the result may be changed to integer or polynomial.
A call to this function is the nomral way of finishing a computation whose
result gives a fraction or rational function which is not necessarily in
lowest terms.

{\tt GEN content(GEN x)}: create the GCD of all the components of x.

{\tt GEN primpart(GEN x)}: x divided by its content.

{\tt void normalize(GEN$*$ px)}: x=$*$px being an unnormalized power series
(i.e. type 11 with all coefficients correctly set except that x[2] is not
necessarily different from zero), normalize x correctly in place. For
internal use.

{\tt void normalizepol(GEN$*$ px)}: x=$*$px being an unnormalized polynomial
(i.e. type 10 with all coefficients correctly set except that x[2] is not
necessarily different from zero), normalize x correctly in place. For
internal use.

\subsec{Binary operators}

{\tt GEN gmax[z](GEN x, GEN y[, GEN z])}: create the maximum of the objects x
and y if they can be compared.

{\tt GEN gmaxsg[z](long s, GEN x[, GEN z])}: create the maximum of the long s
and the object x.

{\tt GEN gmaxgs[z](GEN x, long s[, GEN z])}: create the maximum of the object 
x and the long s.

{\tt GEN gmin[z](GEN x, GEN y[, GEN z])}: create the minimum of the objects x
and y if they can be compared.

{\tt GEN gminsg[z](long s, GEN x[, GEN z])}: create the minimum of the long s
and the object x.

{\tt GEN gmings[z](GEN x, long s[, GEN z])}: create the minimum of the object 
x and the long s.

{\tt GEN gadd[z](GEN x, GEN y[, GEN z])}: create the sum of the objects x and 
y.

{\tt GEN gaddsg[z](long s, GEN x[, GEN z])}: create the sum of the long s and 
the object x.

{\tt GEN gaddgs[z](GEN x, long s[, GEN z])}: create the sum of the object x 
and the long s.

{\tt GEN gaddpex(GEN x, GEN y)}: create the sum of the integer or fraction x
and the p-adic y. For internal use.

{\tt GEN gsub[z](GEN x, GEN y[, GEN z])}: create the difference of the objects
x and y.

{\tt GEN gsubgs[z](GEN x, long s[, GEN z])}: create the difference of the 
object x and the long s.

{\tt GEN gsubsg[z](long s, GEN x[, GEN z])}: create the difference of the long
s and the object x.

{\tt GEN gmul[z](GEN x, GEN y[, GEN z])}: create the product of the objects x 
and y.

{\tt GEN gmulsg[z](long s, GEN x[, GEN z])}: create the product of the long s 
with the object x.

{\tt GEN gmulgs[z](GEN x, long s[, GEN z])}: create the product of the object
x with the long s.

{\tt GEN gshift[z](GEN x, long n[, GEN z])}: left shift of the components of
x by n (if n is nonnegative) or right shift by $-$n (if n is negative). Applies
only to integers, reals and vectors/matrices of such. For other types, it is 
simply multiplication by $2^{\text{n}}$.

{\tt GEN gmul2n[z](GEN x, long n[, GEN z])}: create x multiplied by 
$2^{\text{n}}$. The difference with {\tt gshift} occurs when n is negative and
x is of type integer: {\tt gshift} truncates, while {\tt gmul2n} creates a
fraction if necessary.

{\tt GEN gdiv[z](GEN x, GEN y[, GEN z])}: create the quotient of the objects 
x and y.

{\tt GEN gdivgs[z](GEN x, long s[, GEN z])}: create the quotient of the object
x with the long s.

{\tt GEN gdivsg[z](long s, GEN x[, GEN z])}: create the quotient of the long s
with the object x.

{\tt GEN gdivent[z](GEN x, GEN y[, GEN z])}: create the true Euclidean 
quotient of x by the integer or polynomial y.

{\tt GEN gdiventsg[z](long s, GEN x[, GEN z])}: create the true Euclidean 
quotient of the long s by the integer x.

{\tt GEN gdiventgs[z](GEN x, long s[, GEN z])}: create the true Euclidean 
quotient of the integer x by the long s.

{\tt GEN gdiventres(GEN x, GEN y)}: create a 2-component vertical vector whose
components are the true Euclidean quotient and remainder of x by y.

{\tt GEN gdivmod(GEN x, GEN y, GEN$*$r)}: If r is not equal to 0 or $-1$,
create the (false) Euclidean quotient of x by y, and put the address of the 
remainder in r. If r is equal to 0, do not create the remainder, and if r is
equal to $-1$, create and output only the remainder.

{\tt GEN poldivres(GEN x, GEN y, GEN$*$r)}: same as {\tt gdivmod} but
specifically for x and y polynomials.

{\tt GEN gdeuc[z](GEN x, GEN y[, GEN z])}: create the Euclidean quotient of 
the polynomial x by the polynomial y.

{\tt GEN gmod[z](GEN x, GEN y[, GEN z])}: create the true remainder of x 
modulo the integer or polynomial y.

{\tt GEN gmodsg[z](long s, GEN x[, GEN z])}: create the true remainder of the
long s modulo the integer x.

{\tt GEN gmodgs[z](GEN x, long s[, GEN z])}: create the true remainder of the 
integer x modulo the long s.

{\tt GEN gres(GEN x, GEN y)}: create the Euclidean remainder of the polynomial
x divided by the polynomial y.

{\tt GEN ginvmod(GEN x, GEN y)}: create the inverse of x modulo y when it
exists.

{\tt GEN gpui(GEN x, GEN y, long l)}: create $\text{x}^{\text{y}}$. The
precision l is taken into account only if y is not an integer and x is an
exact object. If y is an integer, binary powering is done. Otherwise, the
result is $\exp(\text{y}*\log(\text{x}))$ computed to precision l.

{\tt GEN ggcd(GEN x, GEN y)}: create the GCD of x and y.

{\tt GEN glcm(GEN x, GEN y)}: create the LCM of x and y.

{\tt GEN subres(GEN x, GEN y)}: resultant of the polynomials x and y computed
using the subresultant algorithm.

{\tt GEN gpuigs(GEN x, long n, long l)}: create $\text{x}^{\text{n}}$ using
binary powering. The precision l is used {\sl only\/} when x is of type 15
(real quadratic forms), hence can almost always be ommited.

{\tt GEN gsubst(GEN x, long v, GEN y)}: substitute in x the object y for the 
variable number v.

{\tt int gdivise(GEN x, GEN y)}:  return 1 (true) if y divides x, 0 otherwise.

{\tt GEN gbezout(GEN x, GEN y, GEN$*$ u, GEN$*$ v)}: create the GCD of x and y,
and put in u and v the adresses of objects such that $*$ux$+$$*$vy$=\gcd$(x,y).


\vfill\eject