While still in J ...

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 12

Nick Cox in a letter (Vector Vol.13, No.2, p.10) compares a Fortran-like J program to the functionally equivalent one-liner +/"1(* x.&^@i.@-@#)\y. and asks whether a quicker and more concise J script can be produced. As far as the second requirement is concerned, Nick should observe that he is in fact evaluating a polynomial in his variable k (x. in the above), and a functionally equivalent verb to the above is >@(#.&.><\). This is at least as fast in execution as either of Nick’s programs. Space depends of course on the length of the series y. — with 1000 terms, the polynomial version requires one-eighth of that needed by Nick’s one-liner (2Mb vs.16MB), although naturally the step by step version is superior to either (30KB).

As a tail-piece, timing experiments such as the above can be interrupted by Ctl-Shift-C. This piece of information, which I have not been able to find in any of the J manuals or Help material, was passed to me by Bill McLean.