Hi Bob, This is a nice discussion piece.

A closely related topic is that of doing computer algebra on constrained devices, a topic on which Elena Smirnova and I have run a few conferences under the banner “Compact Computer Algebra”. See CCA 2008, CCA 2009, CCA 2010, CCA 2012.

Elena did an extended period as a postdoc with us at ORCCA, before heading to Texas Instruments, so the question of computer algebra on handheld devices has been a natural one. This is not exactly the same as doing symbolic computation on your phone, but there is a lot of common ground. Other areas where you might want to do “compact computer algebra” might be smart line-breaking in a word processor, editing gestures in handwriting entry, etc.

For the telephone, have you tried Maxima ported by Yasuaki Honda? It is now in Google Play Maxima on Google Play.

It is great that this port exists, but there is obvioiusly a lot that one would do differently if doing math ab initio on a small touch screen device. I wouldn’t worry too much about a client-service architecture for computational power — we won’t be doing giant problems on phones.

The main reason I would imagine for a service oriented architecture is instead to enhance collaboration, and this I find compelling.

Stephen

That said, the big systems are getting rather old now, though they’ve been significantly and iteratively improved. I do wonder about where the next big software in the area is going to come from. I also think about how we might have done Axiom differently if we had Python, or at least used Java or its virtual machine.

If I were to create a mobile symbolic computation system I would definitely start with an existing system running on a server. Open source Axiom would be an excellent choice. Porting Axiom to a mobile platform, while possible, would be a ton of work. But running it on a Linux server would be trivial. I’d then slap a REST API around it to make it accessible from any mobile or web application.

I wouldn’t want to write a new system from scratch. Creating symbolic computation systems from scratch is hard :-)

Limited bandwidth might be a problem with large data, but I wouldn’t worry about it too much. Mobile data connections are getting faster all the time, and really the data isn’t all that big compared to, say, streaming a video. Also the REST API could batch data in pages since you couldn’t see all of a large expression at once.

For output, using LaTeX is a great option. There would be some nice opportunities for a multi-touch interface for showing large expressions. I’m sure there are great areas for innovation here.

The most interesting area of research for me would be to leverage multi-touch input for writing math expressions. It would be great to just write out expressions in a natural math notation with your finger. It’s a hard problem, but not impossible.

Mobile devices have quite capable GPUs these days, so adding some nice 3d visualization capabilities would be fun too.

Are you thinking about starting a new project?

Will have to get this app. Thanks for the points, Bob!