[SciPy-dev] Numerical Recipes (was tagging 0.7rc1 this weekend?)

Xavier Gnata xavier.gnata@gmail....
Sat Dec 20 14:13:57 CST 2008


>> I am not convinced that their interpretation of copyright law is
>> correct, but either way I think ttest_ind and ttest_rel are safe
>> (according to *their* rules).  The code makes use of constructs such
>> as "where" and broadcasting, that aren't available in C.  The author
>> therefore took an idea from NR and reimplemented it in a novel way.
> 
> I haven't looked at the actual code in question here. Just beware that
> there are Fortran 90 code on the NR CD-ROM and web site. In 1996 the "NR
> in Fortran 90" book was published. Their Fortran 90 code does make use of
> where constructs, broadcasting and array slices.
> 
> 
> Sturla Molden

Come on ;)
All these algo are very standard one.
You can refer to this great book to *explain* these pieces of math but
that's it.
If you want to code let's say a heapshort or a golden ratio based
search, you may refer to one another great book to explain how is it
supposed to work. Fine.
That said, there is no many different ways to code such algo in fortran
(it is of course the same whatever the language could be).
If you think there is a need to obfuscate standard pieces of math well
....that is your choice.

If someone try to claim something about maths, you can just remove all
the ref to this strange guy and put a ref to any classical paper written
 by someone reasonable (or event to wikipedia...)

Xavier


More information about the Scipy-dev mailing list