Newsgroups: comp.lang.apl
Path: watmath!watserv1!utgpu!news-server.csri.toronto.edu!rpi!usenet.coe.montana.edu!decwrl!csus.edu!sfsuvax1.sfsu.edu!vpcsc4
From: vpcsc4@sfsuvax1.sfsu.edu (Emmett McLean)
Subject: Re: %. Torture test ;; Looping ;; General easy ?s
Message-ID: <1992May25.002050.14133@csus.edu>
Sender: news@csus.edu
Organization: San Francisco State University
References: <1992May23.213529.28082@csus.edu>
Distribution: NA
Date: Mon, 25 May 1992 00:20:50 GMT


I see that the variable names, my_inverse and js_inverse,
would have been clear than my_hilbert and Js_hilbert. Also,
my row_norm instruction had a ( 8 + 1) where just 8 was
what I needed. Ugh. I suppose it was clear what I wanted.

Now my torture test program works like a charm. Thanks
very much!

I am curious that for values 5 and higher that

row_norm =. ' >./ +/"1 ((my_hilbert y.) +/ .* (hilbert y.)) - (y.,y.) $ 1, y. # 0 ' : 11
 
provides larger row norms than

row_norm =. ' >./ +/"1 ((hilbert y.) +/ .* (my_hilbert y.)) - (y.,y.) $ 1, y. # 0 ' : 11

Since I get the same result when running a Hilbert torture program 
written in modula, I don't think the explanation has anything to do 
with the implementation of J. Any ideas?

