flint.h – global definitions¶
Macros¶
The file flint.h
contains various useful macros.

__FLINT_VERSION¶

__FLINT_VERSION_MINOR¶

__FLINT_VERSION_PATCHLEVEL¶
The major, minor and patch for current version of FLINT.

__FLINT_RELEASE¶
Equivalent to
10000 * __FLINT_VERSION + 100 * __FLINT_VERSION_MINOR + __FLINT_VERSION_PATCHLEVEL
.

FLINT_VERSION¶
A static text string giving the version number, e.g.
3.1.0
or3.2.0dev
.

FLINT_BITS¶
The constant defining how many bits per limb on the machine. We require this to be either 32 or 64. This constant is set during the configuration.

FLINT_D_BITS¶
A constant set at compile time to be the number of bits per double on the machine or one less than the number of bits per limb, whichever is smaller. This will have the value 31 on 32bit systems and 53 on 64bit systems. Numerous internal functions using precomputed inverses only support operands up to
FLINT_D_BITS
bits, hence the macro.

FLINT_ABS(x)¶
Returns the absolute value of x for primitive signed numerical types. It might fail for least negative values such as INT_MIN and LONG_MIN.

FLINT_MIN(x, y)¶

FLINT_MAX(x, y)¶
Returns the minimum or maximum of x and y for primitive types. This macro is only safe to use when x and y are of the same type, to avoid problems with integer promotion.

FLINT_SWAP(T, x, y)¶
Swaps x and y, both of types T. For instance, with x and y of type
fmpz_poly_t
, one can writeFLINT_SWAP(fmpz_poly_struct, *x, *y)
to swap the content of x with the content of y.
Integer types¶
The char, short and int types are assumed to be two’s complement types with exactly 8, 16 and 32 bits. Although this is not guaranteed prior to C23, it is true on all mainstream platforms prior to this.
Since the C types long and unsigned long do not have a standardised size in practice, FLINT defines slong and ulong types which are guaranteed to be 32 bits on a 32bit system and 64 bits on a 64bit system. They are also guaranteed to have the same size as GMP’s mp_limb_t. GMP builds with a different limb size configuration are not supported at all.

type ulong¶
The ulong type is used for integervalued coefficients that are known to be unsigned, and for values that require the full 32bit or 64bit range. In method names, a ulong parameter is denoted by ui, for example
arb_add_ui()
.The constant UWORD_MAX gives the range of this type. This type can be printed with flint_printf using the format string
%wu
.This is equivalent to GMP’s mp_limb_t.

type slong¶
The slong type is used for precisions, loop indices, array sizes, and the like, even when those values are known to be nonnegative. It is also used for small integervalued coefficients. In method names, an slong parameter is denoted by si, for example
arb_add_si()
.This type can be printed with flint_printf using the format string
%wd
or%{slong}
.This is equivalent to GMP’s mp_limb_signed_t. Furthermore, for UNIXtype systems it is also equivalent to mp_size_t.

UWORD_MIN¶

UWORD_MAX¶

WORD_MIN¶

WORD_MAX¶
The minimum and maximum values that a ulong and slong can hold, respectively.

type flint_bitcnt_t¶
A bit offset within an array of limbs (always nonnegative).

type nn_ptr¶
Pointer to a writable array of limbs.
This is equivalent to GMP’s mp_ptr.

type nn_srcptr¶
Pointer to a readonly array of limbs.
This is equivalent to GMP’s mp_srcptr.
Allocation Functions¶

void *flint_malloc(size_t size)¶
Allocate size bytes of memory.

void *flint_realloc(void *ptr, size_t size)¶
Reallocate an area of memory previously allocated by
flint_malloc()
,flint_realloc()
, orflint_calloc()
.

void *flint_calloc(size_t num, size_t size)¶
Allocate num objects of size bytes each, and zero the allocated memory.

void flint_free(void *ptr)¶
Free a section of memory allocated by
flint_malloc()
,flint_realloc()
, orflint_calloc()
.
Random Numbers¶

type flint_rand_struct¶
A structure holding the state of the FLINT pseudo random number generator.

type flint_rand_t¶
An array of length 1 of
flint_rand_struct
.

