I agree. Plots in Standard are much better. When I do a screenshot, or export a plot for this site,  I usually use Standard. It is sad though that the same improvements were not made in Classic, because other than plots (and help sometimes), Standard seems to be of little use, and copying from it to Classic (or to anything else) works poorly.

Alec

I have a more simple example,

f:=(a-2)/(a-8^(1/3));
                                  a - 2
                           f := ----------
                                     (1/3)
                                a - 8

eval(f,a=2);

                                  0

It is clearly wrong, and even the limit in this case is 1, which Maple gives correctly,

limit(f,a=2);
                                  1

A way to avoid that would be to evaluate fractional powers of integers - at least in the way similar to sqrt (which is also, by the way, far from ideal, but at least such examples would require much larger numbers.)

Another way may be in doing evalf and comparing results. In this example,

eval(evalf(f),a=2);
                           Float(undefined)

as well as in the original example

eval(evalf(q),a=1/3);
                           Float(undefined)

which would also show the error.

For exact integer calculations the correctness should be the main concern - not speed. Anyway, if I did that, I would write it in C (or directly in assembler) and put such things, together with sqrt in the kernel (as Mathematica does, by the way.)

By the way, this example gives another proof that 0=1 in Maple,

is(f=radnormal(f));
                                 true
eval(f=radnormal(f),a=2);
                                0 = 1

Alec

This is an old one and a good one advice. I remember posting about such things myself here and there. That seemed to be a popular advice for 3d plots in Maple V releases 4 and 5. Is it possible that the same old (not gdi) OpenGL driver is still used as 'default' in Classic?

Alec

However,

F:=a->(a-2)/(a-8^(1/3));
                                     a - 2
                         F := a -> ----------
                                        (1/3)
                                   a - 8

F(2);

                                  0

as well as

F(2.);
                                  0.

Alec

Looking at their edits, and the comments on the discussion page, it seems to be pretty clear, whether they were paid to do that specifically, or not.

At least, some of their edits were clearly controversial - otherwise they wouldn't cause long discussions on the discussion page.

Alec

Editing by Maplesoft employes of any Wikipedia pages related to Maple clearly contradicts to Wikipedia Conflict of interest policy.

Alec

I hope I won't have to block the whole Canada :)

Fortunately, with MoinMoin that we are using, the damage can be done only temporarily - everything is saved and logged, and can be easily restored.

Alec

Somebody from mhx.maplesoft.com with that IP address, 199.71.183.2 did that. His (or her) usual edits are removing material, and he (or she) already got a few warnings there, creating for Maplesoft a reputation of vandals.

I think of blocking that IP address in Maple Wiki (but that might make it to be completely unaccessable from Maplesoft network.) Edit: just did that.

From other point of view, it would be better to have some more useful (and nice looking) plot there.

Alec

That's what I thought. NAG library has a good documentation itself. Nevertheless, that, probably, may be useful for somebody using both it and Maple.

What I wanted to have was that Maple (a standard version) had only a part of that toolbox including only the functions available from the part of NAG libraries included in the Maple distribution. Originally (in Maple 8, I think) it wasn't a problem to get the list of NAG functions available from Maple - using such things as dumpbin, or pexports, but later the functions were renamed, and it was not clear what was included and what was not, and under which names. I didn't check that recently though - so who knows, maybe the function names were restored in the dlls.

Alec 

That's what I thought. NAG library has a good documentation itself. Nevertheless, that, probably, may be useful for somebody using both it and Maple.

What I wanted to have was that Maple (a standard version) had only a part of that toolbox including only the functions available from the part of NAG libraries included in the Maple distribution. Originally (in Maple 8, I think) it wasn't a problem to get the list of NAG functions available from Maple - using such things as dumpbin, or pexports, but later the functions were renamed, and it was not clear what was included and what was not, and under which names. I didn't check that recently though - so who knows, maybe the function names were restored in the dlls.

Alec 

Just added Maple in few places there.

Alec

Just added Maple in few places there.

Alec

Well, it's only measure 0. And solutions may also not exist in such cases (as if A=0, for example. Any X would be a solution for A=B=0 though.) By the way, will dtrsyl work for non-invertible A?

I should correct my post above though. If the symmetric X is unique, and A^(-1).B is symmetric, then, obviously, X=A^(-1).B/2. In general, A^(-1).B may be not symmetric, even if A is symmetric.

Anyway, for large matrices with hardware datatype, NAG library procedure should be used, as it was correctly stated by acer.

My old wish related to that was that all NAG functions available were described in the help pages, and NAG dlls were accompanied with their libs and headers (it is not actually a big deal - I could easily create them myself, but still - that would be convenient and it is not clear why that was not done.)

Alec

Well, it's only measure 0. And solutions may also not exist in such cases (as if A=0, for example. Any X would be a solution for A=B=0 though.) By the way, will dtrsyl work for non-invertible A?

I should correct my post above though. If the symmetric X is unique, and A^(-1).B is symmetric, then, obviously, X=A^(-1).B/2. In general, A^(-1).B may be not symmetric, even if A is symmetric.

Anyway, for large matrices with hardware datatype, NAG library procedure should be used, as it was correctly stated by acer.

My old wish related to that was that all NAG functions available were described in the help pages, and NAG dlls were accompanied with their libs and headers (it is not actually a big deal - I could easily create them myself, but still - that would be convenient and it is not clear why that was not done.)

Alec

Thomas,

Thank you! I didn't know about that.

Is complete NAG library included in the toolbox, or it it just a connector from Maple to NAG (similar to Matlab toolbox), and NAG libraries (complete) have to be purchased separately?

Alec