Newsgroups: comp.lang.apl
Path: watmath!watserv1!torn!utcsri!rpi!news.ans.net!europa.asd.contel.com!uunet!uunet.ca!geac!itcyyz!yrloc!rbe
From: rbe@yrloc.ipsa.reuter.COM (Robert Bernecky)
Subject: Re: Dead horse matrix reduction
Message-ID: <1992Jul30.050736.19895@yrloc.ipsa.reuter.COM>
Keywords: None
Reply-To: rbe@yrloc.ipsa.reuter.COM (Robert Bernecky)
Organization: Reuters Information Services (Canada) 
References: <920722044324_70530.1226_CHC117-1@CompuServe.COM> <1110@kepler1.rentec.com> <3j0mbfps@csv.warwick.ac.uk> <1121@kepler1.rentec.com>
Distribution: Snake Island Research Inc   
Date: Thu, 30 Jul 92 05:07:36 GMT
Lines: 44


In article <1121@kepler1.rentec.com> andrew@rentec.com (Andrew Mullhaupt) writes:
>In article <3j0mbfps@csv.warwick.ac.uk> strgh@warwick.ac.uk (J E H Shaw) writes:
>>Unfortunately, the literature (even the `statistical' literature) isn't
>>the place to look for an honest answer; a while ago I submitted a paper
>>on efficient parallel generation (in APL) of quasirandom sequences
>>(Sobol's LP\tau sequence etc.) to the SIAM J. of Scientific & Statistical
>>Computing.  I think an APL or similar implementation gives valuable
>>insights into the structure of such sequences (and is pretty efficient),
>>but the paper was immediately rejected with a comment to the effect that
>>they aren't interested in APL.
>
>But this is pretty normal from my point of view - the stat literature is not
>really interested in APL, _or any other language_. If I wrote an article on
>how you can do ANOVA in a novel way using say, Prolog, I would expect the
>same treatment. In order to care about my advance, a statistician would have
>to learn Prolog first. If I can't explain this method without Prolog, I
>will expect to be able to justify it's use by a number of very significant
>advances in the theory, or I will expect rejection. 

The advantage to using a programming language is that the algorithms are
Verifiable mechanically, whereas handwaving is the only way to verify
pseudocode, etc.

I think that the objection being raised above, however, is that
articles are being rejected because they contain APL, rather than being
rejected because they contain algorithms, Pascal, etc.

This sucks, as we say in my country.


>When it is important to put algorithms in the literature, it is usually
>important to present them in an essentially mathematical form. Take for
>example the power of two FFT, where you pick bits. Look this up in the
>literature, and you will _not_ find bitwise operators from C, FORTRAN, (and

This is true. That's why the damn algorithms are essentially incomprehensible:
Math is NOT a programming language, but APL IS math.

Robert Bernecky      rbe@yrloc.ipsa.reuter.com  bernecky@itrchq.itrc.on.ca 
Snake Island Research Inc  (416) 368-6944   FAX: (416) 360-4694 
18 Fifth Street, Ward's Island
Toronto, Ontario M5J 2B9 
Canada
