Finding the argmaximum of a list in Haskell

Question

I've been doing project euler problems in Haskell, and often I've found the need to get the value that gives the maximum of a function.

For example,

answer = argmaximum $ map collatzLength [1..10^6]

Where collatzLength is a previously defined function.

I came up with this code, which is heavily influenced by learning Ocaml. I run a helper function that has variables that keep track of the best value index and best index so far. It works perfectly for positive lists, but it seems like there's a better way.

Is there a less clunky way to implement argmaximum?

argmaximum :: (Ord a, Num a) => [a] -> a
argmaximum lst = helper lst 0 0 0 where
 helper (x:xs) current bestVal bestIndex
 | current > bestVal = helper xs (current + 1) x current
 | otherwise = helper xs (current + 1) bestVal bestIndex
 helper null _ _ bestIndex = bestIndex

• cole: I believe you are looking for maxmimumBy (comparing collatzLength) [1..10^6] or maximumBy (compare `on` collatzLength) [1..10^6]. I’m not sure I understand how your function works, unless current > bestVal is a typo and meant to be x > bestVal. It’s late so I’m afraid I cannot review more thoroughly - I do hope at least one of my suggestions type checks.

Answer 1

What you have implemented here is strictly speaking not argmaximum, but finding a maximum index in a list. The difference is that your argmaximum will not return the correct value for invocations that don't map from [1..?]

Instead of that, you probably wanted something that's typed as follows:

argmax :: (Ord a, Num a) => (b -> a) -> [b] -> b
argmax f (d:ds) = ...

That way you can write argmax collatzLength [500..10^6] and still get the correct answer :)

By the way: @cole suggested using maximumBy and comparing for the implementation, which I wholeheartedly agree on :)