Issue with feature functions in chapter8/optimal_exercise_bi.py about rl-book

[TikhonJelvis]

Thanks for catching that! This specific example is fixed (in #179); we'll do an editing pass through all the code examples at some point, which should help catch the rest. In the meantime, feedback like this is really helpful.

I recall that, in older versions of Python, i would have been in scope outside the list comprehension 🤯. Looks like at least that has been fixed, probably in the transition from 2 to 3.

An alternative would be to have a named function:

def laguerre(i):
    return lambda s: (np.log(1 + np.exp(-s.state / 2 * strike)) * \
            lagval(s.state / strike), ident[i])

ffs += [laguerre(i) for i in range(num_laguerre)]

(I haven't tested this exact code, so there might be some typos.)

Which version seems clearer? I don't have a feel for which approach would be more pythonic.

[coverdrive]

Fixed