GNU Octave vs. Matlab

Discovered a programing tool called Octave a few weeks ago. The language is almost identical to basic Matlab. The discriminating difference is that Octave is open source and free. For my robotics modeling applications, its more than adequate. I will likely stop spending money on Matlab, especially if the Octave organization develops a basic Simulink-like capability. Here's an interesting thread in the Octave website:

http://octave.1599824.n4.nabble.com/Simulink-for-Octave-td3746480.html

[the cyclist]

Please remove this blatant advertisement for inferior software.

[Walter Roberson]

Previous mentions of Octave have been permitted, including in http://www.mathworks.com/matlabcentral/answers/13687-do-you-think-that-matlab-is-expensive-high-price

[the cyclist]

The thread you cite is, in my opinion, of a different character. It is from a true member of the community, for one thing. This thread is a first post, and I believe it to be a thinly veiled advertisement for another product.

[Paulo Silva]

blatant advertisement indeed but that software is free and works for basic stuff.

[Jan]

Xavion wants to share its positive experiences with the great software. He does not want to earn money with this post.

*But* he forgot to ask a question.

[Walter Roberson]

Having just read the thread Xavion cited, it seems to me that the thread does no real harm to TMW, in that the thread shows that the case for switching is not as simple as Xavion had imagined.

[xavion]

My apologies for not including a clear question. The advertisement charge and spam label is unfounded because Octave is free open source software - I'm not here to make money. This thread is not spam. My question is this... what is Mathworks doing in response to Octave? I hope Mathworks takes it seriously. As a steady Matlab/Simulink user for 16 years, and after using Octave for a few weeks, I can see why organizations are reducing the number of Matlab licenses to save money and use Octave for basic needs. Its easy to dismiss Octave and assume its for non-serious work. However, I think the evidence is clear that its usage is growing at universities and small companies and here to stay. J. Simon, thanks for your response.

[Jan]

I do not dismiss Octave. It is an excellent piece of code. I just trust MATLAB more currently, e.g. when creating programs for clinical decision making. Of course one can write bad programs in any programming language. But the extremly well tested code of MATLAB and the powerful user community support my trust in this product.

Nevertheless, I know, that the majority of internet users rely on the Linux driven Google servers and other free software package reached the maximum level of stability also. Therefore I see an important potential for Octave to get more influence for scientific computing in the future - and for SciLab, SciPy, FreeMat also. And I see what TMW is doing in response to Octave: Improve their stable product in small secure steps and offer an efficient support.

Imagine you develop a brake system for a car (a real one, not a simulation) and get some unexpected results. Then you can call your local TMW consultant and ask, if it is a problem of Matlab or of your program. If the brakes are working finally, this is worth 0.1e6$ or 10.0e6$. Now try this with Octave.

As you've written: "The language [Octave] is almost identical to basic Matlab." When using _advanced_ instead of _basic_ Matlab features, the differences between these platforms become more considerable.

@Readers: I'm *not* payed by TMW and this is *not* a shameless advertisment for MATLAB(R), but my personal opinion. Xavion has other preferences than me and other forum users, but I do not agree that a deviating opinion is worth to be tagged as spam.

[Walter Roberson]

Questions about what TMW is planning in response to Octave and the like should be addressed to your sales representative. "Those who say, don't know, and those who know, won't say".

Anything you read here about TWM's plans should be treated as interpretation or speculative fiction or as bovine excrement. TWM doesn't talk about its competitive plans.

[Jan]

@Walter: Really? Is "Improve their stable product in small secure steps and offer an efficient support" an interpretation, speculative fiction or as bovine excrement? In deed, it is my experience in the past years.

[Walter Roberson]

@Jan: I would say it is an interpretation.

[David Verrelli]

Per Jan's comment above — "When using _advanced_ instead of _basic_ Matlab features, the differences between these platforms become more considerable." — it may have been advisable for the OP to resist endorsing Octave until they had more than "a few weeks" of experience with it.

Answer 1

[Jan]

Fine, Xavion. I like Octave also very much. The availability of the source code is a big advantage, because I can search for bugs for my own and adjust features for experiments. E.g. I cannot get MATLAB to allocate the memory for arrays with an 128-bit alignment, but in Octave I can control this -

- when I spend hours and days and weeks of programming, debugging and testing. And as long as I want to work with a running system, I decide to spend money for Matlab for the serious work.

Answer 2

[Daniel Shub]

One discriminating difference between Octave and MATLAB is the FOSS nature of Octave. Another difference is its lack of advanced features. Last time I looked, support for OOP and GUI development were substantially lagging (possibly because MATLAB is progressing in these areas rapidly) and the IDE (including mlint) were substantially behind. I pay TMW for software maintenance so I can get new features, improvements and bug fixes on a regular basis. If you are willing to wait years for new features to trickle down, then Octave might work for you.

[Jan]

You do not have to *wait* years for new features, if you implement them by your own. I've implemented/improved some features for MATLAB also, as you can see in my FEX submissions. And the FEX allows for including some FOSS (free and open source software) nature to your proprietary MATLAB platform.

Therefore I do agree +1.

Answer 3

[Paulo Silva]

The Mathworks has nothing to fear, their software (MATLAB+Simulink+Toolboxes) is very mature and well documented, it is expensive but it does worth the money, Octave can be the starting point until you need more than it can provide or hit one of the snags it still has, MATLAB is more adequate for demanding professionals and academic students.

Answer 4

[Kayemba Luwaga]

It can be hard to admit, but MATLAB allows amazing "shortcuts" in product development you're going to get anywhere else, you can easily arrive at the most optimal system configuration for your product design with MATLAB; then your can fine tune the configuration with other tools like octave especially if you don't have good processing hardware in case of a system/signal processing/control project, Cheers

[Mamta]

Hi can you help me

Answer 5

[Apple]

i have found octave from a course at coursera. now i am going to use it .

Answer 6

[DuyKy Nguyen]

GNUOctave outperform MLab

Octave uses endif endfor endfunction end while separately but MatLab uses only end for all

so quite hard to folow MLab code

DuyKy

Answer 7

[DuyKy Nguyen]

I spend all day to run my old MLab code under Octave

as end use for all if for while so quite hard to follow

I've been using ML since 1995

I'm done with MLab

MLab is a junk expensive tool

[Mamta]

Hi... Can you please help me