computing a weighted sum

[andreif@mail.dntis.ro]

Suppose I have a list of n floats x and a list of n floats w and I want
to compute x[0]*w[0] + .. + x[n-1]*w[n-1].

Is there some elegant expression (perhaps using lambda) to have it done
in one statement ? As in :
y = lambda x,w : ...

I ask because the way I am doing it now :
y = 0
for i in range(0,n): y += x[i]*w[i]

doesn't seem very pythonic :)

Thanks,
Andrei

[Will McGugan]

Re: computing a weighted sum

andreif@mail.dn tis.ro wrote:[color=blue]
> Suppose I have a list of n floats x and a list of n floats w and I want
> to compute x[0]*w[0] + .. + x[n-1]*w[n-1].
>
> Is there some elegant expression (perhaps using lambda) to have it done
> in one statement ? As in :
> y = lambda x,w : ...
>
> I ask because the way I am doing it now :
> y = 0
> for i in range(0,n): y += x[i]*w[i]
>
> doesn't seem very pythonic :)
>[/color]
I'll take a stab at that!

In Python 2.3

sum( [ _x * _w for _x, _w in zip( x, w ) ] )

or in 2.4

sum( _x * _w for _x, _w in zip( x, w ) )

You may want to use itertools.izip in place of zip if the lists are large.

Will McGugan

[Duncan Booth]

Re: computing a weighted sum

wrote:
[color=blue]
> Suppose I have a list of n floats x and a list of n floats w and I want
> to compute x[0]*w[0] + .. + x[n-1]*w[n-1].
>
> Is there some elegant expression (perhaps using lambda) to have it done
> in one statement ? As in :
> y = lambda x,w : ...[/color]
[color=blue][color=green][color=darkred]
>>> x = [1, 2, 3]
>>> w = [4, 5, 6]
>>> sum(a*b for (a,b) in zip(x, w))[/color][/color][/color]
32

[andreif@mail.dntis.ro]

Re: computing a weighted sum

Thanks Will, the 2.4 expression looks really nice.

[Christos TZOTZIOY Georgiou]

Re: computing a weighted sum

On 16 Mar 2005 06:49:09 -0800, rumours say that andreif@mail.dn tis.ro might have
written:
[color=blue]
>Suppose I have a list of n floats x and a list of n floats w and I want
>to compute x[0]*w[0] + .. + x[n-1]*w[n-1].
>
>Is there some elegant expression (perhaps using lambda) to have it done
>in one statement ? As in :
> y = lambda x,w : ...
>
>I ask because the way I am doing it now :
> y = 0
> for i in range(0,n): y += x[i]*w[i]
>
>doesn't seem very pythonic :)[/color]

Your method seems to be the one closest to what's generally considered as
pythonic.

Anyway, a functional equivalent:

..>> from itertools import starmap, izip
..>> import operator
..>> x= [1,2,3,4]
..>> w=[3.0, 6.0, 9.0, 12.0]
..>> sum(starmap(ope rator.mul, izip(x,w)))
90.0
..>>
--
TZOTZIOY, I speak England very best.
"Be strict when sending and tolerant when receiving." (from RFC1958)
I really should keep that in mind when talking with people, actually...

[Steven Bethard]

Re: computing a weighted sum

Will McGugan wrote:[color=blue]
> In Python 2.3
>
> sum( [ _x * _w for _x, _w in zip( x, w ) ] )
>
> or in 2.4
>
> sum( _x * _w for _x, _w in zip( x, w ) )[/color]

Any reason for the leading underscores? If you're trying to avoid
polluting your namespace, you should note that generator expressions
don't leak their loop variables, so you can write the second as:

sum(x*w for x, w in zip(x, w))

without any worries of overwriting x and w. (Of course I would probably
name them something different anyway, e.g. x_item and w_item...)

