2

In how many ways are we able to arrange $k$ identical non-overlapping dominoes on a circle of $2n$ labelled vertices?


The problem can be reduced to the number of ways to choose $k$ non-consecutive vertices from the $2n$ labelled vertices,and now there are two cases,assuming we are arranging these $k$ identical dominoes counterclockwise:

  • If vertex $1$ in included,then the adjacent vertex (from the left) cannot be chosen,since the dominoes are non-overlopping,so we are left with the other $2n-2$ vertices and we want to choose $k-1$ nonconsecutive vertices ,this can be done in $\binom{2n-2-(k-1)+1}{k-1}=\binom{2n-k}{k-1}$ ways.

  • If vertex $1$ in not included,so we are left with the other $2n-1$ vertices and we want to choose $k$ nonconsecutive vertices ,this can be done in $\binom{2n-1-k+1}{k}=\binom{2n-k}{k}$ ways.

Now summing these two cases gives the answer.


I'm not sure about the proof,besides dos it make difference if we do the process clockwise?

2 Answers2

3

You appear to be off a bit: in your first case $3$ vertices are unavailable, not $2$.

I’ve numbered the vertices from $1$ through $2n$. For my first case I put a domino on vertices $1$ and $2$. Now I need to choose $k-1$ of the $2n-3$ vertices $3,4,\ldots,2n-1$, ensuring that no two chosen vertices are adjacent. This can be done in

$$\binom{(2n-3)-(k-2)}{k-1}=\binom{2n-1-k}{k-1}$$

ways.

For my second case I put a domino on vertices $2n$ and $1$; the analysis is the same, so we get another $\binom{2n-1-k}{k-1}$ arrangements.

Any other arrangement must avoid vertex $1$ entirely. In that case we need to choose $k$ of the $2n-2$ vertices $2,3,\ldots,2n-1$, ensuring that no two chosen vertices are adjacent. This can be done in

$$\binom{(2n-2)-(k-1)}k=\binom{2n-1-k}k$$

ways. I get a total of

$$2\binom{2n-1-k}{k-1}+\binom{2n-1-k}k=\binom{2n-1-k}{k-1}+\binom{2n-k}k$$

arrangements. I’ve checked this by hand with $n=4$ and $k=3$.

Brian M. Scott
  • 616,228
1

Let's use $d_k$ to denote the number non-overlapping arrangements of $k$ dominoes on a circle of $2n$ vertices.

Your calculation contains an error. If vertex $1$ is included, then three vertices must be excluded, not two. Since the domino will cover vertices $1$ and $2$, neither of these may be included. In addition, as you point out, vertex $2n$ must be excluded. So the correct result is $$ d_k=\binom{2n-k-1}{k-1}+\binom{2n-k}{k}. $$

Here's an alternative method. Let's place a token on each vertex that is not covered by a domino. So every vertex is now covered either by a token or by a domino and the number of dominoes and tokens taken together is $k+2(n-k)=2n-k$.

Let's now make a special mark on one object, either domino or token. Since there are $2n-k$ objects we could choose to mark, the number of marked arrangements is $(2n-k)d_k$.

Marked arrangements can be counted in a different way, however, and equating the two answers will allow us to find $d_k$. This second method is to lay the $k$ dominoes and $2(n-k)$ tokens out in a line rather than a circle. There are $\binom{2n-k}{k}$ ways to do this. The first object in the line will be the marked one. Now we wrap the arrangement on the circle. Let the vertices be numbered $1$ to $2n$. We just need to pick one of these vertices to be the starting point. So there are $2n\binom{2n-k}{k}$ marked arrangements.

Equating this to our previous answer shows that $$ d_k=\frac{2n}{2n-k}\binom{2n-k}{k}. $$ Since $\binom{2n-k-1}{k-1}=\frac{k}{2n-k}\binom{2n-k}{k}$, the two methods agree.

Will Orrick
  • 18,220
  • Or I guess we can say "Assume the dominoes are constructed of two glued equal sized tiles, since the dominoes are non-overlapping,hence there does not exist any two dominoes with their tiles overlapped,which means there is at least $1$ unchosen vertex between each neighbouring pair (between the last tiles of every two consecutive dominoes there is at least $1$ vertex),so we are looking for the number of arrangements of $k-1$ identical dominoes with their tiles non-overlapped on a circle of $2n$ labelled vertices,hence we remove the vertices before and after the vertex $1$ ...." –  May 18 '20 at 17:06
  • Yes, you can think of dominoes as two tiles glued together and tokens as a single tile. I'm not sure I understand the last part of your comment. – Will Orrick May 18 '20 at 17:38
  • Wish I could show that by a picture,then everything would be clear,however I tried that on a paper and now I understand the process,just a few questions,$\color{red}{1)}$Do we divide the cases into two parts just because it lets us to use the formula $\binom{n-k+1}{k}$? (it looks that this trick is a good way which makes the problem easier to deal with,since we end up with a simpler and more famous problem which asks for the number of ways to choose $k$ non-consecutive objects from an $n$-element set. –  May 18 '20 at 17:49
  • $\color{red}{2)}$The problem is related to relaxed ménage problem,what will happen if the dominoes were labelled ?I think then the answer would transfer to:$$ \frac{2n}{2n-k}\binom{2n-k}{k}. k!$$ –  May 18 '20 at 17:53
  • I agree with your formula for the number of configurations with labeled dominoes. In the problem of choosing $k$ non-consecutive objects from a set of size $n$, you are not working on a circle. That is, if item $n$ and item $1$ were both chosen, that would not be considered a violation of non-consecutivity. But on the circle you do need to consider that case to be invalid. The case analysis is a way handling this issue. The marking procedure in my method is an alternative way of dealing with the same issue. The "seam" in circle problems is usually problematic and must be dealt with somehow. – Will Orrick May 18 '20 at 18:05
  • you are right,and after choosing the vertex $1$ we are left with $2n-3$ vertices, and we need to choose $k-1$ non-consecutive vertices of them,which is the same as what we do in the linear fashion. –  May 18 '20 at 18:25