I have recently encoutered 2 errors with Mathcad's symbolic solver

Author

1-Visitor

Very interesting Roger. I wonder why does changing the constant terms works?

R

Raiko

17-Peridot

I second that. Conversion is a crucial feature that has to work and to work seamlessly.

It's beyond me why PTC appears to write code without bothering to listen to users. In their advertisement they seem to ignore that the cases they're making for MathCad Prime are all cases for MC15 but not Prime!

Raiko

Author

1-Visitor

I guess there is nothing we can do Raiko other than continuing to point out how PTC's Mathcad development is impacting users.

25-Diamond I

david snelling wrote:
I guess there is nothing we can do Raiko other than continuing to point out how PTC's Mathcad development is impacting users.

Sad but true and there is only a very, very dim chance that this will change anything, I fear.

BTW, I played again with your integral and tried to simplify it to track down the problem. While I wqas able to reproduce the effect with a more simple integrand even without that additive constant, I encountered some strange effects during my attempts. Sometimes changing simple things like a variable name would make the i disappear but on relading the sheet it was there again. I also found two workarounds for my simplified example which both did not work on your original integrand.

One way which works for your integral, too, is to separately calculate numerator and denominator and then put both together. Thats somewhat more elaborate than the way with using Re() showed in my earlier post so I would prefer the latter.

1_Dave3c-xmcdz.zip

25-Diamond I

I agree with your feelings about Prime not being an improvement so far and that jumping off the MC15 bus and converting old valuable worksheets (many of them won't work anymore) is not an option yet.

Concerning symbolics - Mathcad always was known as an all purpose program with a nice user interface (I am speaking of MC15 and below, not Prime), but there always where programs with better symbolics or programs with better numerics available (but then you would have to go without units or without the wordprocessor like layout, etc.). So symbolics ever was an addon not created by the makers of Mathcad. Older versions used a subset of Maple to do the symbolic evaluations and when Mathsoft changed that and included a subset of muPad instead (one of the few things you cannot blame PTC for) it got worse in many respects, better in some others. Looking at Prime it doesn't seem realistic to me that PTC would be able and willing to spend the ressources necessary to develop its own symbolic engine as you suggested.

Concerning your problem: No! Mathcad does not give you an imaginary solution as you wrote in your file. Just because an expression contains the imaginary unit does not mean its imaginary itself. Mathcad simply had decided that its simplified enough (I also tried "simplify,max" without success). So either Mathcad "thinks" that eliminating the imaginary unit would make the expression less simply or its not capable enough to do so (as seems to be the case here, because the real representation of the solution is shorter/simpler). In contrary to Maple muPad defaults to deomain complex and does not consider an imaginary unit as something bad or unusual

You can verify that its a real solution by taking the imaginary part of your calculation which Mathcads simplifies correctly to zero.

Taking the real part seems to give you the solution in the form you expected.

1_Dave3b-xmcdz.zip

Author

1-Visitor

Werner, I will try your interesting solution with future symbolic problems. I had realized that the term was not necessarily imaginary because it had imaginary components in it. I had even tried selecting the complex result and calculating the imaginary component of it, but got no result.

I tried the solution you had for the integral problem on the non integral problem, but got a “No solution found” for the imaginary part and a “The symbolic solution is too large to display....” for the real part.

You evidently know more about the symbolic solvers than I do, but isn’t Maple a more robust and capable system than MuPad? I never had these problems with the symbolic solver when Mathcad used Maple, although to be fair I may be trying to solve more complex equations with MuPad.

25-Diamond I

muPad sure is not that bad but most experienced long term users consider Maple to be the better of the two. But thats history and it does not look we will get any better in the near (and also not so near) future.

Author

1-Visitor

I think one outcome of this posting is that using Mathcad's symbolic solver can be like spinning a roulette wheel.