STeVe

[andreif@mail.dntis.ro]

Re: computing a weighted sum

Even if language permits
sum(x*w for x, w in zip(x, w))
would seem confusing for anyone watching the code

Maybe
sum(xi*wi for xi, wi in zip(x, w))
would be more appropiate

Andrei

[Fernando Perez]

Re: computing a weighted sum

andreif@mail.dn tis.ro wrote:
[color=blue]
> Suppose I have a list of n floats x and a list of n floats w and I want
> to compute x[0]*w[0] + .. + x[n-1]*w[n-1].
>
> Is there some elegant expression (perhaps using lambda) to have it done
> in one statement ? As in :
> y = lambda x,w : ...
>
> I ask because the way I am doing it now :
> y = 0
> for i in range(0,n): y += x[i]*w[i]
>
> doesn't seem very pythonic :)
>
> Thanks,
> Andrei
>[/color]

import Numeric
print Numeric.dot(x,w )

Best,

f

[John Machin]

Re: computing a weighted sum


Fernando Perez wrote:[color=blue]
> andreif@mail.dn tis.ro wrote:
>[color=green]
> > Suppose I have a list of n floats x and a list of n floats w and I[/color][/color]
want[color=blue][color=green]
> > to compute x[0]*w[0] + .. + x[n-1]*w[n-1].
> >
> > Is there some elegant expression (perhaps using lambda) to have it[/color][/color]
done[color=blue][color=green]
> > in one statement ? As in :
> > y = lambda x,w : ...
> >
> > I ask because the way I am doing it now :
> > y = 0
> > for i in range(0,n): y += x[i]*w[i]
> >
> > doesn't seem very pythonic :)
> >
> > Thanks,
> > Andrei
> >[/color]
>
> import Numeric
> print Numeric.dot(x,w )[/color]

Indeed. Horses for courses. Anyway, people who reinvent the wheel often
fall to arguing among themselves whose polygon is the best
approximation to a circle, and forget to reinvent the axle. Wouldn't
happen in this newsgroup, of course :-)

[Raymond Hettinger]

Re: computing a weighted sum

[Christos TZOTZIOY Georgiou][color=blue]
> Anyway, a functional equivalent:
>
> .>> from itertools import starmap, izip
> .>> import operator
> .>> x= [1,2,3,4]
> .>> w=[3.0, 6.0, 9.0, 12.0]
> .>> sum(starmap(ope rator.mul, izip(x,w)))
> 90.0[/color]

Gack! starmap() is only for situations where the data is already in tuple form.
If it inputs are already distinct, imap() is the preferred form.

FWIW, the answer was already in the docs (itertools recipes):

def dotproduct(vec1 , vec2):
return sum(imap(operat or.mul, vec1, vec2))


Raymond Hettinger

[Christos TZOTZIOY Georgiou]

Re: computing a weighted sum

On Thu, 17 Mar 2005 08:11:11 GMT, rumours say that "Raymond Hettinger"
<vze4rx4y@veriz on.net> might have written:
[color=blue]
>[Christos TZOTZIOY Georgiou][color=green]
>> Anyway, a functional equivalent:
>>
>> .>> from itertools import starmap, izip
>> .>> import operator
>> .>> x= [1,2,3,4]
>> .>> w=[3.0, 6.0, 9.0, 12.0]
>> .>> sum(starmap(ope rator.mul, izip(x,w)))
>> 90.0[/color]
>
>Gack! starmap() is only for situations where the data is already in tuple form.
>If it inputs are already distinct, imap() is the preferred form.
>
>FWIW, the answer was already in the docs (itertools recipes):
>
> def dotproduct(vec1 , vec2):
> return sum(imap(operat or.mul, vec1, vec2))[/color]

What, you're some kind of expert on itertools now? :-)

You are of course absolutely correct. Let my post stand as an example of
itertools misuse.
--
TZOTZIOY, I speak England very best.
"Be strict when sending and tolerant when receiving." (from RFC1958)
I really should keep that in mind when talking with people, actually...