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

jaropis jaropis@gazeta...
Sun Nov 4 05:50:36 CST 2007


> Given that you felt for the sake of publication you must use Matlab, was
> it worth it then to use scipy to begin with? That is, if you knew you
> would need to convert to Matlab, was scipy worth the trouble?

A very good question. I use scipy because my research is multidisciplinary
and multicenter. This means that I have to write code that will be used in
various places all around the world. I cannot expect all my friends to buy
Matlab. The second reason is that I do not own a copy of Matlab - I used
the copy owned by my University only to test the code. If I had wanted to
use Matlab for this paper I would have had to spend weeks at the
University. Using Python I could work at home. The third reason is probably
the most important: Python is a real programming language and therefore it
is much more convenient. I get medical time series in various formats,
usualy some kind of text files, and python is great at processing them and
unifying all the formats I come accross to the single format which is used
by my software. Everyone who has ever tried to process gigabytes of text
files with Matlab knows that this is extremally difficult. It is also much
easier to write small scripts that have a specific function and send them
over the net to those who need them. 

As I said - the use of Matlab was necessary to include the code for the
various descriptors introduced in the paper - unfortunately Scipy code
would not have been accepted (I know this for a fact). The Python software
used in this paper was much more complex than the Matlab one-liners - I had
to process the recordings and do the calculations. Also, I had to write a
speparate program to produce the animation. Of course this software was not
converted to Matlab - just the mathematical expressions. 

Jarek



More information about the SciPy-user mailing list