# $Id$

This small package, only composed of header files, is used to offer a generic
interface to different numbers representation. It took its inspiration from
pk_int (integers in Newpolka library from Bertrand Jeannet) and oct_num
(integers in Octagon Abstract Domain library from Antoine Min).

It requires the GMP and MPFR libraries. If bounds are used, it also requires
the APRON library.

It may be freely distributed and/or modified under the conditions of the LGPL
license.

Numbers
=======

The main file is num.h. Its inclusion should be done after one of the following
macros has been defined:

- NUM_LONGINT		long int
- NUM_LONGLONGINT	long long int
- NUM_MPZ		mpz_t (GMP library)
	Defines num_t, numint_t.
	Possible operations are num_XXX and numint_XXX 
	(signatures in num.h and numint.h)

- NUM_LONGRAT		rationals using long int
- NUM_LONGLONGRAT	rationals using long long int
- NUM_MPQ		rationals using mpq_t (GMP library)
	Defines num_t, numrat_t
	Possibles operations are num_XXX and numrat_XXX
	(signatures in num.h and numrat.h)
	Also defines numint_t (the underlying integer types).

- NUM_DOUBLE		double
- NUM_LONGDOUBLE	long double
- NUM_MPFR              floating-point of arbitrary precision (MPFR library)
	Defines num num_t, numflt_t
	Possibles operations are num_XXX and numflt_XXX
	(signatures in num.h and numflt.h)
	
Bounds
======

Bounds are *superior* bounds defined on top of num_t. They add +oo to num_t.
First include properly num.h with the right macro, and then include bound.h.
Signatures of operations in bound.h

It requires the APRON library (more precisely, the module ap_scalar, for
conversion functions).
