Python: numpy and matplotlib anomaly

Question

This is the first time I am using matplotlib and numpy.

Here goes the problem:

If I goto python cli, the intended code works fine. Here is that code

>>> from numpy import *
>>> y = array([1,2])
>>> y = append(y, y[len(y) - 1]+1)
>>> y
array([1, 2, 3])

But if I use it with matplotlib in a script I get this error.

line 26, in onkeypress
y = append(y, y[len(y) - 1]+1)
UnboundLocalError: local variable 'y' referenced before assignment

Here is my script:

from matplotlib.pyplot import figure, show
from numpy import *
figzoom = figure()
axzoom = figzoom.add_subplot(111, xlim=(0,10), ylim=(0, 10),autoscale_on=True)
x = array([1, 2  ])
y = array([1, 10 ])
def onkeypress(event):
    if event.key == "up":
        y = append(y, y[len(y) - 1]+1)
        x = append(x, x[len(x) - 1]  )
        axzoom.plot(x,y)

I tried "append"ing to a different array,say y1, and then y = y1.copy(). But I still get the same error. I must be missing something trivial here???!!!

Answer 1

It may work if you change the variables to global

def onkeypress(event):
    global y, x
    ...

Answer 2

Unless you include global y in your onkeypress() function, the y you're assigning to is scoped locally to the function. You can't use y on the right side of the assignment statement in which you're defining the local variable.

Answer 3

When you assign to a variable inside a function, python creates a new variable that has local scope, and this new variable also hides the global variable.

So, the x and y inside onkeypress are local to the function. Hence, from python's point of view, they are uninitialized, and hence the error.

As GWW points out - declaring x, y as global will solve the problem. Also, if you do not assign x, y any new value, but only use their previously existing value, those values will refer to the global x, y.