@halirutan I agree completely with the sentiment that performance in the system is too unpredictable. On the other hand, I do think the Mathematica ecosystem is hard to beat. Jupyter is nice and all, but in my experience it's not quite all-encompassing in the same way. I think it's got one major flaw, though, which I think people have noted: WRI is trying to do too much and is spitting out a bunch of surface-level products...

Per Wikipedia there are ~700 employees, but most of those clearly aren't going to be core developers. There are probably only a few hundred of those. And if you look at some of the attribution in the paclets, it's, say, one or two people working on a full paclet. And lots of names appear on many of them:

On the other hand, this strain could be alleviated if there were a better way for non-WRI employees to get their work into the ecosystem via some form of functional, simple package manager.

That would also, I think, increase the incentives for people to build packages well, rather than just to what they need at the moment. Most of what I build is junk, because I never expect anyone else to use it. If I thought that were the case it'd be much better developed.

But Wolfram can't really control that flow, and, moreover, it's not really eye-catching in any way. Imagine the press release: "Mathematica 22.10.3.2: Now with a package manager!"

@b3m2a1 Jupyter has one major difference: It is open source. If I'm passionate about something, especially if I use it for work, I'm always willing to contribute. So you might ask why not switching to Python. The reason is simple: Mathematica has very good mathematicians for ages. This is why things like integration, simplification, numerical solving of equations or diff.-equations works so well.

This is something that is hard to reproduce in an open-source project.

I believe when it comes down to graphics, user-interface or in general software development, it wouldn't be impossible to create something that can compete with Mathematica.

What is indeed hard is to get the mathematical foundation behind it working. Unfortunately, my work strongly relies on being able to analyse things analytically.

@b3m2a1 Yes, headlines like "bullet-proof packaging system, framework for building integrated documentation, ..." only catches the users you already have for a very long time.

@b3m2a1 And before I forget it: There is no way there will ever be a reasonable solution for Jupyter to support a Mathematica kernel. We shouldn't forget, that everything about Dynamic and Manipulate is not part of the kernel. It's part of the front end.

@halirutan I agree with all of that, although I think the for an open source project the commitment to building something like Mathematica in terms of its integration across its disparate parts would be a challenge. Another thing on the front-end, since someone mentioned the cloud, I can see how WRI thinks the cloud might bring in new users, but it's just so crushingly slow I don't see how this would work.

@b3m2a1 Yes, there probably wouldn't be support for graphs, neural networks, image processing, etc. but for those things, there are libraries out there. Indeed, Mathematica itself calls ITK for some image filters. So to support one of these topics it only needs one extremely dedicated member and probably some people that join. Working out complicated integrals on the other hand or getting expression transformations correctly, you need good mathematicians.

I tried some of the cloud stuff. Using Mathematica online is absolutely no choice for me. Regarding the other things like publishing snippets of code, so other users can run them online. I just don't need this.

@halirutan I feel that point about needing good maths people, although I think that issue is not insurmountable using, say, sage or something.

My main use-case for the cloud is as a web-hosting platform that's natively built into the language. It seems to handle anything that's HTML, CSS, and JS just fine. I only use their APIFunction and cloud notebook interface if I absolutely have to, though.

Tread carefully, y'all: mathematica.stackexchange.com/questions/153883

Anyone else having issues with Wolfram Cloud at the moment?

This morning I was using an online object and it was working, but later in the day my friend was getting HTTP 500 errors when trying to view all of his cloud deployments

Now I'm trying to use the same object I was using this morning and it's not responding

the console is also saying the server responded with 500 Internal Server Error

Looks like it's somewhat working now

@numbermaniac if the cloud goes down contact Wolfram's support. They've thrown lots of resources at it recently so hopefully they'll bother to at least keep it up and running. For what it's worth, my static pages are being served fine. Maybe there was a mass kernel crash. And one of my form functions is fine too so they must have gotten things back up.

@halirutan In the short term yes, but in the long term the proliferation of easy-to-obtain quality packages would build up the user base. See R for example.

@ChrisK OTOH, reading the source of a number of user-contributed R packages tell me that they work as advertised, but are a PITA to maintain. :D

hi @xzczd

i want to plot Re@w[x,y], where x is in the starting and ending points of the interval

Any thoughts?

How can I center the rows of an array plot when each row has a different number of elements?

@user76284 you can use the function CenterArray on each row, i.e. CenterArray[#, maxRowLength] & /@ matrix – the function is new as of Mathematica 11.

@JasonB nothing, for some reason when i tried it i wrote _Subscript and i didn't realize it . thank you for your answer

@Alucard I might still make the replacement rule more specific, Subscript[val_, k] :> Subscript[ val, p]

oh cool

i was trying to generalize my answer to mathematica.stackexchange.com/questions/153918/… to the case where there is the a0 coefficient) without using the second part of kglr's code, which is too advanced for me . That's why i asked the previous silly question.