Generators were a mystery to me when I first learned Python. I learned them through an excellent explanation on Stackoverflow about Python yield keyword and generators in general. Generators are more memory efficient than a regular list because they return a sequence lazilily, meaning each item in the sequence gets computed at the last minute.
The other day, I was reading the book Python for Data Analysis and found an interesting usage of Python generators in the appendix.
It was the classic coin change problem and Python generators were used to solve it.
def make_change(amount, coins=[1, 5, 10, 25], hand=None):
hand = [] if hand is None else hand
if amount == 0:
yield hand
for coin in coins:
# ensures we don't give too much change,
# and combinations are unique
if coin > amount or (len(hand) > 0 and hand[-1] < coin):
continue
for result in make_change(amount - coin,
coins=coins, hand=hand + [coin]):
yield result