Re: C++ Matrix & Linear Algebra library


"Lou Pecora" <pecora@xxxxxxxxxxxxxxxxxx> wrote in message
news:pecora-EF822E.14420122082008@xxxxxxxxxxxxxxxxxx
In article
<f4f784af-5c62-411c-9c1a-d4f1f08caf40@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Evgenii Rudnyi <usenet@xxxxxxxxx> wrote:

On Aug 22, 9:12 am, Gert Van den Eynde <gvdey...@xxxxxxxxx> wrote:
Dear Evgenii,
It seems that, even several years after the introduction of C++ in
scientific computing, people have moved back to "old style" Fortran
because "the software is already there and it works" or they go for
"full option packages" like Matlab, Mathematica, Maple,...

I would say that the reason is not Fortran but rather Matlab,
Mathematica, Maple,.... What is necessary for research is actually
rapid prototyping and neither Fortran nor C++ are qualified for this.

A good student after a month or two time can obtain something useful
in Matlab or Mathematica - I mean from a scientific viewpoint. What he/
she could have done for a comparable time in Fortran/C++?

Also there is a trend for the software development that there should
be a scripting gluing all pieces together. This also changes the way
the software is developed. C++ is left for relatively low level things
where comfortability is not that important. It is after all for a Real
Programmer. The comfort is for the scripting.

Sounds like a job for Python (and/or it's friend Sage). Check it out.
Rapid prototyping, easy to learn language which is a good glue and has
come to often be used as such in the coding community in general as well
as science community. Just do a Google and check out SourceForge for
starters.

--
-- Lou Pecora

Hi Lou

A pox on fortran!! Iused it in 1966 and as soon as C came along I moved up
to that. Then when C++ came along, I moved to that. Each time I moved up
my efficiency went up by 4. So I am 16 times more efficient in C++. There's
a reason why major applications like Microsoft Office, Firefox, and
Internet Explorer are writen in C++ --- efficiency !! --- fewer lines of
code and less programming time. You can get all the functionality of LAPACK
in
8000 lines of C++. LAPACK has >100,000 lines. Fewer lines means more
reliable code easier to maintain.
The new multicore processors will cause major work in multithreading. I
bet the code will bei in C++. I will sit quietly and wait for the Standard
Template Library
to be multithreaded. Bingo---- just plug it in.
Rumors are that a new exotic version of C++ is developed and that
Nvidia has
a super cruncher card that can spawn 10,000 threads gee! one thread for
each
column of a matrix. I wonder what ppLinear can do with that.

regards pecos



.

Relevant Pages

  • Re: C++ Matrix & Linear Algebra library
    ... On this list used to be a reasonable effort in ... "full option packages" like Matlab, Mathematica, Maple,... ... I would say that the reason is not Fortran but rather Matlab, ...
    (sci.math.num-analysis)
  • Re: C++ Matrix & Linear Algebra library
    ... scientific computing, people have moved back to "old style" Fortran ... "full option packages" like Matlab, Mathematica, Maple,... ... I would say that the reason is not Fortran but rather Matlab, ...
    (sci.math.num-analysis)
  • Re: What do you LISPers think of Haskell?
    ... modern CL implementations is in the same league as C and Fortran". ... in the kind of scientific computing I ...  This is where Lisp has a spectacular ... good debugging facilities if something goes wrong. ...
    (comp.lang.lisp)
  • Re: HPC Considered Harmful
    ... Helps to explain many nasty Fortran programs I've seen! ... As computers and programming languages evolve, ... IMO commercial and scientific computing do have some things in ... In other forums others often trash older languages like Fortran ...
    (comp.lang.fortran)
  • Re: Fortran vs C++ for scientific computing
    ... Blitz++ library performs better than Fortran (e.g. lattice QCD ... This subroutine consists of three nested loops for handling 3-D arrays. ... optimization of scientific code. ... Fortran vs C++ for scientific computing? ...
    (comp.lang.fortran)