WSTP server on iPad Pro

Many have requested a fully featured version of Mathematica on the iPad. The new iPads with M1 chip definitely have the resource to run it. I understand the challenges for Wolfram to create an entirely new notebook UI for tablets. On top of that most people belonging to the intersection of iPad owners and Mathematica users have a Desktop or a laptop anyway, so it is really easy for them to just use a Remote Desktop program on their iPad to use Mathematica, so the need may not seem so urgent. On the other hand, using a mighty M1 chip to run a RDP or VNC client is a bit of a waste : what if we considered the implementation on the server side of things ?

Wolfram could develop a very simple app - from a UI standpoint - that launches a WTSP server (with the appropriate license of course) on the iPad, which could be used as a remote kernel by another machine.

I see two use cases here :

1. The iPad processor could help use more kernel for parallel computation. A typical user would have for instance a desktop and an iPad (which practically as powerful as a desktop nowadays) and could thus halve the processing time of parallel friendly computations.
2. The iPad WSTP Kernel could be the main kernel of a notebook instance running in a machine with a processor that is comparatively slower than the iPad. In particular, I think about the Raspberry pi here. As mentioned by the Raspberry foundation , a Raspberry Pi can be connected to the Ipad with a simple USB-C cable carrying power and high speed (gigabit) / Low latency (sub millisecond) IP connection. It is even possible to run a WebDAV server on the iPad to give the Pi access to the faster SSD storage of the iPad (much faster than Pi memory card) ! The Linux notebook side on the Pi would then connect to the high performance kernel on the iPad : familiar desktop UI and high performance, it is a win-win ! An additional advantage is that such configuration doesn’t need to be connected to the internet to work, so is functionally equivalent to a laptop. It is actually ergonomically better for intensive computations, because on laptops, the chip is located below the keyboard and make it hot, whereas on the iPad, it is the screen which becomes hot…

very curious how the latest developments of an M1 processor and 16MB in the new iPad Pro will enable mobile use of Mathematica. All software development on the iPad has been hindered by Apple policies, but now that computational resources are no issue anymore, I hope to see some changes. Would love to hear from the Wolfram Team how they see the future.

I think that the main issue is the front-end on the iPad, judging by the versos demos I have seen and the implementation of the reader on the iPad. There is no longer any technical issue I am aware of that would prevent a full implementation of the kernel on the iPad. There is definitely sufficient RAM, which was an issue with early versions of the iPad. There may be sandboxing issues, or commercial issues (providing it in the App store, for example).

The virtual keyboard is nice, but it takes app too much of the screen. Requiring a hardware keyboard would eliminate that problem, but would add expense. The other issue is with the touch-interface, which tends to be orthogonal to the widgets created for mouse manipulation for desktop systems.

I think that the solution would be to completely rethink the interface to be optimal for the iPad, and then worry about 'translating' this into the resulting document when (if?) it is opened on the desktop.

The whole project would be worth doing, if only to justify the expense of using Apple-specific technology, such as Metal and the neural frameworks available with the Apple Silicon chips in the iOS devices and not the desktops. I, for one, would like to see an interface where I could scribble notation or graphs using my finger (or Apple pencil for more refinement) and seeing it transformed into proper mathematical notation and plots. It would be a further step in shortening the line from thought to execution.

I would very much like to have a working version of just the documentation. I enjoy exploring the documentation. It is a form of recreational computational thinking--and sitting at my laptop is what I do for work.

It wouldn't be necessary to have saveable files. Evaluations could have default TimeConstrained and MemoryConstrained.

I believe that my iPad has more than enough storage for the documentation.

I'd pay $100-$200 for a documentation only app.

Thanks, George, for the tip on sidecar. I've been using Wolfram Player to look at the documentation. But, I believe the documentation is not being stored locally. In any case, navigating the documentation is very slow and the openers are not responsive.

If the documentation was stored on the ipad, it would be better for "recreational reading". Interactivity would be nice, but necessary (for my purposes at least).

I agree that the pace of adopting new Apple technologies has slowed -- it too far too long, in my opinion, to make the front end Cocoa and 64 bit, and to dump OpenGL in favor of Metal for graphics. The days are long gone where Apple would introduce new technologies only to abandon them, so no company need fear that efforts to adopt the APIs would be wasted.

Also agree that free isn't necessary, or even desired. Student/hobbyist pricing would be fine. (I have had a commercial license since 1989.)

I also understand that Wolfram is a small company, and that they can't do everything. While I like the new stuff in Wolfram Language, I would like it better if Mathematica could make full use of the hardware I run it on. That includes both my MacBook Pro and my iPad -- if given the chance to run a native Mathematica on iPadOS.

I believe that Wolfram Player for iOS has essentially a complete implementation of Mathematica running on iOS. At various WTCs over the years, I have seen Mathematica working on iPads. The main issue is tweaking the UI to handle touch. This issue would be gone with ARM macs.

Wolfram has already implemented Metal for graphics (replacing OpenGL), and I think that it would be possible for a small team to make use of all of Metal for neural networks, and even parallel processing using GPUs. I looked into doing this, and while it is beyond what I can do by myself, it is certainly do-able.

I only use the cloud app occasionally, due to the lag and the crippling of interactive graphics, necessitated by the round trip to the cloud, of course. Not everyone has broadband all the time, so this is not a viable choice for 'serious' work.

Wolfram Research figured prominently in the WWDC that introduced the transition to Intel. Theo Gray and Rob Raguet-Schofield took part. It would be great if Wolfram were able to take part in this (rumored, but quite likely) transition to ARM. Remember that Wolfram was alone in the major CAS platforms who stuck by the Mac in its darkest hour.

My feeling is that if Apple makes the transition from Intel to ARM, Wolfram would have an incentive to deliver Mathematica to iOS, or iPadOS. We already know that it is technically possible, so the only issue would be with licensing and App store policies. Since Wolfram essentially gives away Mathematica for Raspberry Pi, I think that it would make sense to license Mathematica for iPad for non-commercial use. Professional users will still want the extra resources available with a laptop or desktop -- at least I would.

Some might say that Apple is still a minor player in the PC business. However, if you add in the iPad to the MacBooks, iMacs. etc., Apple is now a major player. With Catalyst, developing for macOS and iPadOS is now much easier in the past. It seems to me that Mathematica for iPad makes a lot of sense.