Open Scientific Library -- libosl.org down

[NickFort]

Hi all,

I've been trying to track down the Open Scientific Library for a while now; it is officially hosted at

http://libosl.org,

but the site has been down for about a week now, as far as I can gather. I've emailed the guy who created it, but have yet to receive a reply. I don't want to pester him with more emails, so I thought I'd ask you guys: does anyone have a copy of the library (probably called libosl.tar.gz) lying around on their hard drive?

I can't find any online mirrors of it either.

I know it's a long shot, but I thought I'd try anyway.

Thanks!
NickFort

--------------------------------------
Background: Chemical engineer, familiar mostly with MATLAB, but now branching out into real programming.

 

[mikrom]

I have no knowledge about this library, but maybe it's not very often used when the server is permanently down.

Maybe it would be better when you say what special problem you need to solve: Linear algebra, ODEs, PDEs, optimization (linear, non linear), ...
As I worked with numerical mathematics (it's about 8 years ago) I used to search for the software in Netlib Repository. It contains the essential packages as BLAS, LAPACK, ODEPACK, etc.

 

[NickFort]

Well, the library was released to the public in September 2009, so it hasn't really had a chance to succeed or flop.

I know of Netlib, and I've had a look at it before, but the idea of (a) an all-in-one library, and (b) one written in Fortran 95 (as opposed to F77, which is what most of Netlib appears to be) is what I find appealing.

To answer, "Maybe it would be better when you say what special problem you need to solve," -- what I want is the to have a library with much of the same basic functionality as MATLAB, most of which is what you mentioned above. I think the F77 aspect of it is what I find most off-putting, what with it's common blocks instead of modules, etc. but maybe it's just time to get over my F77 phobia...

--------------------------------------
Background: Chemical engineer, familiar mostly with MATLAB, but now branching out into real programming.

 

[mikrom]

Many of the free Fortran77 numerical libraries was later rewritten in C or C++, so they are now available in both languages: Fortran77 or C/C++. I don't know exactly the situation with Fortran90/95, but it seems, that there are not so much numerical libs freely available as it was with Fortran77.
For example in your previous thread

http://www.tek-tips.com/viewthread.cfm?qid=1609150&page=1

you wrote about GNU scientific library. This library is written in C - available here

http://www.gnu.org/software/gsl/

and has only the Fortran90 interface available here

http://www.lrz.de/services/software/mathematik/gsl/fortran/

On the 1. link above they write that "A compiled version of GSL is available as part of Cygwin on Windows...". However, when you want to create windows native executables, you need to compile it under MinGW.
Btw, I found here an older version 1.8 of GSL binaries for windows:

http://gnuwin32.sourceforge.net/packages/gsl.htm

The actual version of GSL is 1.14
Here is the installer for 1.13

http://ascendwiki.cheme.cmu.edu/Binary_installer_for_GSL-1.13_on_MinGW

and here seems to be the package for version 1.14 mingw32-gsl-1.14-1.zip available:

http://sourceforge.net/projects/mingw-cross/files/

You can try the binary builds above or try to bulid the library self under MinGW.

I found here a free pure Fortran90 library

http://sourceforge.net/projects/afnl/

but it's smaller in comparition with GSL.

 

[NickFort]

Hey mikrom,

Thanks for some of that info; admittedly, I hadn't found either the GSL binaries or the AFNL.

I've had a more serious look at both now, and GSL has a lot of stuff -- well-programmed, I'm sure -- the way of using it with F90 is too convoluted, and would involve me adapting all the code in the project heavily. I'm still considering this option.

AFNL is the kind of thing I was looking for, but it's not great, unfortunately: very much the "textbook" sort of algorithms, and not even the more advanced ones you can find in numerical methods textbooks. Numerical Recipes have better code, although also licensing issues. I honestly don't know if OSL is any better, but I thought it might be worth a shot.

Thanks for hunting down those links, though.

--------------------------------------
Background: Chemical engineer, familiar mostly with MATLAB, but now branching out into real programming.