Desambiguation and know more.
Cubic spline "ainterp" and the likes are little useful
but more generally "useless".

Jean

Sorry all Collabs, I want to correct myself.

linterp: just didactic for straight junction
ainterp and cinterp are much differents.

Occasionally either one will do apparently same, but no so !

In the attached "Quadratic spline", cinterp does a much better job.
in the attached "Hermite cubic", just the opposite: cinterp is
totally zombie while ainterp is excellent .

What's gorgeous about either the 3 [linterp, ainterp, cinterp] is the
populating of sparse matrix for surface plot. Will post in few days.

What puzzles me is the derivative operator in Smath. It is not scalar
at all and excludes plotting either "spline interp" unlike Mathcad.
I have the basic Mathcad code of the derivative operator but not sure
yet if it can be made Smath companion. Please read carefully "Hermite"
the local definition operator is the code that highlights the granularity
of the Smath derivative operator.

The work sheet "Granularity" exemplifies what it means in all aspects
of applied numerical maths. I bet some plugins are 64 bits extended
floating while Smath is 32 bits, so that they don't digest the same food.

Cheers, Jean

Hello Radovan,

I read your general comment, word by word. Quite right in all,
but my reserve is that like in Mathcad, most users go to bed
before exhausting their work or like reading papers, too much
in diagonal. True: Allan Razdow [the creator of Mathcad associated
with the creators of TK Solver] was very good coach at time of Mathsoft.
Many Collabs of the "Mathsoft Collaboratory" were "Neerds".

On the other end the core Maple of Smath does a lot better job than
Mathcad 11, especially for inversing functions which is an essential
CAS tool for Scientific/Engineering. My Platinum RTD in Smath is
just "Bingo Quick" ,,, gorgeous compared to Mathcad 11 !

I wish Martin can help coding the remaing of the attached l_p_cspline
Smath work sheet. Up to where it is in the attached, superb for Edu.
RemToDo: discretise, and work around the "while loop ... return"

Cheers, see you soon ... Jean

Another very nice example Good . Thank you.

Just to say for myself, I also managed to transfer from Mathcad to Smath almost all I needed.

I am not sure what did you want to say here [Radovan].
_________________________________________

The Smath Maple Wrapper is more powerful than Mathcad/Maple 11 ... examples: root, solve ...

Maxima Plugin almost does not work in my Smath version June 2015.

What you mean about Smath "own symbolic" is probably few book style integrals, derivatives, exponetial conversions ... etc.

In the attached, the l_p_c splines work completely except for the granularity in the derivative,
nothing to do in there unless Andrey reads us. The bugger is the segmentation wrt the derivative code . If it works in Mathcad that's surely because the designers included an epsilon that freaks the segmentation but not invalidates the general code.
Or an otherwise related "return" code
Do you still have Mathcad active in your PC ? which version ?
My web site [200 MB] of Mathcad 11 Single User might still be active for visitors to download.
It is dead for me because on May 30 3:42 PM PTC deactivated all Single User Edition and by same token the download from the web site is dead so that I can't even open in *.PDF or print
,,, I have PTC where you know !

Google for : mathengjmg ,,,,, that will direct you to my now dead former Mathcad 11 repository

Cheers, Jean

Hello Jean,

Maple Wrapper and Maxima Plugin might do almost all the job the users need but they have quite "rough edges" at the moment. We should be aware and have respect that these plugins are maintained by the good willing people in their spare time. Therefore, when they will work depends of those people.


I think I am not the right person to answer this question. I just think that the Andrey had some intention about extending the SMath symbolical calculation capabilities (just look under Calculation main menu) but it was never done. To repeat, this is just my opinion. The complete answer Andrey could give you.


All my respect for the Mathcad files you have made. There is lot of work involved there. By converting some of them into SMath you might find where the SMath week spots are ant to have some proposals to improve it. I did that this way and nagged here on the Forum when I found problems .

As for the presented spline functions, I would like have them them packaged (lspline(), pspline(), cspline) including the mentioned and wanted dspline() and intspline(). I mentioned that few times because these things are quite basic calculations one needs in any numerical analysis.

Moreover, there are also few numerical plugins (mostly made by uni) which implements some of the functions from the numerical libraries like Matlab-C-Math-Library, GNU-Scientific-Library etc. Actually, at the moment there are many numerical functions in the plugins and the problem is to find them in SMath. SMath still misses some kind of organized functions list from various plugins.

Regards,
Radovan

Edited by user 25 September 2015 17:42:49(UTC)  | Reason: Not specified

Thank you Jean

Welcome to the "dark corners" of SMath. Here, it is probably the line() command gives you headache. It should be s:=M^-1*br and not s:=|M^-1*br. If I leave that way, the calculation will be quite long (I did not have a patience to wait that long - it might crash though). The problem might be that whenever you use 's' defined with line(), then 's' will be called like a function - every time you call it, the calculation will be repeated (imagine what is going one in a loop with 's' then with large matrix and matrix inversion).

On the other hand, it is always advisable to right click on a region and choose Optimization->Numeric or None or to use eval() if you need strictly the numerical answer. If you browse the Forum there were lots of examples when this helped.

Regards,
Radovan

Edited by user 26 September 2015 17:52:06(UTC)  | Reason: Not specified

Thank you Jean

I respect your persistence

Regards,
Radovan