[SciPy-user] Subject: Re: Scientific Python publications

C.J.Lee@tnw.utwen... C.J.Lee@tnw.utwen...
Sun Nov 4 11:28:12 CST 2007


Perhaps a compromise(for the moment) is to provide both. So minimal matlab scripting to allow publishable and the full python code also available. That way the phd students learn that there is an alternative available and the matlab *only* attitude will die out with retirements. This way you meet the requirements for the peer review and you can grow the python neuroimaging community

It might also be interesting to attempt publishing results which show the differences in functionality between the matlab based code and python based code, particularly if you can find a test case where the matlab toolbox is definitively wrong. (I am not involved in neuroimaging so I can't really say anything about the feasibility of this)

Cheers
Chris


-----Original Message-----
From: scipy-user-bounces@scipy.org on behalf of Matthieu Brucher
Sent: Sun 11/4/2007 5:52 PM
To: SciPy Users List
Subject: Re: [SciPy-user] Subject: Re: Scientific Python publications
 
Hi,

It's even worse. If you design a new algorithm with results that cannot be
verified with SPM, you can't ublish anything (worse : the neurologists do
not want to publish results with your algorithms).
The fact that SPM is GPL is not relevent for me. What bothers me is that
there is a conflict of interest between the SPM team and the editorial board
of some journals et the fact that even you demonstrated that your algorithms
are correct, you can't publish something that SPM cannot coroborate (knowing
that SPM relies on some heavy hypothesis that are not even correct).

Matthieu

PS : I really hope nipy will go on growing and that people at my lab will
collaborate, at last, even if it is not for mainstream algorithms (SPMs,
...)

2007/11/4, Matthew Brett <matthew.brett@gmail.com>:
>
> Hi,
>
> > > In my field, the same problem arises with a Matlab toolbox, SPM. You
> > > can't write an article without using this toolbox, especially if you
> > > want to publish it in NeuroImage (the editor in chief is one of the
> > > writter of the toolbox).
> > This is really SAD:
> > you're doing medical research, investigating and describing some
> > physiological phenomena,
> > well I can assume they are interested what algorithm you used,
> > to verify your conclusions,
> > but it's absolutely absurd, they make it obligatory that you write your
> > manuscript with a pen of a certain brand !!
>
> Hmm - I'm in the same field, and trying to work out a way to fix this
> with:
>
> http://neuroimaging.scipy.org
>
> The problem is not as silly as it seems.  Neuroimaging is moderately
> complex, and it's difficult for reviewers to understand what has been
> done if you have used a variety of software packages.  The advantage
> of SPM, in this case, is that the code for the neuroimaging part is
> open source (GPL), it covers the whole range of processing (more or
> less), and it's fairly easy to extend.  The disadvantage is that - you
> need matlab - and that matlab is so clunky as a programming language
> that it deals very badly with the increasing need for complexity,
> meaning it is harder and harder to develop in SPM as time goes by.
> Our thought was that the only way out of this was to develop a system
> that would allow you to call the core algorithms of SPM within python,
> and develop new ones that you could validate against SPM (and other
> tools) as the standard.  But it's hard work getting that right,
> because our community is so fragmented.
>
> Best,
>
> Matthew
>
> Matthew
> _______________________________________________
> SciPy-user mailing list
> SciPy-user@scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
French PhD student
Website : http://miles.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher



More information about the SciPy-user mailing list