meinardmueller / pcp Goto Github PK

In the recent versions of the Python documentation, the options that are emphasized for string formatting are:

1. formatted string literal (f-strings),
2. str.format method,
3. string slicing and concatenation

The printf-style %-based method is only briefly mentionned and referred to as "old-style formatting". (See https://docs.python.org/3/tutorial/inputoutput.html#formatted-string-literals.)

While the feature is not formally deprecated, I'd rather have students learn f-strings instead of printf-style formatting, as in:

>>> string = f'x = {x}'

(I would not use the recent syntax string = f'{x = }' in such a course, but YMMV). Of course, since this preparation course is not meant to be comprehensive, if the f-string approach is considered too heavy, the 3rd option (concatenation) could be an option too:

>>> string = 'x = ' + str(x)

Problem description:
Binder fails building our environment with the following error:
package nbconvert-5.6.0-py27_1 requires python >=2.7,<2.8.0a0, but none of the providers can be installed

This is strange because we fix python=3.7.* in our environment, but it seems to load a python=2.7 version of the package.

Unit 3 states that row vectors and columns vectors are special cases of matrices (with either 1 column or 1 row). While this is also how I see things, however idiomatic NumPy code would probably represent matrices as 2D arrays but row and column vectors as 1D arrays. The same object can then be interpreted and used as a row or column vector in different contexts.

Instead of

>>> A = np.array([[0, 1], [2, 3]])
>>> x = np.array([[4], [5]])
>>> y = np.array([[6, 7]])
>>> np.dot(A, x)
array([[ 5],
       [23]])
>>> y.dot(A)
array([[14, 27]])
>>> np.dot(y, A)
array([[14, 27]])
>>> np.dot(y, x)
array([[59]])
>>> np.dot(x, y)
array([[24, 28],
       [30, 35]])

one would generally do

>>> A = np.array([[0, 1], [2, 3]])
>>> x = np.array([4, 5])
>>> y = np.array([6, 7])
>>> np.dot(A, x)
array([ 5, 23])
>>> np.dot(y, A)
array([14, 27])
>>> np.dot(y, x)
59
>>> np.outer(x, y)
array([[24, 28],
       [30, 35]])

I'd rather see that latter kind of idiom used (and have the solution of exercise 2 updated accordingly).