In reply to this post by Boost - Users mailing list
On 26/10/2018 06:03, A Z via Boost-users wrote:
> I am aware the the boost multiprecision library is a headers library.
> -Is Boost multiprecision accurately arbitrary precision, ie.
> can have its number precisions and scales always made to a larger number,
> contingent only to the amount of available OS memory?
This is a very vague question. But:
* Yes for some integer types - for example cpp_int and mpz_int are both
true arbitrary precision. But...
* There are no arbitrary precision floating point types - such types do
not really exist in any case (how many digits does Pi have??). But...
* There are *variable* precision floating point types - for example
mpfr_float where you set the current working precision with the
::precision and ::default_precision member functions.
> -If I want to use boost Boost multiprecision on its own,
> what is the name of the Boost .dll or .lib files that have all the headers
> in it? I need a minimal set of files in order to run multiprecision
> on its own.
Huh? .dll's and .libs do not "contain" headers.
On it's own, Boost.Multiprecision *does not need any external dll's or
.lib's* - that's what "header only" means. But....
Some types in the library may make use of external libraries.... GMP,
MPFR etc. You will need to look at the documentation for those
libraries for the linking options required, but in general it's pretty
simple on Unix like systems:
Anything from boost/multiprecision/gmp.hpp requires -lgmp
Anything from boost/multiprecision/mpfr.hpp requires -lmpfr -lgmp
Anything from boost/multiprecision/mpfi.hpp requires -lmpfi -lmpfr -lgmp