tips-computer-programming-scipy

Tips and gotchas for people moving from Octave (or MATLAB) to Python+Scipy. I already know Python well, so i'll probably leave out many things that Python beginners might find surprising, although i'll include the ones i can think of.

• If you don't know Python, read http://docs.python.org/tutorial/ at least up until the chapter "Brief Tour of the Standard Library". At the time of this writing, this is about 60 screenfuls
• Use ipython. Read http://ipython.scipy.org/doc/manual/html/interactive/tutorial.html
• Start ipython with the -pylab argument. This loads pylab, a matlab-style "plot" library
• Read http://ipython.scipy.org/doc/manual/html/interactive/tutorial.html
• Python is always call-by-reference, not call-by-value, and assignments of object to variables also assign refs rather than copying values:

a = array([1,2,3])
b = a
a
b
b[0] = 4
b
a

->

In [1]: a = array([1,2,3])

In [2]: b = a

In [3]: a
Out[3]: array([1, 2, 3])

In [4]: b
Out[4]: array([1, 2, 3])

In [5]: b[0] = 4

In [6]: b
Out[6]: array([4, 2, 3])

In [7]: a
Out[7]: array([4, 2, 3])

If you need to copy a matrix, use the "copy" method, i.e. a.copy():

a = array([1,2,3])
b = a.copy()
a
b
b[0] = 4
b
a

->

In [1]: a = array([1,2,3])

In [2]: b = a.copy()

In [3]: a
Out[3]: array([1, 2, 3])

In [4]: b
Out[4]: array([1, 2, 3])

In [5]: b[0] = 4

In [6]: b
Out[6]: array([4, 2, 3])

In [7]: a
Out[7]: array([1, 2, 3])

• In Python, you gotta use square brackets to index into matrices
• The command "set_printoptions(threshold=nan)" makes it so that entire matrices are printed out, rather than abbreviated
• Matrix shapes are specified using Python tuples, so it's not zeros(3,3), it's zeros((3,3))
• In Python, you can conveniently flatten any matrix with the syntax "a(:)". In Python, "a[:]" doesn't work, and you have to do "reshape(a, prod(shape(a)))" (there's probably a better way that i don't know about; please tell me if you find it)
• Python has a built-in function called "max", but it doesn't work on matrices the way that matlab's max does. To get the maximum element of a matrix, use the max method, not the max built-in function:

a = zeros((3,3))
a[1,1] = 3
max(a)
a.max()

->

In [1]: a = zeros((3,3))

In [2]: a[1,1] = 3

In [3]: max(a)
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)

/home/bshanks/sync/<ipython console> in <module>()

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

In [4]: a.max()
Out[4]: 3.0