MathGroup Archive 2008

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Mathematica and F#

peter lindsay wrote:
> not in my experience - no, not at all. Much of the work my colleagues are
> involved in uses Fortran, IDL and to a much lesser extent C++ with
> occasional scripting in perl or python. My Statistics colleagues use SAS
> and my pure maths colleagues use GAP and Haskell amongst others. However,
> my point was: does anyone really want to tie them selves in to Microsoft .
> By the way, we also tend not to massage data.

I am surprised. My experience of scientific computing is very different. We
used a handful of monolithic applications long-since written in Fortran and
spent all of our time writing many small programs in C++, Mathematica and
OCaml to analyze the results. Massaging was required when other academic
groups gave us data generated by other monolithic programs (albeit often
ones they developed in-house) than were in an incompatible format.

>>The risks associated with vendor lock-in are then tiny, of course.
> with respect - not at all. The risks of lock-in are high and I would
> recommend anyone considering "buying-in" completely  to some proprietary
> brand [ even including Mathematica ] to be extremely careful if their
> professional careers are to depend on that vendor.

Lock-in does not tie your career to a particular platform, only one of your
programs. For example, we are "tied in" to several platforms including

and C#:

and F#:

and it is precisely because we are not tied to any given platform as a
company that I do not consider it to be a danger. On the contrary, in fact,
our Mathematica product often sells well when our Microsoft products are
waning so there is stability in it for us.

Granted we cannot easily port our 250kLOC of visualization software from F#
and Windows Presentation Foundation to another platform but we found it
practically impossible to port C++ I do not feel we have lost anything.

>> So I would not hesitate to recommend both Mathematica and F# to working
>> scientists. Not least because they compliment each other extremely well.
> for the moment, maybe ....

I cannot really see that changing. F# will not get a standard library
anything like as comprehensive as Mathematica's and Mathematica will not
get a modern static type system and concurrent run-time. The combination of
all of those features would be a wonder to behold though. :-)

Dr Jon D Harrop, Flying Frog Consultancy Ltd.

  • Prev by Date: Re: How can I do a "grep -v" equivalent in Import[]?
  • Next by Date: Re: Mapping A Condition Across Two Different Lists
  • Previous by thread: How to do efficient NDSolve Error control
  • Next by thread: How to remove duplicate solutions (Solve)?