[SciPy-user] Avoiding For Loops Question

Whitcomb, Mr. Tim tim.whitcomb@nrlmry.navy....
Fri Jun 5 13:19:59 CDT 2009


> > All right guys, last one:
> > 
> > for l in xrange(1,1000):
> >     for m in xrange(0,l+1):
> >         Alm[l][m] = alm[l][m]*cl[l]
> >          Blm[l][m] =alm[l][m]*cl[l]
> > 
> >    Here is one where the second index depends on value of the first.
> 
> In this case, you are still summing across columns (i.e. the 
> first index of all parts of the expression is constant).  The 
> only variation is that in row N, you only want the first N 
> columns.  The output pattern that you get from your for loop 
> looks like:
> x
> x x
> x x x
> ...
> which, in a rectangular matrix, means that it's lower 
> triangular, so you can do something like:

Whoops - got the signs mixed up.  This should be 
A = numpy.tril(alm * cl[:,numpy.newaxis])

Tim


More information about the SciPy-user mailing list