[ANN] mlabwrap-1.0final

[Alexander Schmolck]

I'm pleased to finally announce mlabwrap-1.0:

Project website
---------------
<http://mlabwrap.sourceforge.net/>


Description
-----------

Mlabwrap-1.0 is a high-level python to matlab(tm) bridge that makes calling
matlab functions from python almost as convenient as using a normal python
library. It is available under a very liberal license (BSD/MIT) and should
work on all major platforms and (non-ancient) python and matlab versions and
either numpy or Numeric (Numeric support will be dropped in the future).


Examples
--------

Creating a simple line plot:

from mlabwrap import mlab; mlab.plot([1,2,3],'-o')

Creating a surface plot:

Creating a neural network and training it on the xor problem (requires netlab)

net = mlab.mlp(2,3,1,'logistic')
net = mlab.mlptrain(net, [[1,1], [0,0], [1,0], [0,1]], [0,0,1,1], 1000)

What the future holds
---------------------

Please note that mlabwrap-1.0 will be the last non-bugfix release with Numeric
support. Future versions of mlabwrap will require numpy be a part of scipy's
scikits infrastructure (so the package name will be ``scikits.mlabwrap``) and
use setuptools rather than distutils so that it should be possible to
automatically download and install via EasyInstall.

The next major version of mlabwrap should also bring more powerful proxying
and marshalling facilities, but the default conversion behavior might be
changed to reflect the fact that matlab is becoming increasingly less
``double`` (-matrix) centric; although wrappers for old-style behavior will be
provided if backwards-incompatible interface changes are introduced, for
upwards compatibility it is recommended to explicitly pass in float64 arrays
rather than e.g. lists of ints if the desired input type that matlab should
see is a double array (i.e. use ``mlab.sin(array([1., 2., 3.])`` rather than
``mlab.sin([1,2,3])`` for production code in order to be on the safe side)).
Please have a look at <http://www.scipy.org/Mlabwrap> if you're interested in
the ongoing development of mlabwrap and planned features.


Feedback and support
--------------------

The preferred formum for users to request help and offer feedback and keep
informed about new releases is mlabwrap-user:

<https://lists.sourceforge.net/lists/listinfo/mlabwrap-user>

the list is low-volume and subscription is recommended.

Discussion of mlabwrap development takes place on the scipy-dev (please
mention mlabwrap in the subject line):

<http://projects.scipy.org/mailman/listinfo/scipy-dev>

cheers,

Alexander Schmolck, mlabwrap author and maintainer

 

[Stef Mientki]

I'm pleased to finally announce mlabwrap-1.0:

Project website
---------------
<http://mlabwrap.sourceforge.net/>


Description
-----------

Mlabwrap-1.0 is a high-level python to matlab(tm) bridge that makes calling
matlab functions from python almost as convenient as using a normal python
library. It is available under a very liberal license (BSD/MIT) and should
work on all major platforms and (non-ancient) python and matlab versions and
either numpy or Numeric (Numeric support will be dropped in the future).

Probably quit a large and nice job, to embed MatLab in Python.
But, I've the idea I'm missing something ...
.... coming from MatLab, a few months ago I tried Python,
.... and at the moment I've decided to move completely from MatLab to Python.
All new programs will be written in Python and the old programs will be translated.

Is my decision wrong ?

What can do MatLab, that can't be done in Python (SciPy) ?
(Ok I accept there's no SimuLink/PowerSim).

Both environments can create any program you like.
If MatLab is your standard / favorite why not stick to MatLab ?

Please enlighten me.

cheers,
Stef Mientki

 

[Aldarion]

What can do MatLab, that can't be done in Python (SciPy) ?
(Ok I accept there's no SimuLink/PowerSim).

What about SVD(Singular value decomposition) for Sparse Matrix.
I can't find this in numpy/scipy, someone told me it's in matrix.
though I don't know how to get it in mlabwrap.

 

[Alexander Schmolck]

Probably quit a large and nice job, to embed MatLab in Python.

Actually, it's not much code -- mlabwrap.py is less than 500 lines of code.

But, I've the idea I'm missing something ... ... coming from MatLab, a few
months ago I tried Python, ... and at the moment I've decided to move
completely from MatLab to Python. All new programs will be written in Python
and the old programs will be translated.

Is my decision wrong ?

Depends on your circumstances, but in many cases just switching to a different
language and rewriting everything from scratch (possibly even at a single go,
if there are many interdependencies) isn't practical. Mlabwrap help here,
because it allows you to either migrate piecemeal or mix and match (i.e.
mostly use python, resorting to matlab libraries where necessary).

What can do MatLab, that can't be done in Python (SciPy) ?
(Ok I accept there's no SimuLink/PowerSim).

There are plenty of commercial and free libraries and environments for matlab
that have seen considerable development by domain experts don't have a
comparable python equivalent. I'm currently sitting in a cafe in Berkeley,
because researchers at the universities of Berkeley and Cambridge are
interested in doing their neuroimaging analysis in python, but still need to
interface to existing matlab packages like SPM.

Both environments can create any program you like.

Not really.

If MatLab is your standard / favorite why not stick to MatLab ?

Because matlab is much limited as a programming language once you stray
outside linear algebra. Many people also seem to be unhappy about licencing
issues . Finally, how often did matlab crash on you? And python?

Please enlighten me.

My pleasure.

'as