Maple is an interpreted language with a fair bit of overhead.  With today's machines, it would take a rather large XML file (not hard to do, I know) before it made a noticeable difference.  Furthermore you would have to be extremely careful not to hang on to any of that data (to let the gc work), otherwise you might be able to process some of the file but would eventually choke anyways.

In other words, I doubt this is high on anyone's priority list.  That and XML processing seems to have disappeared off the radar.

Digits is an environment variable, so any assignment made to Digits at any level outside the top-level will not persist.  ModuleLoad is a procedure just like any other, so assignments to Digits will not have any effect outside that scope.

There are some very evil tricks that one can pull to (unreliably!) do this, but they are not worth it.

Digits is an environment variable, so any assignment made to Digits at any level outside the top-level will not persist.  ModuleLoad is a procedure just like any other, so assignments to Digits will not have any effect outside that scope.

There are some very evil tricks that one can pull to (unreliably!) do this, but they are not worth it.

If you had a previous definition of A (or in some strange cases N) in your document, it might have produced that.

If you had a previous definition of A (or in some strange cases N) in your document, it might have produced that.

There many generalizations, starting with Hermite-Pade, Matrix Pade, and so on.  One of the experts in this area is actually a Maple person, namely George Labahn.  George has supervised many students who have made significant contributions to Maple.

There are multivariate versions of Pade, but I must admit that I really don't know about them, I mostly stuck to 1-variable stuff.

There many generalizations, starting with Hermite-Pade, Matrix Pade, and so on.  One of the experts in this area is actually a Maple person, namely George Labahn.  George has supervised many students who have made significant contributions to Maple.

There are multivariate versions of Pade, but I must admit that I really don't know about them, I mostly stuck to 1-variable stuff.

The reason is that Maple is really all about computation, or at least was designed that way.  diff is from the very early days, long before 'assume' and 'is'. 

Basically every function that appears in practical problems is C².  Only in an educational setting or extreme research problems don't satisfy this.

Having said that, it was long thought that many other "simplifications" that Maple did were crucial, like the old nugget of sqrt(x^2) = x.  But it turns out this wasn't as needed as once thought.  Same for many other assumptions from the naive early days. So who knows, maybe this one too is unecessary.

The maple tag does seem broken.  It is nice for visual effect (when it works).  But for reuse, plain ASCII is best.  I really meant for you to put in some Maple input lines in 1D.

The maple tag does seem broken.  It is nice for visual effect (when it works).  But for reuse, plain ASCII is best.  I really meant for you to put in some Maple input lines in 1D.

(Chebyshev) Pade approximants give, in some ways, the "best" numerical approximants over a range.  See numapprox[chebpade].

The heart of gfun[guessgf] is Pade approximants!  If you want to guess that a sequence is rational, you set up a generating function, compute its Pade approximant, and if you have enough terms, bingo, you find it. 

(Chebyshev) Pade approximants give, in some ways, the "best" numerical approximants over a range.  See numapprox[chebpade].

The heart of gfun[guessgf] is Pade approximants!  If you want to guess that a sequence is rational, you set up a generating function, compute its Pade approximant, and if you have enough terms, bingo, you find it. 

The GXL and dot (see GraphViz) seem even more pervasive to me.

Maple 11 can do some amount of parallelism -- but independing testing showed that this was not quite ready for prime time as no speed gains to speak of were detected.  Low-level numerical routines can apparently use multiple processors, and real gains might be possible, but I have never seen benchmarks that show this.

There are various versions of distributed and grid Maple around, but that is probably too coarse for what you want.

It this move also the reason the <maple> tag, which used to work fine before the site upgrade, is now broken?