If they are making money somehow (sure they are), how can they use Smath or its engine without permission of Andrey? Even if they are not making money it does not matter. This is freeware, not gnu-gpl. If it was gnu, then they should mention about Andrey too.

The widespread use of SMath is a good thing, but like this? With stealing it? If Andrey have not permitted them about this (this is an option too) I think they must be sued.

I have already stated the option that Andrey could permitted (or sold) Smath engine, or part of it. But if he is not, I do not want a beatiful thing like Smath slip away from its dedicated creator (and users) to some big company.

Nope, it does not work. See the video
Maybe something is blocking me, but can not figure out what.

Regards,
Radovan

I already did. It seems that my antivirus software (avast) reported false alarm when connecting to Knovel.

Regards,
Radovan

EDIT: It seems that avast caused problem. I downloaded few "Knowel Worksheet (.sm)" files and as expected, they can be opened in SMath

Edited by user 17 March 2014 01:49:23(UTC)  | Reason: Not specified

Thank you, Gabriela!

SMath Studio core is built-in into Knovel Interactive Equations to provide ability to make calculations of all kinds in the browser window with no need to instal any third-party products.
It is a pleasure for me to work together with Knovel on such an interesting project. For me it is also a great opportunity to make math more accessible for people.

Just want to mention, it is important, SMath Studio remain to be absolutely free software, independent form any company, so nothing changed for SMath community.

Best regards, Andrey Ivashov.

Edited by user 17 March 2014 22:47:29(UTC)  | Reason: Not specified

Their sheets are really trivial, I agree. I also would support the request for more reliable numeric functions. My observation is that for the students, it is easier to use FindRoot() than Solve(), because the latter may spit out mathematically correct but useless complex and negative solutions, whereas in most of the cases it is easy to get reasonable guesses for the initial values as required in FindRoot().

Derive, for example, demonstrated, that a very lean code could equally serve symbolic and numeric needs. Thus I guess that there is hope to get both. But not unless Andrey finds ways to extend the ressources of his current one man show.

Thank you Andrey for the reply

I hope that we understood each other. I just wanted to mention once more that I feel we are in a loop for years. Let me try to put that as simple as possible.

1. It happened that when a calculation in SMath took an unreasonable long time or simply do not work, than the SMath symbolic core is blamed.
2. One of the first advice then is to use eval() or numeric optimization. This will sometimes greatly reduce the time of calculation.
3. Sometimes eval() or numerical optimization simply do not work, because of the way SMath is working - and then we are stuck here. We are forced to use eval() but we can not solve our problem if we do not use symbolical engine. Go back to 1.

There are many post from me and many other users regarding this. Many times I was told that this is because SMath is not float point numbers oriented but symbolically oriented with its own symbolic engine. At least I was understood that way. I can not comment to much on this because I am not a programmer nor an expert in all these things. Moreover, I just have the impression that if SMath engine do not change, in order to work in strictly numerical fashion, the same problems will repeat over and over.

Regarding numerical algorithms, I think there is not the point about their availability. You said that everything is prepared for their implementation. I still have the impression that there is still something missing. I do not know what is this in particular, and I think that the plugin makers oriented to numerical algorithms (uni, Davide, Martin) can say much more about this. By the way, if I understood all the things well, uni used many available libraries and made quite a lot of numerical functions. Davide made few very valuable plugins by using his own code based on the available algorithms. SMath evolved in this way significantly in the past few years. However, if we are in the SMath environment using these functions we are again in the symbolic-eval()-numerical- triangle. It happened to me few times that I just had to give up. You can browse my posts here on the Forum and you will find where I struggled quite a lot.

I might be wrong about all of this. Actually, I would like to be wrong.

Regards,
Radovan

Edited by user 23 March 2014 19:57:10(UTC)  | Reason: Not specified

Andrey, I agree, it is in your hands.

I am not really worried about plugging in whatever nice stuff might be around there, uni has demonstrated quite a lot of that. Perhaps we should focus on license-wise clean solutions for essential components of the system, such as numerics and plotting.

What really is missing is sort of white paper on how the optimization, the evaluation, the line operator, function definitions, name spaces are meant to work. From the pure user perspective we might make observations and try to do some analysis. But how can we tell bugs from features? I don't know if you just try to avoid to much detail in order to not give away the real assets of your concept or if it is just lack of time that we do not have such a document.

When I today reviewed my old forum posts I had to face that for some of them (e.g. http://smath.info/bts/Is...s/IssueDetail.aspx?id=56) I felt as stupid as two years ago. I see that this is a plain bug but in the same example there is the miracle of a for loop from 1 to 4, ending up with no change to the loop counter. How am I expected to understand from observation under which circumstances this happens? I have written 340 pages of handbook but I know nothing about the essentials. Isn't that a waste of time?

I am not a professional programmer or mathematician and there are a lot of formal concepts out there which I might never have heard of but which might be essential for understanding how SMath works. Only after having understood these and finding them consistent with the observations, only then we can think of how to get the best out of it and what rules the users should stick to.

However, in the end, it is your project, you are the pace maker and it is in your hands to change things. Everybody is free to engage based on his trust in you and the project.

The alternative would be to pick another project, perhaps open source like sympy and wrap it in a Mathcad style interface, my handbook lists some of such attempts. Again, engagement would be based on trust in the right choice and in the guys behind the project.

Edited by user 24 March 2014 01:38:16(UTC)  | Reason: Not specified

I'm not a programmer at all, but electric engineer (hardware developer).
I have been keeping an eye on Smath for about 2 years.
The major advantage of Smath over other mathematical software is an easiness of using symbolic and numeric calculations in the same worksheet.
And the performance of symbolic calculations is of the same order as in Maple or Mathematica.
For example, I got analytical expressions for coefficients of my 8-order sigma-delta modulator faster than using Mathematica !
Each coefficient is a formula of about 1-2 page long.

What is really frustrating:
1) Every time I make a small change, I must to recalculate entire worksheet, wasting a lots of time ! Please implement "selective" recalculation.
2) There is no visible distinction between symbolic and numeric operators. I would suggest something like a small square in the corner, used for flagging enable/disable option.
3) So long I can't understand how the performance depends on usage of eval() function. Sometimes it helps a lot, but sometimes it makes no difference at all.
Why the result of pure NUMERICAL matrix functions like mwabs() is not equal to eval(mwabs()) in terms of worksheet performance ?

Anyway, Smath is a great software !!
My best wishes to Andrey.

Edited by user 24 March 2014 22:03:50(UTC)  | Reason: Not specified