> you can just initialize with a constant or variable or just use a procedure to initialize with.

Is there an option to leave something uninitialized? I often find the allocation of explicitly uninitialized objects to be a performance necessity in tight loops when I'm working with numpy.