When I reported the problem before, I used the Maplesoft Customer Service Tech Support request page.  Per Robert's suggestion, I have put it in again using the Submit Maple Software Change Request form.  Will this affect how the report is treated?  Time will tell.

By the way, you have to choose from a list of Maple versions when putting in the SCR.  Maple 12 is not on the list.

Robert,

 Yes,  I have reported this, leading to an exchange between a Maplesoft tech support person and me in which I tried to convince him it is a bug in Maple, and he tried to convince me it is just a documentation error for failing to correctly identify things you are not supposed to try to do in Maple.  Neither of us succeeded.

Bob

I have resurrected this dead thread because I stumbled across my old solution.  It turns out my memory was not that great when I said I recognized my solution in the expression Robert posted.  I wrote my expression for a*b as 1/c + 1/c - 1 - a - b, where c = 1/(a+1) + 1/(b+1) - 1/(1 + 1/(a+b) + 1/(1/a + 1/b)).  Should it be a surprise how many odd-looking solutions there are?

Robert, in my case the most important question at the outset was whether a solution is even possible, and your solution answers that, even if it's not "very nice".

I didn't get the part about 3/2*b being equal to b+1/(b+b), but I could see that 3/2*b=1/(1/(b+b+b)+1/(b+b+b)), so I didn't worry about it.

Doing that would provide a real solution, so I have to change my opinion.  You CAN use Maple to solve this problem, if you've been using Maple longer than I have.

Joe, please tell me you enjoyed doing this, so I won't feel guilty about instigating it.

If you change 500 to 50, what are the odds of x^2 showing up?  (It took me 9 tries.)

Thanks, Robert.  When I saw your expression a light went on and I recognized my old solution.  I haven't found my notes on it though.  They're around here somewhere.

I solved it by fiddling around with various expressions for a few weeks until I stumbled onto it.  Joe's way seems a lot more elegant.

As Joe said, Maple is probably no use in finding the expression.  But it sure is good at verifying it.

Joe, from your remarks it sounds like this may be a known problem that has been written up or discussed somewhere.  Care to share?  Or, like me, did you think of it and work it out on your own?

I'm not positive, but I think my questions have been answered.  In reading the replies, I did get a couple of surprises:

Why would a widely-used command like 'solve' have undocumented options?

Since the Explicit option was undocumented, how did so many of you already know about it?

Bob Baker

Thanks, John, this method works too and seems more intuitive.  But in ?solve/details I can't find any reference to the solve command having more than two arguments.  Where should I be looking for this?  Where would I find the choices for the third argument?

Bob Baker

Thanks, Georgios, allvalues is the command I was looking for.

Bob Baker

Acer & Joe,

Thanks much for checking it out.  Now all I have to do is convince Maplesoft to sell me the student version.

Bob