PDL: Perl Data Language
I have a new tool to play with: The Perl Data Language. The idea is to provide a numerical-processing backbone underneath Perl. All the tough stuff is written in C and abstracted away. Less memory and fewer CPU cycles. Conceptually, it seems right on the money. Perl may not be the prettiest thing out there, but I’m quite used to it and very productive with it. Also, there are a zillion libraries out there for it, ready to download. I played with it for a while as Matlab replacement, but it doesn’t really have a complete featureset from my perspective. Also, while the documentation is comprehensive, it is very badly organized. I thought it was quite difficult just to look up simple functions.
In addition, setting up plotting was nontrivial: The PDL module itself was easy, but installing PGPLOT is pretty arcane. After fiddling with it for a bit, I found this page that recommended a script that does everything for you. Then I installed the PGPLOT perl library and everything seemed to be fine, except for a complete lack of features. As far as I can tell, one can’t even plot using the second y-axis. I tried to install the alternative library, PLplot, but couldn’t get the perl module to install, after successfully installing the libraries themselves.
What I had been hoping for was something end-to-end for doing math in Perl. Right now, I use a lot of Data::Dumper reads and evals, combined with sending data to stdout for Gnuplot. It is rather cumbersome, so I typically choose the speed (and horrible syntax) of Matlab, particularly now that I have broken down and purchased a copy for my home laptop. Anyway – the PDL guys are definitely on the right track, but it has a little ways to go before it can really replace Matlab for me.
October 31st, 2006 at 7:42 pm
I use perl and PDL for most of my time series and graphic analysis. It took a bit of time to learn the PGPLOT
functions, but I use them now to plot multiple functions on a graph, graph in colors, and even draw false color
maps. I eventually learned how to make my own mapping color schemes, stored in FITS files. Once you make the
transition to PDL, you’ll find you depend often on features lacking in other math languages. I have significant
experience also in Octave, R, and Yorick. I use Yorick sometimes for its speed, but use perl PDL the most.