# Higher order functions - map, fold and filter

Higher order functions are functions which take in other functions as input, or return a function as output. takeWhile and dropWhile which I covered in the page on Prelude functions are examples of higher order functions. But the three most useful higher order functions are map, fold and filter.

Main> map (+1) [1..5] [2, 3, 4, 5, 6] Main> map (toLower) "abcDEFG12!@#" "abcdefg12!@#" Main> map (`mod` 3) [1..10] [1, 2, 0, 1, 2, 0, 1, 2, 0, 1]

2 - fold

fold takes in a function and folds it in between the elements of a list. It's a bit hard to understand at first - try this example.

Let's add up the first 5 numbers with fold. The command for this is

foldl (+) 0 [1..5] How does this work? First, take the elements of ths list and write them out separately with a space between each.

 1 2 3 4 5

And now we fold the function into the elements. To do this, we write the function in the empty spaces between the elements.

 + 1 + 2 + 3 + 4 + 5

Note that there's nothing before the first (+). That's because we can specify a starting point. So that's the second input foldl asks for - a starting point. In this case, we said zero, so we'll put it in there.

 0 + 1 + 2 + 3 + 4 + 5

And all this gives us the final result
Main> foldl (+) 0 [1..5] 15 And if we change the starting point, we get the corresponding answer. Main> foldl (+) 1 [1..5] 16 Main> foldl (+) 10 [1..5] 25 fold is NOT limited to 'infix' functions! (Infix functions are those where you write the function between the inputs. Examples are 'mod', `div`, + and - ) With a non-infix function you won't write the function between the elements like the example above - you'll write it to the left of a pair of elements, and put lots of brackets everywhere. For example, foldl (max) 0 [3,10,14,6] is the same as (max (max (max (max 0 3) 10) 14) 6) (This is basically how the inbuilt function maximum is implemented in Prelude.hs)

We can also go through the list from right to left, rather than left to right. That's why we use foldl and foldr. In the example with (+) we get exactly the same result since (0 + 1 + 2 + 3 + 4 + 5) is exactly the same as (0 + 5 + 4 + 3 + 2 + 1) However, there are cases where folding from the left gives us different results to folding from the right. It all depends on what function you're folding. For example,
Main> foldr (div) 7 [34,56,12,4,23] 8 Main> foldl (div) 7 [34,56,12,4,23] 0 Now, the fact that I've written heaps about fold doesn't mean it's the most important of the three - just the hardest to explain! I actually use fold a lot less than I use map and filter.

3 - filter

filter is easy. It takes in a 'test' and a list, and it chucks out any elements of the list which don't satisfy that test.

Main> filter (isAlpha) "\$#!+abcDEF657" "abcDEF" Main> filter (even) [1..10] [2, 4, 6, 8, 10] Main> filter (>5) [1..10] [6, 7, 8, 9, 10] A lot of the time, you'll want to write your own tests. The other day I saw a list comprehension which didn't work. allThings:: [Content] -> [Content] allThings contents = [ x | x <- contents, x == (Thing _)] This is basically an attempt at filtering with a list comprehension. This didn't work - remember that you can only use the underscore (the _) on the left hand side of a pattern match, and never on the right hand side of = . What you normally do to work around this problem is to write your own test, like so. isThing:: Content -> Bool isThing Thing _ = True isThing _ = False And then you can rewrite allThings allThings:: [Content] -> [Content] allThings contents = filter (isThing) contents

Go get some practice!

Back to Supp. Notes