5
$\begingroup$

Imagine I have two lists of elements like the following:

listOne = RandomReal[{0,2}, {100,2}];
listTwo = RandomReal[{0,2}, {100,2}];

Let $f: X \to Y$ be a bijection or one-to-one correspondence between the elements in listOne and the elements in listTwo.

I can shuffle the elements in listOne as follows:

listOne = RandomSample[listOne, Length[listOne]];

However, is there a way for me to perform the above shuffling operation while preserving the bijection $f$? In other words, can I shuffle the elements in listOne and listTwo in precisely the same manner?

| improve this question | |
$\endgroup$
  • 1
    $\begingroup$ p.s. There's no need for Length[listone]. $\endgroup$ – Kuba♦ Oct 29 '13 at 21:07
  • $\begingroup$ @MikeHoneychurch I just saw your comment. I was a little slow on the pickup for rule replacement, but I think I get it now. $\endgroup$ – RTaylor Oct 29 '13 at 21:42
  • $\begingroup$ Personally I don't see this as a duplicate. Could one of those who closed the question show how one of the answers to the other question can be applied to this question? $\endgroup$ – Mike Honeychurch Oct 29 '13 at 21:59
  • $\begingroup$ @MikeHoneychurch To my mind, both of the highest-voted answers there apply to this question quite directly, so much so that I don't know what else to add. $\endgroup$ – Leonid Shifrin Oct 29 '13 at 22:15
  • 1
    $\begingroup$ @LeonidShifrin I agree with Mike Honeychurch; those seem to require that one list start off life ordered. [If I'm wrong, please accept Mike's humble apology..] $\endgroup$ – Daniel Lichtblau Oct 29 '13 at 23:41
9
$\begingroup$

Just create an ordering list and use Part:

listOne = {1, 2, 3};
listTwo = {4, 5, 6};

order = RandomSample[Range[Length[listOne]]]
(* {2, 3, 1} *)

listOne[[order]]
(* {2, 3, 1} *)

listTwo[[order]]
(* {5, 6, 4} *)
| improve this answer | |
$\endgroup$
  • $\begingroup$ +1 Geez I normally always choose Part over other methods but overlooked that. My morning coffee hasn't kicked in yet -- brain still frozen $\endgroup$ – Mike Honeychurch Oct 29 '13 at 21:11
  • 1
    $\begingroup$ @MikeHoneychurch, I have a time-zone advantage, my brain has been steeping in coffee all day :-) $\endgroup$ – Simon Woods Oct 29 '13 at 21:17
6
$\begingroup$

If it's a bijection you can for example do this:

s1 = RandomSample[listOne]
s2 = s1 /. MapThread[Rule, {listOne, listTwo}, 1]
| improve this answer | |
$\endgroup$
  • 3
    $\begingroup$ I would have gone with Dispatch@Thread[Rule[listOne, listTwo]]. Not sure but might be faster for large lists. $\endgroup$ – Mike Honeychurch Oct 29 '13 at 21:08
  • $\begingroup$ @Kuba Thanks it's good to know that you can do this. $\endgroup$ – RTaylor Oct 29 '13 at 21:20
  • $\begingroup$ This doesn't work if listOne contains duplicates. (Since the question is about Mathematica lists/arrays, other site users should be aware of this limitation.) $\endgroup$ – Michael E2 Oct 30 '13 at 9:58
  • $\begingroup$ @MichaelE2 OP gives the assumption that it is bijection. So there are no duplicates, or if there are they are reffering to duplicate positions in the second list. Or I missed something? $\endgroup$ – Kuba♦ Oct 30 '13 at 11:10
  • $\begingroup$ A one-to-one correspondence between two Lists, which are ordered and permit duplicates, is different than a one-to-one correspondence between sets. The site being about Mathematica, it seems fair to presume that an array is a List, not a mathematical set. Because the Q&A format is for everyone, I think the distinction ought to be made clear for those who would adapt a solution to their own use-case. If sets are meant - I agree the possibility is suggested by "bijection" - then the question should make that clear. I had a similar disagreement about another of the OP's questions. :) $\endgroup$ – Michael E2 Oct 30 '13 at 12:49
6
$\begingroup$

Although I have used Simon's method many times I think it is somewhat cleaner in this case to transpose the data, shuffle, and transpose again:

l1 = {a, b, c, d, e};
l2 = {v, w, x, y, z};

RandomSample[{l1, l2}\[Transpose]]\[Transpose]

{{e, c, b, a, d}, {z, x, w, v, y}}

The \[Transpose] symbol appears as T in the Front End:

enter image description here

| improve this answer | |
$\endgroup$
  • $\begingroup$ I agree that this is cleaner. In fact I'd normally only use the method I posted if there was some need later on to undo the shuffle. $\endgroup$ – Simon Woods Oct 31 '13 at 22:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.