void flint_rand_init(flint_rand_t state)¶

void flint_rand_clear(flint_rand_t state)¶
Initialises or clears a
flint_rand_t
:.
Thread functions¶

void flint_set_num_threads(int num_threads)¶
Set up a thread pool of
num_threads  1
worker threads (in addition to the master thread) and set the maximum number of worker threads the master thread can start tonum_threads  1
.This function may only be called globally from the master thread. It can also be called at a global level to change the size of the thread pool, but an exception is raised if the thread pool is in use (threads have been woken but not given back). The function cannot be called from inside worker threads.

int flint_get_num_threads(void)¶
When called at the global level, this function returns one more than the number of worker threads in the Flint thread pool, i.e. it returns the number of workers in the thread pool plus one for the master thread.
In general, this function returns one more than the number of additional worker threads that can be started by the current thread.
Use
thread_pool_wake()
to set this number for a given worker thread.See also:
flint_get_num_available_threads()
.

int flint_set_num_workers(int num_workers)¶
Restricts the number of worker threads that can be started by the current thread to
num_workers
. This function can be called from any thread.Assumes that the Flint thread pool is already set up.
The function returns the old number of worker threads that can be started.
The function can only be used to reduce the number of workers that can be started from a thread. It cannot be used to increase the number. If a higher number is passed, the function has no effect.
The number of workers must be restored to the original value by a call to
flint_reset_num_workers()
before the thread is returned to the thread pool.The main use of this function and
flint_reset_num_workers()
is to cheaply and temporarily restrict the number of workers that can be started, e.g. by a function that one wishes to call from a thread, and cheaply restore the number of workers to its original value before exiting the current thread.

void flint_reset_num_workers(int num_workers)¶
After a call to
flint_set_num_workers()
this function must be called to set the number of workers that may be started by the current thread back to its original value.
Input/Output¶

