0
$\begingroup$

I need to generically define a strict preference profile with Condorcet cicles when the number of players and alternatives coincide. To illustrate my problem, consider the following four-player & four-alternative example.

Let $A=\{a_1,a_2,a_3,a_4\}$ be a set of four alternatives, let $N=\{1,2,3,4\}$ be a set of four players and let $P=(P_i)_{i\in N}$ be a preference profile (i.e., linear order profile) on alternatives such that \begin{gather} a_1P_1a_2P_1a_3P_1a_4\\ a_2P_2a_3P_2a_4P_2a_1\\ a_3P_3a_4P_3a_1P_3a_2\\ a_4P_4a_1P_4a_2P_4a_3 \end{gather}

In the preference profile above, the top-ranked alternative of player $i$ becomes the bottom-ranked alternative of player $i+1$, and so on.

In order to generalise the strict preference profile above to an arbitrary (but equal) number of players and alternatives, I had thought of writing the following: let $P=(P_i)_{i\in N}$ satisfy, for every player $i\in N$, \begin{gather} a_iP_ia_{i+1}P_i\dots P_ia_{n+i-1} \end{gather} Then, for player $i=1$, we obtain $a_i=a_1$, $a_{i+1}=a_2$, $a_{i+2}=a_3$ and $a_{n+1-i}=a_{4+1-1}=a_4$; and thus, \begin{gather} a_1P_1a_2P_1a_3P_1a_4 \end{gather} Hence, for player $i=1$, this approach works.

However, for player $i=2$, we obtain $a_i=a_2$, $a_{i+1}=a_3$, $a_{i+2}=a_4$ and $a_{n+1-i}=a_{4+1-2}=a_3$; and thus, \begin{gather} a_2P_2a_3P_2a_4P_2a_3 \end{gather} Hence, for player $i=2$ (and by a similar token, for all players $i\neq 1$), my approach does not work.

Hence, how to formally define the preference profile in which the top alternative of one player becomes the bottom alternative of the next player?

EDIT A friend of mine suggested using modular arithmetic to do what I need to do; and while I think he’s right, I don’t see how to proceed.

$\endgroup$

2 Answers 2

2
$\begingroup$

We have $a_i P_j a_k$ when $i-(j-1)\mod n<k-(j-1)\mod n$. Here, we represent $0$ by $n$.

$\endgroup$
7
  • $\begingroup$ Thank you very much for your helpful answer. However, could you please add some context that helps me understand it? Also, is the modulus $N$ or $N+1$? I personally came with $a_iP_ia_{(i+1)\bmod n}P_ia_{(i+2)\bmod n}P_i\cdots P_ia_{(i+n-1)\bmod n}$, but my and your expressions seem to be quite different. Could you please point out the error in mine? $\endgroup$
    – EoDmnFOr3q
    Commented Aug 11 at 6:33
  • $\begingroup$ Also, for $n=4$ and $i=2$, I obtain $(1-(i-1)) \bmod (n+1)=0<3=(4-(i-1)) \bmod (n+1)$; and thus, $a_1P_ia_4$. But instead, we should have $a_4 P_i a_1$ $\endgroup$
    – EoDmnFOr3q
    Commented Aug 11 at 8:23
  • 1
    $\begingroup$ You are right, it should be mod $N$. Then $i-(j-1)$ mod $N$ is the rank of alternative $j$ for $i$ when there are $N$ alternatives, and a lower rank is better. And you need to interpret $0$ as $N$. $\endgroup$ Commented Aug 11 at 12:18
  • 1
    $\begingroup$ That is just the ordering used $1<2<3<...<0$. $\endgroup$ Commented Aug 11 at 12:39
  • 1
    $\begingroup$ Given that the labels are $1,2,3,\ldots,n$ instead of $0,1,2,\ldots,n-1$, that is the natural thing. Most people I know with an analog watch have 12 at the top, not 0. $\endgroup$ Commented Aug 11 at 15:48
3
$\begingroup$

For $1\le i\le n$ let $a_{n+i}:= a_i$. For $i\in N$ define $P_i$ by $a_iP_ia_{i+1}P_i\cdots P_ia_{i+n-1}$.

$\endgroup$
2
  • 1
    $\begingroup$ Thank you very for your answer. However, shouldn’t it be “For $1<i \leq n$, let $a_{n+i}=a_i$” instead of “For $1\leq i \leq n-1$, let $a_{n+i}=a_i$”? $\endgroup$
    – EoDmnFOr3q
    Commented Nov 2 at 7:42
  • 1
    $\begingroup$ @EoDmnFOr3q, you're right, I edited the answer. $\endgroup$
    – VARulle
    Commented Nov 3 at 8:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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