08 December 2007

SAGE: open-source mathematics software

"Open Source 'Sage' Takes Aim at High End Math Software", from Slashdot. (The actual site, sagemath.org, is Slashdotted right now.) Here's the article that Slashdot links to (by the way, the links in that article go to malformed URLs, but they're not hard to fix). The leader of the Sage project is William Stein, who with David Joyner wrote an opinion piece in November's Notices stating that mathematical software should be open source. From what I can tell, SAGE includes both new code written specifically for the project and a lot of old code that's been previously used (it appears to incorporate Maxima, GAP, etc.)

I agree with this; although it may very rarely be practical to check all the work that goes into establishing a mathematical result, it should always be at least theoretically possible. It seems acceptable for the internal workings of mathematical software to be closed when it's just students that are using it, or when the serious researchers are just using it to do experiments. But as we start seeing more and more proofs which reduce to "so I reduced the problem to some finite but very long computation, and then I ran this program, and it said my theorem is true!", it becomes more important that someone can actually verify that the program does what the author claims it does. For comparison, see Wolfram on why you don't need to know what's going on inside Mathematica, which is mildly amusing.

I haven't actually used SAGE, but it certainly seems more in line with the open-information ethos of the mathematical community than the proprietary software that's out there. It's a shame I know Maple pretty well, because that creates a disincentive for me to switch if I wanted to.


Rod said...

There was also an article on this on PhysOrg:


Pretty much the same stuff that came on that Networkworld article, only a tad more informative.


Efrique said...

I think the issue is slightly overblown.

Mathematical software being open source just shifts the blackbox back to the compiler.

If that's open source, you just shift the blackbox back to the firmware and hardware.

Two orders of nested blackboxes is doubtless an improvement over three orders of blackbox, but it's still naive to act as if removing the outer layer is taking care of the problem.

Robert Talbert said...
This comment has been removed by the author.
Robert Talbert said...

My January project (we have a Winter Term at my college and I'm off this year) is to learn SAGE. I've downloaded and installed it and dinked around some and... well, it's pretty much just like GAP so far. That's not a bad thing -- bringing GAP and Maxima and other scattered open-source projects under one roof is pretty compelling. But I can tell you now that it won't take off unless it has the ease of use (read: GUI) that Maple, etc. have. There appears to be such a thing for Sage, but I haven't gotten that far yet.

The incentive for me to switch to Sage is that Maple 10 still won't run on my Macbook Pro running OS 10.5, and Maplesoft appears singularly disinterested in offering real help for Leopard users in getting their software to work. I bet I'd get help fast, though, if I were the IT director at a university with 10K+ licenses of Maple on the line. I am personally getting tired of the big math software companies treating end users with utter dismissiveness.