int flint_printf(const char *format, ...)¶

int flint_fprintf(FILE *fs, const char *format, ...)¶

int flint_vprintf(const char *format, va_list vlist)¶

int flint_vfprintf(FILE *fs, const char *format, va_list vlist)¶
These functions are extensions of the C standard library functions
printf
,fprintf
,vprintf
, andvfprintf
.The first extension is the addition of the length modifier
w
, used for printing the typesulong
,slong
andulong
. As these types are either defined as signed and unsignedlong int
orlong long int
, this comes in handy. Just likelong int
andlong long int
, the conversion format specifier are allowed to bed
,i
,o
,x
,X
andu
.The second and final extension is printing of FLINT types. Currently supported types are the base types
ulong
,slong
,fmpz_t
,fmpq_t
,mag_t
,arf_t
,arb_t
andacb_t
as well as the context structures for modulo arithmeticnmod_t
andfmpz_mod_ctx_t
. We also support the GMP typesmpz_t
andmpq_t
.We currently support printing vectors of pointers to the following base types:
slong
,ulong
,fmpz
,fmpq
,mag_struct
,arf_struct
,arb_struct
andacb_struct
.We also support printing matrices of the following types:
nmod_mat_t
,fmpz_mat_t
,fmpq_mat_t
,arb_mat_t
andacb_mat_t
.Finally, we currently support printing polynomial of the following types:
nmod_poly_t
,fmpz_poly_t
,fmpq_poly_t
,arb_poly_t
andacb_poly_t
.
ulong bulong;
slong bslong;
fmpz_t bfmpz;
fmpq_t bfmpq;
mag_t bmag;
arf_t barf;
arb_t barb;
acb_t bacb;
nmod_t bnmod;
fmpz_mod_ctx_t bfmpz_mod_ctx;
mpz_t bmpz;
mpq_t bmpq;
/* Initialize and set variables */
flint_printf(
"ulong: %{ulong}\n"
"slong: %{slong}\n"
"fmpz: %{fmpz}\n"
"fmpq: %{fmpq}\n"
"mag: %{mag}\n"
"arf: %{arf}\n"
"arb: %{arb}\n"
"acb: %{acb}\n"
"nmod: %{nmod}\n"
"fmpz_mod_ctx: %{fmpz_mod_ctx}\n"
"mpz: %{mpz}\n"
"mpq: %{mpq}\n",
bulong,
bslong,
bfmpz,
bfmpq,
bmag,
barf,
barb,
bacb,
bnmod,
bfmpz_mod_ctx,
bmpz,
bmpq);
slong * vslong; slong vslong_len;
nn_ptr vnmod; slong vnmod_len; /* The base type for nmod is ulong */
fmpz * vfmpz; slong vfmpz_len;
/* fmpz_mod vectors are given by the type `fmpz *' */
fmpq * vfmpq; slong vfmpq_len;
mag_ptr vmag; slong vmag_len;
arf_ptr varf; slong varf_len;
arb_ptr varb; slong varb_len;
acb_ptr vacb; slong vacb_len;
/* Initialize and set variables */
flint_printf(
"slong vector: %{slong*}\n"
"nmod vector: %{ulong*}\n"
"fmpz vector: %{fmpz*}\n"
"fmpq vector: %{fmpq*}\n"
"mag vector: %{mag*}\n"
"arf vector: %{arf*}\n"
"arb vector: %{arb*}\n"
"acb vector: %{acb*}\n"
vslong, vslong_len, /* They require a vector length specifier */
vnmod, vnmod_len,
vfmpz, vfmpz_len,
vfmpq, vfmpq_len,
vmag, vmag_len,
varf, varf_len,
varb, varb_len,
vacb, vacb_len);
nmod_mat_t mnmod;
fmpz_mat_t mfmpz;
fmpz_mod_mat_t mfmpz_mod;
fmpq_mat_t mfmpq;
arb_mat_t marb;
acb_mat_t macb;
/* Initialize and set variables */
flint_printf(
"nmod matrix: %{nmod_mat}\n"
"fmpz matrix: %{fmpz_mat}\n"
"fmpz_mod matrix: %{fmpz_mod_mat}\n"
"fmpq matrix: %{fmpq_mat}\n"
"arb vector: %{arb_mat}\n"
"acb vector: %{acb_mat}\n"
mnmod,
mfmpz,
mfmpz_mod,
mfmpq,
marb,
macb);
nmod_poly_t pnmod;
fmpz_poly_t pfmpz;
fmpz_mod_poly_t pfmpz_mod;
fmpq_poly_t pfmpq;
arb_poly_t parb;
acb_poly_t pacb;
/* Initialize and set variables */
flint_printf(
"nmod polynomial: %{nmod_poly}\n"
"fmpz polynomial: %{fmpz_poly}\n"
"fmpz_mod polynomial: %{fmpz_mod_poly}\n"
"fmpq polynomial: %{fmpq_poly}\n"
"arb polynomial: %{arb_poly}\n"
"acb polynomial: %{acb_poly}\n"
pnmod,
pfmpz,
pfmpz_mod,
pfmpq,
parb,
pacb);
Note
Printing of FLINT types does not currently support any flags.
Note
Any use of %n
flags will be invalid, but will not generate any error.
Note
Invalid formats using variable minimum field width and/or precision such as
"%* p"
may be wrongly parsed, and may result in a different result
compared to the C standard library functions.

int flint_sprintf(char *s, const char *str, ...)¶
This functions is an extensions of the C standard library functions
sprintf
. It is currently advised to not use this function as it is currently not coherent withflint_printf()
.
Exceptions¶

void flint_abort(void)¶
FLINT version of the C standard function
abort
.

void flint_set_abort(void (*func)(void))¶
Sets the
flint_abort()
function to callfunc
instead ofabort
.

enum flint_err_t¶
An error code with one of the following values

FLINT_ERROR¶
Describes a generic error.

FLINT_OVERFLOW¶
Describes an overflow.

FLINT_IMPINV¶
Describes an impossible inversion.

FLINT_DOMERR¶
Describes a domain error.

FLINT_DIVZERO¶
Describes a division by zero.

FLINT_EXPOF¶
Describes a exponent overflow.

FLINT_INEXACT¶
Describes an inexact operation.

FLINT_TEST_FAIL¶
Describes a test fail.

FLINT_ERROR¶

void flint_throw(flint_err_t exc, const char *msg, ...)¶
Throws an error of type
exc
with messagemsg
and aborts viaflint_abort()
. The printing backend function isflint_fprintf()
, and so it allows for printing of FLINT types as well.