Comments on: The basics of…Matlab

By: Rich

Rich — Tue, 24 Feb 2009 14:09:50 +0000

Hi Scott,

Thanks for the input some very useful points! To respond in order:

- I admit I’ve never explicitly speed-trialled the FOR loops (sorry!), but I’m very glad to hear that they’re becoming faster. Do you publish bench-marking data?

- Great news about the multi-threading! This is a feature (eg. in IDL) that I *really* like. I’m very happy to hear that more and more Matlab functions are using it.

- we’ve used the old version of the Matlab OO. I’ve got a book about the new version sat on my desk as I type! :-)

By: Scott Hirsch

Scott Hirsch — Tue, 24 Feb 2009 13:38:22 +0000

Thanks for a great discussion of MATLAB, along with other popular programming languages for scientists. It’s nice to hear where you find it useful, and which things you’d like to see improved.

While for loops are slower in MATLAB than in C, we hope that you’ve noticed them getting faster over the years as we’ve continued to enhance the JIT Compiler and Accelerator that silently compile much of your code down to machine code.

I’d also like to mention that MATLAB also offers implicit multithreading for many operations and functions, allowing your code to leverage multiple cores. We first introduced multithreading in R2007a, though it was not turned on by default until R2008a. We are multithreading a handful of additional functions each release, based on our studies of where we can have the best impact on code performance.

Finally, I’m curious which incarnation of the OO programming you guys have used. I completely agree that the system we introduced back in the mid-90s never felt well integrated into the language. We introduced a completely revised system in R2008a that caught MATLAB up to other OO languages, with (hopefully!) a much more integrated feel.

Thanks again, and happy blogging!

By: LeoNazDo

LeoNazDo — Thu, 19 Feb 2009 21:51:06 +0000

we are taking a class called IP,
not Internet Programing ,but Image Processing.
and i got a nice new info about this language,
since our study is based on it,
thx a lot,

Best Regards,
Nazeeh.

By: Will Dwinnell

Will Dwinnell — Thu, 11 Dec 2008 12:11:52 +0000

Though MATLAB has a reputation for slowness, I find it speedy enough for use with re-coding in other languages. It’d be interesting to see an examination of execution times for tasks typically performed in MATLAB, implemented in MAATLAB and other langauges.

By: gioby

gioby — Tue, 25 Nov 2008 22:31:58 +0000

I have used octave and liked it.
It lacks some of the proprietary libraries from matlab (maple?), but I prefer to use an open source program if it is possible.

By: Rich

Rich — Thu, 20 Nov 2008 09:48:29 +0000

Thanks for your comments! You both make some very valid points.

I’ve not used GNU Octave, although I know some people who swear by it. I do use R quite a bit and find it very good. We’ll be doing a “basics of…R” post at some point in the near future.

Thanks again for your input!

By: Guy

Guy — Thu, 20 Nov 2008 01:12:12 +0000

If you’re addicted to Matlab but don’t want to spend the money, there’s always GNU Octave (www.gnu.org/software/octave/), which has almost identical syntax.

By: nico

nico — Tue, 18 Nov 2008 07:26:50 +0000

Personally, I try to convince people NOT to use Matlab. There are opensource equivalents that are as (or even more) powerful like R or Scilab with which you can achieve exactly the same.

These packages have plenty of support from the online community and are updated regularly. So, unless you absolutely need Matlab (because you can’t be bothered to change your preexisting code) think twice before spending money on a licence.