Part of my program requires me to be able to randomly shuffle list elements. I need a function such that when i give it a list, it will pseudo-randomly re-arrange the elements in the list.
A change in arrangement Must be visible at each call with the same list.
My implementation seems to work just fine but i feel that its rather long and is increasing my code base and also, i have a feeling that it ain't the best solution for doing this. So i need a much shorter implementation. Here is my implementation:
-module(shuffle). -export([list/1]). -define(RAND(X),random:uniform(X)). -define(TUPLE(Y,Z,E),erlang:make_tuple(Y,Z,E)). list(L)-> Len = length(L), Nums = lists:seq(1,Len), tuple_to_list(?TUPLE(Len,[],shuffle(Nums,L,[]))). shuffle([],_,Buffer)-> Buffer; shuffle(Nums,[Head|Items],Buffer)-> {Pos,NewNums} = pick_position(Nums), shuffle(NewNums,Items,[{Pos,Head}|Buffer]). pick_position([N])-> {N,[]}; pick_position(Nos)-> T = lists:max(Nos), pick(Nos,T). pick(From,Max)-> random:seed(begin (case random:seed(now()) of undefined -> NN = element(3,now()), {?RAND(NN),?RAND(NN),?RAND(NN)}; Any -> Any end) end ), T2 = random:uniform(Max), case lists:member(T2,From) of false -> pick(From,Max); true -> {T2,From -- [T2]} end.
On running it in shell:
F:\> erl Eshell V5.8.4 (abort with ^G) 1> c(shuffle). {ok,shuffle} 2> shuffle:list([a,b,c,d,e]). [c,b,a,e,d] 3> shuffle:list([a,b,c,d,e]). [e,c,b,d,a] 4> shuffle:list([a,b,c,d,e]). [a,b,c,e,d] 5> shuffle:list([a,b,c,d,e]). [b,c,a,d,e] 6> shuffle:list([a,b,c,d,e]). [c,e,d,b,a]I am motivated by the fact that in the STDLIB there is no such function. Somewhere in my game, i need to shuffle things up and also i need to find the best efficient solution to the problem, not just one that works.
Could some one help build a shorter version of the solution ? probably even more efficient ? Thank you
Please note that karl's answer is much more concise and simple.
Here's a fairly simple solution, although not necessarily the most efficient:
For example, one could probably do away with the
++
operation innth_rest/3
. You don't need to seed the random algorithm in every call torandom
. Seed it initially when you start your program, like so:random:seed(now())
. If you seed it for every call touniform/1
your results become skewed (try with[shuffle:list([1,2,3]) || _ <- lists:seq(1, 100)]
).main()-> shuffle(lists:seq(1, 10)).
Associate a random number R with each element X in L by making a list of tuples {R, X}. Sort this list and unpack the tuples to get a shuffled version of L.