Just thought I'd mention how I got here, probably not the commonest
route, but then again, whose is?

I'd heard of APL (as an 'array' language) and always meant to look into
it at sometime, but that never happened what with one thing and the
other.

Recently, however, I was looking at Scmutils, a package for MIT-Scheme
that has been created for applying Scheme in a quite generic way to
mathematical concepts - none of the ideas there would seem strange to
the J community.

The really interesting bit is the book that 'accompanies' Scmutils,
Sussman & Wisdom's "Structure and Interpretation of Classical
Mechanics" (a play on Sussman's earlier classic "Structure and
Interpretaion of Computer Programs":  It is the authors' contention that
the imprecise notation of teaching physics is a serious hinderance to
in-depth understanding (as a physics drop-out I would agree).
A great little intro specifying Lagrange's equations with their system
is available as "The role of Progamming in the Formulation of Ideas".
Their solution for the "impressionistic mathematical notation" commonly
used, is an executable notation that can be tested on a computer, the
ultimate controller for ambiguity or fuzzy specification.  

To this end they have designed Scmutils, which basically has three
features they mention.  I list them here, since J also seems to support
them:
1) A functional notation that distinguishes the function from the its
value when applied to particular arguments.  CHECK
2) A generic arithmetic in which the basic operations are extended to
apply to wide variety of mathematical types PRETTY MUCH CHECK
3) Introduction of a "tuple arithmetic" as an alternative way of of
manipulating aggregate quantities.  This is basically a rather neat
extension of tensor arithmetic.

The last of the three is of particular interest to me as I have just
started a course on tensors.  Soon as my J is strong enough I will
attempt to port the scmutils "tuple" system into J.

If this works, it could be that the whole of SICP, could be doable in J,
which would be very interesting.  MIT-Scheme and scmutils only work on
Intel, and it's likely to remain that way.

Well, that's how I got to J!
By looking for an alternate / parallel solution to Scmutils.
I really want a functional (and preferably portable) environment for
experimenting with maths.  J seems to be the first candidate that ticks
all the boxes.

Anyway, if anyone else is using J in the teaching of Physics or Maths I
would very much like to hear about.  It seems to me that a lot can be
done in this area.




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

Reply via email to