large arrays in python (scientific)

[Travis Oliphant]

> > it'll help me predict memory requirements for constructs.
>
> Oh, it's terrible.  Each float is a heap-allocated object;
> then each tuple is another heap-allocated object containing
> pointers to the objects it contains.
>
> I was surprised that SciPy requests the input data in that form.
> Numeric does have an efficient array type, after all...
>

This is Scientific Python not SciPy.  They are different.  SciPy is
"bigger" and requires Numeric to work as most things are done with the
more efficiently stored Numeric arrays.

-Travis