Sieve of Eratosthenes in Erlang [closed]

Question

I'm in the process of learning Erlang. As an exercise I picked up the Sieve of Eratosthenes algorithm of generating prime numbers. Here is my code:

-module(seed2).
-export([get/1]).

get(N) -> WorkList = lists:duplicate(N, empty),
          get(2, N, WorkList, []).

get(thats_the_end, _N, _WorkList, ResultList) -> lists:reverse(ResultList);
get(CurrentPrime, N, WorkList, ResultList) -> ModWorkList = markAsPrime(CurrentPrime, N, WorkList),
                                              NextPrime = findNextPrime(CurrentPrime + 1, N, WorkList),
                                              get(NextPrime, N, ModWorkList, [CurrentPrime|ResultList]).


markAsPrime(CurrentPrime, N, WorkList) when CurrentPrime =< N -> WorkListMod = replace(CurrentPrime, WorkList, prime),
                                                                 markAllMultiples(CurrentPrime, N, 2*CurrentPrime, WorkListMod).

markAllMultiples(_ThePrime, N, TheCurentMark, WorkList) when TheCurentMark > N -> WorkList;
markAllMultiples(ThePrime, N, TheCurrentMark, WorkList) -> WorkListMod = replace(TheCurrentMark, WorkList, marked),
                                                           markAllMultiples(ThePrime, N, TheCurrentMark + ThePrime, WorkListMod).

findNextPrime(Iterator, N, _WorkList) when Iterator > N -> thats_the_end;
findNextPrime(Iterator, N, WorkList) -> I = lists:nth(Iterator, WorkList),
                                        if
                                          I =:= empty -> Iterator;
                                          true -> findNextPrime(Iterator + 1, N, WorkList)
                                        end.

replace(N, L, New)-> {L1, [_H|L2]} = lists:split(N - 1, L),
                     lists:append(L1, [New|L2]).

This code actually works :) . The problem is that I have this feeling that it is not the best possible implementation.

My question is what would be the "erlangish" way of implementing the "Sieve of Eratosthenes"

EDIT: OK, Andreas solution is very good but it is slow. Any ideas how to improve that?

Answer 1

Here's a simple (but not terribly fast) sieve implementation:

-module(primes).
-export([sieve/1]).
-include_lib("eunit/include/eunit.hrl").

sieve([]) ->
    [];
sieve([H|T]) ->          
    List = lists:filter(fun(N) -> N rem H /= 0 end, T),
    [H|sieve(List)];
sieve(N) ->
    sieve(lists:seq(2,N)).