[PYTHON MATRIX-SIG] Final conventions for alpha2

[Carlos Fonseca]

On Sun, 18 Aug 1996, Jim Hugunin wrote:

> Carlos Fonseca wrote:
> 
> > ix=argsort(a)
> > A=take(a,ix)
> > B=take(b,ix)
> >
> > This should be enough.
> 
> You have a good point here, but I think that the issue is you want a
> different function from Konrad.  Alright, writing functions is something
> I can do.  What would you like it to be called?

How do you feel about index(a,indices,axis). Alternatively, select(...) 
would also make sense, I think. Thanks for looking into this.

> > My experience is different. But Matlab doesn't let me write something
> > like the above even for 2d arrays, so I end up using a for-loop.
> 
> So at least we aren't in danger of having you defect back to matlab :)

:-)

> > Something else:
> > 
> > How can I achieve
> > 
> > rank = argsort(argsort(cost))
> <
> > without having to call argsort twice? Something like an inverted take
> > (ekat below :-)
> 
> I see what you're looking for here, but I'm not convinced that having to
> call argsort twice is enough of a burden to justify the function you're
> asking for.  I'd either like to see more examples of how this would be
> useful, or have you wait and ask for this in version 1.1.

Plain indexing should always be more efficient than sorting, but this is
certainly not a problem for me at the moment. I agree that there is
more than enough to concentrate on until v1.0.

Carlos



=================
MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
=================