Reputation: 633
I want to define a simple function in Haskell:
nzp :: [Int] -> (Int,Int,Int)
that accepts a list of integers as input and returns a triple (a,b,c) where a is the amount of numbers in the list less than 0, b is the amount equal to 0 and c is the amount higher than zero. For example,
nzp [3,0,-2,0,4,5] = (1,2,3)
I have to define this function recursively and I can only traverse the list once. How can I do this? I can't seem to grasp the concept of recursively creating a tuple.
Most Regards
Upvotes: 1
Views: 3869
Reputation: 714
I am a little over a couple of months into Haskell. A helper/auxilliary function would make running this solution easier.
s3 [] (l,g,z) = (l,g,z)
s3 (x:xs) (l,g,z) = if x<0
then (s3 xs (l+1,g,z))
else if x>0
then (s3 xs (l,g+1,z))
else (s3 xs (l,g,z+1))
This is run with s3 [1,0,-2,3,4,-7,0,-8] (0,0,0)
producing
(3,3,2)
.
Just about any primitive recursive function can be translated into a fold.
Upvotes: 1
Reputation: 672
Here are some pointers:
To use recursion to build up a value, you need to pass the previous version of the value as an argument, so write
nzp_helper :: [Int] -> (Int,Int,Int) -> (Int, Int, Int)
Decide what the answer is when the list is empty
nzp_helper [] runningTotals = -- what's the answer if the list is empty?
Decide what the answer is when there's something in the list
nzp_helper (i:is) (negatives, positives, zeros) =
| i < 0 = -- write stuff here
| i == 0 = -- I hope you're getting the idea
Kick the whole thing off by defining nzp
using nzp_helper
.
nzp is = nzp_helper is -- one final argument - what?
Upvotes: 5
Reputation: 144136
Instead of thinking of trying to create a single tuple recursively, you should think about updating an existing tuple containing the counts based on given value. This function would look something like:
update :: (Int, Int, Int) -> Int -> (Int, Int, Int)
update (l,e,g) x | x < 0 = (l+1, e, g)
update (l,e,g) x | x == 0 = (l, e+1, g)
update (l,e,g) x | x > 0 = (l, e, g+1)
Then you can traverse the input list using foldl
and accumulate the output tuple:
nzp :: [Int] -> (Int, Int, Int)
nzp = foldl update (0,0,0)
EDIT: As @leftroundabout points out in the comments, you should avoid using foldl
since it can lead to space leaks - you can find an explanation in Real World Haskell. You can use the strict version of foldl
, foldl'
in Data.List
import Data.List
nzp = foldl' update (0,0,0)
Upvotes: 3
Reputation: 67507
New to Haskell too! Perhaps not proper but here is my solution which works. Define an auxiliary function that accumulates the n,z, and p values
let f (x:xs, (n, z, p)) | x < 0 = f (xs, (n+1, z, p))
| x == 0 = f (xs, (n, z+1, p))
| otherwise = f (xs, (n, z, p+1))
f ([], (n, z, p)) = ([], (n, z, p))
and define nzp in terms of the auxiliary function
let nzp x = snd $ f (x,(0,0,0))
to verify
Prelude> nzp [-1,1,1,-1,0]
(2,1,2)
Upvotes: 1
Reputation: 120711
I can't seem to grasp the concept of recursively creating a tuple.
I shouldn't think so – in fact, it's impossible! (At least not without GHC extension havoc.)
No, you need to create a tuple in one go. The important thing needed for your function: since you may only traverse the list once, you need to pull through the entire tuple at once as well. As this is apparently homework, I shall instead show how it works with a fold (that is in fact preferrable, but translates quite directly to recursion):
nzp = foldr switchIncr (0,0,0)
where switchIncr x (negatives, zeroes, positives)
| x<0 = (succ negatives, zeroes, positives)
| x==0 = (negatives, succ zeroes, positives)
| otherwise = (negatives, zeroes, succ positives)
Upvotes: 2