Comments on: Remove duplicate values from a List in F#

By: FSharp.it » Another interview question

FSharp.it » Another interview question — Wed, 26 Aug 2009 13:24:48 +0000

[...] made of the two lists created. The rest of the job is calling some standard library functions to filter duplicates, sort and count the elements of the [...]

By: Jon Harrop

Jon Harrop — Fri, 17 Jul 2009 07:31:24 +0000

I think anocka’s comment is wrong. Your algorithm is O(n log n) whereas his is O(n^2).

Also, you can write:

let nub = set >> List.of_seq

By: OJ

OJ — Fri, 18 Jul 2008 06:12:42 +0000

You could make it shorter:

let nub = Set.of_list >> Set.to_list

By: Daniel Furrer

Daniel Furrer — Mon, 30 Jun 2008 11:40:25 +0000

Leaving performance aside, the version that anocka proposed can be adapted to make sure elements in the list do not change order, which is important in some cases.

Here’s what I am using:

let remove_duplicates list =
let add_elem acc_list elem =
if List.mem elem acc_list then
acc_list
else
acc_list @ [elem]
List.fold_left add_elem [] list

By: anocka

anocka — Sun, 27 Apr 2008 14:04:44 +0000

Well, I did not think it through : you algorithm has a O(n*log(n)) complexity while mine is O(n²)…

By: anocka

anocka — Sun, 27 Apr 2008 13:40:10 +0000

This may seem elegant, but I’m afraid this would be to slow on large lists. Another elegant solution uses the fold operator.
First define a function that adds an element to a list if it does not already belong to it :

let add_elem elem list =
if List.mem elem list then list else elem :: list

Then you use it to accumulate on the list to build one without duplicate values :
let nub list = List.fold_right add_elem list []

Fold power !
PS : Actually, that’s ocaml code, but I think it should work on F#.

By: claudio

claudio — Thu, 24 Apr 2008 20:33:21 +0000

I totally agree with you…

By: Akiva

Akiva — Thu, 24 Apr 2008 20:29:18 +0000

It’s precisely this kind of thing that makes me love F# more and more each day.