> I would suggest the following rewording:
> 
> "A function which is not 1-1 is restricted to a principal domain.  The
> limits on principal values of arcsine and arccos may be obtained as
> follows..."

This is not correct either.   For example,
11 o. y , which computes the imaginary part of y,
is not 1-1 but is not restricted to a principal
domain.  (11 o. y works for any complex number y.)
What is restricted is that the inverse is an inverse 
only for a significant subdomain.  So _11 o. y is the
inverse of 11 o. y for the set of imaginary numbers,
complex numbers with real part 0.

See the dictionary page for / for a related discussion
on neutrals (identity elements).



----- Original Message -----
From: John Randall <[EMAIL PROTECTED]>
Date: Friday, June 30, 2006 5:02 am
Subject: Re: [Jbeta] _11&o. is not the inverse of 11&o.

> Henry Rich wrote:
> > The interpreter thinks
> >
> >    11&o. b. _1
> > _11&o.
> >
> > but it ain't so.  Circle functions 9-12 have no inverse that
> > I can see.  Circle functions _9-_12 do, but they're not
> > the functions 9-12&o. that the interpreter uses.
> >
> 
> The following explanation at the bottom of the Dictionary page for
> o. appears flawed:
> 
> "The principal domain of the inverse of a non-monotonic function is
> restricted. The limits on real domains of arcsine and arccos may be
> obtained as follows..."
> 
> If a function is monotonic, then it is 1-1, but it is the latter
> condition that is necessary for it having an inverse.  Monotonicity
> makes no sense if the domain is the complex numbers, as with 9&.o. .
> 
> The usual terminology is to take a function f and restrict its domain
> and codomain to D and C so that the new function f* is 1-1 and onto,
> and thus has an inverse.  D is called the principal domain of f* 
> and C
> the (set of) principal values.  The inverse of f* has domain C and
> codomain D and is what we loosely call "the inverse of f".  In the
> second sentence quoted, it is not the limits on real domains of arcsin
> and arccos that are being described, but the limits on the real
> domains of sin and cos.
> 
> I would suggest the following rewording:
> 
> "A function which is not 1-1 is restricted to a principal domain.  The
> limits on principal values of arcsine and arccos may be obtained as
> follows..."
> 
> Since conventions differ as to what principal domains are, even with
> quite standard functions, it might be worth spelling them out, as is
> done for sin and cos, the easiest ones to guess. This is even more
> worthwhile for less obvious cases like 9&o. .


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

Reply via email to