@Nasser If you remember the history of this site, what happened was that the bulk of these committers were really pissed about having the previous iteration of the proposal closed. Anger can be a good motivator in some circumstances.
@J.M. I did not know that! (or may be I knew but forgot). I thought area51.stackexchange.com/proposals/37304?phase=commitment was the first and only attempt to make Mathematica web site. This is interesting. I do know that Maple had once such proposal and that was closed after sometime. I was hoping this one will not have same fate. It will be nice to have a Maple own site here. Competition is good :)
@Nasser Oh, actually I used to try to learn Ansys about 5 years ago, but gave up after some miserable struggling, it's just... hard to learn in my impression. I agree the simple syntax of "FiniteElement" is very attractive, and its growth is expected.
@xzczd WHen I used ANSYS, I did not like the GUI much, so I used its scripting language called APDL. Which I find much easier to use than the GUI actually.
APDL is huge language on its own and has 1000's of commands! I only learned few I needed.
@xzczd ANSYS is now free for academic and students. I have it here on my PC. The full ANSYS, free of charge. It can be downloaded from ANSYS student web site. I do not use it much now since I am not taking a course that needs it now.
@J.M. I was in the middle of a 1500 mile move when the site ticked over into beta. That's part of the reason my user number is as high as it is. :)
@Nasser We also broke all the speed records up to that point, from time of proposal, to commitment phase, to private beta, to full site. Ubuntu previously held the record, and we took about 2/3's of the time for each one.
Also, anger motivation is a necessary ingredient. If this is correct, two MATLAB proposals have already been shot down, and the third one is still struggling.
@rcollyer OTOH, a little bird told me that a number of those users are actually regular habitués who just didn't want to ask embarrassing (to them) questions under their actual account. :D
Thanks for the beginer of MMA.SE to propose for Mathemtica.Just a un-perfect name,and I think the Wolfram.SE will be better in case of always mixing with Mathematics.SE up to now.
I read about the Julia language more than once and yesterday night I looked over the syntax as it seems a very nice language. However, what do I find on the front-page of their website? A speed comparison with different languages that is exactly one of those speed comparisons. A little pissed I went over to GitHub and looked at their test implementations just to find what I expected to find.
Long story short, if you have a GitHub account then you might want to give this issue I reported a thumbs up.
I believe you can do this by using the smiley on the page, but I'm not sure
@MMM It's getting to the end of the academic year (at least in the US). Projects are coming due; exams are happening. The pressure brings out bad questions, and some site users recognize that (every year). One effect of the downvotes is to communicate to other users (to me, e.g.) that there are better things to do. I find that helpful.
@MichaelE2 I never knew. Its good that some people here dont hesitate to down vote. I was just wondering that suddenly what happened that people start down voting (ofcourse not random). Your comment make things clearer for me. Thanks
@halirutan I looked at those benchmarks years ago. I think the original implementations were reasonable. They were not always the fastest, but they absolute fastest versions was sometimes so convoluted that it would not have made a fair comparison with other languages.
Later people tried to "improve" them and ended up with stuff like this nonsensical Compile
@halirutan Actually that compiled Fibonacci is faster than a non-compiled one.
@halirutan The author of the benchmark code said here that he was willing to submit a rewritten version of the Mathematica code, but no one stepped forward and suggested another version. Ever since I have accepted those benchmarks as acceptable.
@MMM I usually don't just downvote but instead just vote to close, maybe with an additional comment, but from someone asking a question about PWM i expect them to be able to RTFM
@Szabolcs @C.E. I cannot recreate the results in this comparison table. One of the simplest functions they use is pisum which is nothing more than calculating this 500 times in a loop:
Sum[1.0/(k*k), {k, 10000}]
I set up Julia on my system and compiled the equivalent C function. For the equivalent code of
In fact, there seems to be something strange with AbsoluteTiming. Let us take an implementation of pisum that is exactly like they used it in C and Julia
pisum2 = Compile[{}, Module[{sum = 0., n = 10000.0},
Do[
sum = 0.0;
Do[
sum += 1.0/(k*k)
, {k, n}], {500}];
sum], CompilationTarget -> "C", RuntimeOptions -> "Speed"
]
AbsoluteTiming[pisum2[]] gives for the first run always a value of about 0.05 s.
But if you iterate pisum2 and measure the overall time and dividing later by the number of iterations we get
n = 300;
First[AbsoluteTiming[Do[pisum2[], {n}]]]/n
Out[29]= 0.0189038
