The behavior of tellsimpafter as you describe it sounds peculiar and we
should fix it.
Here is an example of the current behavior:
tellsimp(integrate(f(x),x),g(x))$
integrate(f(x),x) => no change (verb is evaluated, not simplified;
returns noun)
'integrate(f(x),x) => no change (doesn't match noun)
'(integrate(f(x),x)) => g(x) <<< matches; what use is this?
It seems to me that there is no good reason to be defining simplification
rules on verbs. Yes, I can construct scenarios where it would be useful
(see below), but even if someone does that someday, I'd argue it is a very
very specialized and unusual case. The usual case is application of
tellsimp rules during general simplification of whole expressions, in which
there should be no verbs at all (remember that simplification doesn't reach
inside lambdas including function definitions).
So I would suggest that tellsimpafter(f(...),...) should be interpreted as
applying to the *noun* f. Same for any embedded functions, e.g.
tellsimpafter(f(diff(x,y))...). The only argument I can think of against
this is that if for some reason you *do* want to define a rule on a verb,
you won't be able to.
-s
-------------
Here's the unusual scenario where simplification rules on verbs would be
useful. Suppose I'm writing a transformation system for Maxima programs
(non-Maxima programs don't have such a thing as a noun/verb system...).
For example, I might want to transform the code fragment (thru 3 do x)
within a program to (x,x,x) ("loop
unrolling<http://en.wikipedia.org/wiki/Loop_unwinding>;
"). In that case, the pattern would have to match a verb, not a noun.
On Sun, Nov 25, 2012 at 4:44 PM, Robert Dodier <robert.dodier at gmail.com>wrote:
> On 2012-11-25, Stavros Macrakis <macrakis at alum.mit.edu> wrote:
>
> > 1) tellsimpafter attaches the simplification property to the *verb*
> laplace,
> > not the *noun* laplace (cf. noun and verb section of
> > manual<http://maxima.sourceforge.net/docs/manual/en/maxima_6.html#SEC31
> >),
> > but the simplification needs to be on the noun. It is unclear whether
> this
> > is a bug; it is conceivable that one would want different transformations
> > on the verb form and the noun form.
>
> Well, I think tellsimpafter's treatment of nouns and verbs can only be
> described as a bug. I'm tempted to say that a rule should always apply
> only to either the noun or verb form, in the spirit of "you got what you
> asked for". Here's a description of how it works at present (as
> determined by reading the code & trying some examples):
>
> The treatment of noun and verb forms is slightly confused. If a
> rule is defined for a noun (or verb) form and a rule for the
> corresponding verb (or noun) form already exists, the
> newly-defined rule applies to both forms (noun and verb). If a
> rule for the corresponding verb (or noun) form does not exist, the
> newly-defined rule applies only to the noun (or verb) form.
>
> Whatever we decide about tellsimpafter, we should ensure that tellsimp,
> defrule, and defmatch act the same.
>
> > 2) when the laplace function returns a result, it marks it as simplified
> > without giving the simplifier a chance to apply tellsimpafter rules.
> This
> > is a bug, which should be corrected.
>
> laplace constructs a lot of expressions with the SIMP flag, which should
> just be struck out, right? laplace shouldn't mark expressions with the
> SIMP flag.
>
> best,
>
> Robert Dodier
>
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>