Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
1 Like
I don’t often watch the ULS, but I did notice The Target Audience for the 16GB Raspberry Pi 5 excerpt that the TWiT editors and put on YouTube. Thanks, editors!
@jp_bennett’s comments about this 16GB Pi for education are right on target. A great candidate for this 16GB Pi config Is the Wolfram Language – Mathematica. It’s especially great for home schooling and supplementary public school education in general. Students turning in their homework in a Wolfram Notebook would be the cat’s meow.
Mathematica is a fantastic platform for visualizing/manipulating all sorts of STEM concepts, and for authoring your own visualizations. Wolfram provides a massive amount of free educational materials for download: lessons, video classes, e-books, etc. They have a bunch of hybrid Wolfram/RPi STEM projects and a boatload of courses designed to run in their software. The MAKE:Magazine archive has their own set of hardware and software projects you can use there, including Python. I think that students are better prepared (for anything!) with the Wolfram Language, but that’s my personal preference. Students that go on to University will typically find a University Mathematica site license available for use on their Macs, PCs, and Linux computers.
Stephen Wolfram announced Mathematica on the Pi back in November 2013: Putting the Wolfram Language on Every Raspberry Pi. That project made a big splash back then, but has pretty much flown under the radar in recent times. Ironically, the ~100x increase in computational capacity of a 16GB RPi5 has made the 64-bit free Mathematica far more “real” than the glacial ports on earlier hardware. The Pi Foundation’s utilities to download, configure, and image the OS on SSD means you never have to bother with hooking up a kbd/mouse/monitor to fire up a networked Pi. A pre-configured Pi is a nice gift you can give to your less tech-oriented friends. If you’re feeling extremely generous, you could outfit the Pi5 with a 512GB NVMe SSD in an Argon40 case for a monster-sized network-ready machine. That is a gargantuan tool for learning.
With the ~100x performance improvements in the last 10 years, I’ve been deathly afraid that Stephen Wolfram would back down on his company’s free RPi software. They have not. Wolfram Research has a fundamental commitment to education of our children. Conrad Wolfram’s book and TED talk oh the “Maths Fix” suggest we should shift our focus from mindless hand-calculations to computational problem-solving and visualization: let the damn computer do the crunching. Hear, hear! There are other computation engines available at this point, but Mathematica with its visualizations and touch-capable 3D pan & zoom – and manipulate – provides an incredible opportunity to provide kids to author their own Aha! joy. As an aside, Rich Cameron and Joan Horvath’s MAKE math books provide another way to achieve alternative learning with fantastic mathematical visualizations: a different way to hack your “maths fix” (MAKE podcast episode available here). Which is better? Who cares? You can go hybrid!
1 Like
I’ve been thinking of getting into ULS. I don’t subscribe to the TWIT youtube channel (I’m anal about my sub feed) but I went and watched that clip after you mentioned it.
I liked their take “why would anyone be mad about that”?
2 Likes
But how much memory does Mathematica need?
There are 4GB and 8GB model’s already, which have the same performance as the 16GB model. The only difference is the amount of RAM. You only mention performance, not memory, that is why I am asking.
If it really benefits from that extra RAM, no question, the 16GB model is a no-brainer. But a lot of people are questioning whether the extra 8GB is worth the money. I don’t understand them, it isn’t as if Raspi are going to stop selling the 4GB and 8GB versions…
There are doubtless applications out there that will benefit from the extra memory (perhaps Mathematica is one of them) and if you use them, the new 16GB Pi will be worth every penny.
For these non-power-users, they are looking at the 16GB and saying it is a waste of money. Others are looking at it and saying, “finally”. Many people can’t seem to see past their own domain these days, to see that other people have other needs.
(Yours is a great post by the way.)
I don’t know the memory limits of Mathematica. It varies a lot. The software is over 35 years old. it grew up in an age when memory was a scarce resource. I have the sense they have maintained that frugality over time. At the same time, it is a huge API that can be tapped with a vast amount of data: geography, cosmology, anatomy.
This is the first Pi with 16GB available. The 5 is the first one that supports NVMe SSDs. That complicates the performance question even further, because any swapping/paging operations will be heavily dependent on storage performance, and speed of a MicroSD card is very bad. If I were getting one, I’d definitely add an Argon40 NEO case with an 512GB M.2 SSD. That would total around $225 or so.
A funny thing happened over the weekend. Explaining Computers YouTuber Christopher Barnett was reviewing the Raspberry Pi 500, a Pi 5 with 8GB. Christopher slammed the micro-HDMI connectors that Raspberry Pi used instead of regular HDMI. He then flipped the unit over to show the M.2 NVMe port – which does not exist! He had drawn it as an animation. Christopher presentation is incredibly dry; this mischievous misdirection was unexpected and delightful – and I hope he doesn’t repeat that stunt for a very long time.
Thanks. It was really helpful drafting this message. There is a line between Mathematica and the Makers; it should be reinforced. They need to know more about each other; they need to shout out loud about each other. I dropped a line to a friend at Make; we may be able to make something spectacular here. This is a tiny audience, but IMHO it is one heck of a sounding board. Thank you – everybody here.
I wanted to show one thing I had done with Mathematica. Two years ago, a friend launched a 110-city tour across the US and Canada for his half-day functional anatomy course. Gil calls himself a somanaut; you can look that up if curious. Aha! This is just like the “traveling salesman problem”; I’ll call it the Rogue Academic’s Roadshow Problem. Thanks, Chat.
Gil’s tour can be mapped easily with his tab-spaced list of cities and dates:
The tour can be visualized with simple tools, but it took over 4 minutes (!!!) for Mathematica to grind this. I think a lot of that time was pulling various city data from the network:
Here’s the map:
Note how Mathematica sweated out all the deets. Too many city names would mess op the image; skip 'em. I then took the sorted city list, calculated the distance, and compared to the actual shortest path:
entitycitylist=Table[Interpreter[“City”][Normal[datesortedtour[All,“City”]][]],{x,1,110}];
TravelDistance[entitycitylist]
21447.2 mi
And then the shortest path:
shortesttour= FindShortestTour [Table[EntityValue[entitycitylist[],“Position”],{x,1,110}]]
{TemplateBox[{“12520.462869711497`”, “"nmi"”, “nautical miles”, “"NauticalMiles"”}, “Quantity”],
12520.5 nmi
Oops. That’s in nautical miles. I’d better change that. 
Also, Gil deliberately chose warm-weather cities during cooler months (and vice versa). Checking the closeness of his tour to the ideal would take some added structuring of the data – maybe a warm/cold bit. For now the Rogue Academic Roadshow Problem remains unsolved.
Can you imagine the thrill of a high-schooler diving in and producing a computation like this: some question that he asked himself and then answered himself? That’s the Maker credo!
I know you can do a lot of this stuff in Python. I believe that Wolfram’s solution is far more cohesive. It’s a proprietary solution, but that funding allows them to curate their world-class database in a way that’s harder to provide in Python. The fact that they provide access to all this stuff through the free Raspberry Pi port is kinda mind blowing – especially when you factor in the computational capabilities of the Raspberry Pi 5.
Sorry for the long message; I just didn’t know how to capture the feeling of beautiful computations.
2 Likes
as a Montrealer I’m deeply offended Montreal(in name only. because I do see the Montreal dot there) was left out of this map.
That’s overall super cool though. I love that kinda thing
2 Likes