tkoenig1 / fmpfr

A Fortran binding for the GNU MPFR library.

FMPFR - a Fortran binding for MPFR

This library offers a Fortran binding to the GNU MPFR Library multi-precision library. The aim of the library is to make the use of MPFR convenient for a Fortran user.

Status

This is an initial work, expect bugs and (especially) missing features and possibly some instability, at least in binary compatibility. Source code compatibility should be fine.

Compiler requirements

As of now, you need a compiler which supports `ERROR STOP for the testsuite.

What is supported

• Arithmetic expressions can be used the usual way; expressions can be mixed with real and integer expressions.

• Mathematical intrinsics, like sin, cos etc.

• Array expressions

• Setting and changing a default precision of variables

• List-directed output

• Reading values from strings

What is not supported

• Input (user-defined I/O is too unevenly implemented in existing compilers)

• Array intrinsics like MAXLOC and MAXVAL

An example

Here is a short exampe, if your compiler supports user-defined I/O:

program memain
  use fmpfr_oper
  implicit none
  type (fmpfr), dimension(:), allocatable :: a
  call set_default_prec (128)
  allocate (a(2))
  a(1) = fmpfr("1.3")
  a(2) = a(1) + 2
  print *,a
  print *,sin(a)
  print *,get_str(cos(a(1)),10)
end program memain

whose output is

 1.299999999999999999999999999999999999998E0  3.299999999999999999999999999999999999998E0 
 9.635581854171929647013486300395548153418E-1  -1.577456941432483820116542776024823708430E-1 
 2.674988286E-1

Installation

Type ./configure and make. If everything was successful, type make check.

Modifying the code

If you want to modify the code, it is best if you make your changes to the src/generate.pl file and run

$ grep ^@deftypefun mfpr.texi | perl generate.pl

where mpfr.texi is the MPFR documentation.

License

This is covered under the MIT license.

fmpfr's Issues

Improve handling of precision and rounding

It would be nice to be able to specify precision and rounding, both optional, to allow something like FMPFR("1.23456",100,mpfr_rndz) where both the 100 and the mpfr_rndz are optional.

Have to think a bit how Fortran generics rules allow for that...

Blocker: Problem with library with -O2

Seems the library is miscompiled with -O2.

So, it seems not to be usable at the moment. Investigating...

Enable a subset for older compilers

Right now, the program needs derived type I/O and ERROR STOP (the latter only for the testsuite). It could still be useful for older compilers. It would be better to not compile those parts, or chose an alternative.

add topics

I suggest adding topics such as mpfr, multiprecision in the About section at https://github.com/tkoenig1/FMPFR, Thanks for your project.

Create test cases

Issues Using FMPFR

Hi!
I am having some issues using your wrapper library that I am hoping you can clear up. I have successfully installed and configured FMPFR, but it does not work. The main problem is within the src files of FMPR itself. When a function in an FMPFR file is called that is defined within a different file, the whole thing crashes. It appears as if there has been no effort to link the files within the library together. Now, I will say that I have never had to install a library like this before. This means that there is a 99% chance that this is my fault and I am just lacking the knowledge of how to correctly call and use the wrapper. If that is the case, then some advice and assistance would be GREATLY appreciated!

memory leaks due to buggy finalization in gfortran

The interface is essentially unusable when using gfortran (Version 12.1.1), because it does not handle finalization properly (compiler bug). So on each assignment, or for temporary results from operators in formulas, memory is leaked. You can check this easily with valgrind.