My friend who was visiting last weekend has returned home safely. He left me with something I must have heard before, but I didn’t actually remember: “the plural of anecdote is not data.”
While he was here, we checked the appearance of this blog on Windows under Firefox and Internet Explorer… it looks fine in Internet Explorer, but some of my Mathematica command images are too wide in Firefox (as I said, under Windows). I’m looking at other templates to see if I can fix that. It has always looked fine to me in both Safari and Firefox on my laptop Macs.
Although I do not encourage friends to send me cute links via e-mail, most of them know that I appreciate their sending the best of the best a couple of times a year. Here is a mathematical version of “the 12 days of Christmas”.
In minor news this week, I have an assignment due next Friday in the “computational investing” Coursera course, so I’ll need to put some time into it in the evenings… I started yesterday. (You could search on Coursera in this blog for previous discussions of MOOCs, Massive Open Online Courses; or just google Coursera.)
There is a related class starting up on December 16… I’ve already gotten 3 of the books on its reading list, and a 4th is on the way. Whatever else I may think of the current course, it has done me good just by introducing me to the “capital asset pricing model”. While I still expect to put out an introductory post about the CAPM, I more than suspect that I have only begun to scratch the surface.
I wasn’t at all sure that I wanted the latest book… 752 pages just about asset pricing and portfolios. Then I decided that with tens of thousands of students enrolled in the class, I should buy one of Amazon’s 7 remaining copies before they disappeared. If I think of it, I’ll keep checking during the course to see if Amazon runs out.
In major news this week, Mathematica® version 9 is out. In fact, I have installed it and looked at it. Actually, the 1st thing I did was modify Mathematica’s menu to assign a keystroke to “underscore”. Then I looked at it.
I am amused that Wolfram technical support sent me an e-mail telling me that they were providing a new installer for the Windows version – not a revision of Mathematica 9, but a new installer for it. For one thing, Wolfram knows perfectly well that I run on Macintosh not Windows. For another, no one ever bothered to tell me that version 9 was available. I found out about it by reading mathgroup (the USENET newsgroup for Mathematica. Strictly speaking, I think “mathgroup” refers to the Wolfram version – the USENET version that I actually follow is comp.soft-sys.math.mathematica. And you see why I call it “mathgroup”.
There are a lot of additions in version 9… more than I will talk about today. A couple of them I expect to get to in the next few months: image processing and signal processing, in particular, which I’ve never played with and which now have even more capabilities. I expect to try them out under the lash of 2 Coursera classes.
On the other hand, there were several things I had to take a look at.
Units. I can’t say I’m at all fond of what they’ve done. Instead of typing, for example,
and then asking Mathematica to convert it to PSI…
we have to start with something like
and go from there. Since Dave Park’s “extend units” package still runs, and the old “physical constants” still loads, I’ll stay with the old-fashioned way, even in v9.
Time series. It looks to me like they’ve just barely begun establishing functionality… and the add-on time series package for version 8 cannot be loaded in v9… so I’ll have to stay with version 8 for time series. A shame, since unless you have the add-on, and version 8, you can’t duplicate most of what I might show you.
Symbolic tensors. They have an impressive stress/strain example, but it doesn’t look all that useful to me in general.
Colors. They say they have some new color schemes, but they didn’t leap off the page at me. More significantly, from my point of view, Mathematica’s translation from RGB to CMYK is still not invertible: I have gone back and checked the failed conversions described in this post. They still fail, exactly the same way.
the R statistics package. I’m glad to know they have an interface to it, now that I’ve had an introduction to R… but it remains to be seen if I will actually need to combine Mathematica and R. (Hmm. R can do time series analysis….)
Finally, control theory. They’ve added a couple of fancy things that I will want to play with: PID tuning and a Smith predictor. On the other hand, my 1st attempts at Bode plots with multiple colors have not fixed the problems described in this post. Nevertheless, I will keep playing around with version 9.
But… they have added one thing which I am delighted to see. I do not yet know how pervasive it is. Although I’ve never mentioned it, both “control system professional” and Mathematica version 8 cannot handle delays – that is, negative exponentials – in transfer functions. Yes, the inverse Laplace transform works fine on such a transfer function – that is, until you close the loop. But Mathematica’s control theory commands could only handle rational functions, the ratio of two polynomials, so it was necessary to use Pade approximations for exponentials.
Well, I have confirmed that with v9 I can now get a Bode plot for the most common of all chemical engineering transfer functions: first-order plus delay. (That is, .) Maybe before I put this out, I’ll find out if “output response” can cope with delays…. I tried quickly, and it didn’t work, but I’m not sure I tried carefully.
Don’t get the wrong impression: version 9 has a lot of nice new features or capabilities; I have not, however, checked to see if my version 7 notebooks are broken by it. I hope not. Truth to tell, I haven’t yet checked to see what version 8 did to most of my notebooks – I moved to version 8 for control theory and wavelets, and haven’t run much of anything else in a while.
And that’s all for today.