Prior discussions have focused on MATLAB and J as languages.  In this,
there is no comparison: J is much more consistent, efficient and
elegant.  However, this is irrelevant to most MATLAB users, who want
to solve particular problems in scientific computing.  From their
point of view, MATLAB is more an application than a language.

At my university, the four main areas in which MATLAB is used are

solving differential equations
digital signal processing
image processing
finance

and there are "toolkits" which allow a novice to immediately solve
practical problems of interest in these domains.  Most actual
programming in connection with MATLAB is done in other languages
(notably C and Fortran), with MATLAB routines being called from the
language or vice versa.

Performance is important to heavy MATLAB users.  Since actual language
constructs will use insignificant amounts of time in comparison to the
libraries, a lot of effort has been expended on making the latter work
efficiently. Good performance on variant hardware is achieved through
using a tunable BLAS, and there is an active secondary market.  MATLAB
also squeezes out extra precision on Intel hardware by using native
IEEE extended doubles (64-bit mantissa).

As far as I can tell, J wants to be a language, not an application,
and MATLAB wants to be an application, not a language.  If this is not
kept in mind, comparisons will be futile.

Best wishes,

John




----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to