1
$\begingroup$

The other day at an online bridge site I got dealt a $13$-card hand containing $3,4,5,6,7,8,9,10,J$ of spades, i.e. a $9$-card-long straight-flush (s-f). This is so rare that I decided to count how many $13$-card hands contain a s-f of length $9$ or longer.

For the purpose of this question:

  • A s-f is a consecutive sequence in the same suit, and aces always count high (since I was playing bridge!) so $A,2,3,4,5$ of spades is not a $5$-card s-f for this purpose.

  • Let us just count s-f in spades. If we want to count all suits we can simply multiply by $4$ since a hand cannot contain two s-f's of different suits each of length $\ge 9$.

I was trying to use inclusion-exclusion but got very confused. Finally I decided to count explicitly as follows. My questions:

(1) Can someone verify my answer below?

(2) Can someone show how this can be counted via inclusion-exclusion, or explain why that's not a good idea?


My attempt:

  • $13$-long s-f: No. of hands $=1$.

  • exactly $12$-long s-f: There are two such ($2-K$ or $3-A$), and each must be completed with an extra card (to make a $13$-card hand) but the extra card must not complete a $13$-long s-f. No. of hands $= 2 \times {39 \choose 1}$.

  • exactly $11$-long s-f: Three such; the $2-Q$ hand must not include the $K$ (leaving $40$ valid cards to choose $2$ to fill out the hand), the $3-K$ hand must not include $2$ or $A$ (leaving $39$ valid cards), and the $4-A$ hand must not include the $3$. No. of hands $= {40 \choose 2} + {39 \choose 2} + {40 \choose 2}$.

  • exactly $10$-long s-f: Similar to above, no. of hands $= {41 \choose 3} + {40 \choose 3} + {40 \choose 3} + {41 \choose 3}$.

  • exactly $9$-long s-f: Similar to above, no. of hands $= 2 \times {42 \choose 4} + 3 \times {41 \choose 4}$.

According to wolfram alpha total $= 571130$.

$\endgroup$
1
  • $\begingroup$ looks good to me. the +1 is for work shown, and clean exposition, regardless of whether there is an error (which I don't think there is). $\endgroup$ Commented Nov 24, 2020 at 0:19

2 Answers 2

1
$\begingroup$

Your count is correct, and here is the inclusion-exclusion approach, where the five properties correspond to the five straight flushes: \begin{align} &\binom{5}{1}\binom{52-9}{13-9} \\ &-\left[4\binom{52-10}{13-10}+3\binom{52-11}{13-11}+2\binom{52-12}{13-12}+1\binom{52-13}{13-13}\right] \\ &+\left[3\binom{52-11}{13-11}+4\binom{52-12}{13-12}+3\binom{52-13}{13-13}\right] \\ &-\left[2\binom{52-12}{13-12}+3\binom{52-13}{13-13}\right] \\ &+\binom{5}{5}\binom{52-13}{13-13} \\ &= 617050 -48461 +2623 -83 +1 \\ &=571130 \end{align}

$\endgroup$
3
  • $\begingroup$ thanks!! so in the union terms, the distances (between the s-f's) do matter and i must carefully keep track. i was thinking PIE would simply say all the 9-long s-f's minus the 10-long ones plus the 11-long ones etc, but that is not to be. your solution makes sense. hard to say if this approach is simpler or not compared to mine. $\endgroup$
    – antkam
    Commented Nov 24, 2020 at 3:39
  • $\begingroup$ wait a minute.... your entire long expression, after cancellation of terms, has only two terms left! $5 {52-9 \choose 13-9} - 4 {52 - 10 \choose 13 - 10} = 571130$...! What gives? this cant be just a coincidence, and yet i dont know how to interpret (in a combinatorial sense) these two remaining terms! $\endgroup$
    – antkam
    Commented Nov 24, 2020 at 3:41
  • $\begingroup$ I answered my own question of how to interpret the two remaining terms after cancellation. I thought this whole question was just a matter of tedious computations -- either my original way or your PIE-based answer -- but it turns out there is a simple (non-PIE-based) solution after all! $\endgroup$
    – antkam
    Commented Nov 24, 2020 at 6:13
0
$\begingroup$

Inspired by the PIE-based answer of @RobPratt I found the following much simpler (computationally speaking) solution!

As usual all s-f's considered here are in spades.

Consider the value $5 \times {52-9 \choose 13-9}$. This is the number of pairs of sets $(X,Y)$, where $|X| = 9$ and the contents form a $9$-long s-f and $|Y| = 4$ and the contents are arbitrary, as long as $X, Y$ are disjoint. The first factor in the product is $5$ because there are $5$ possible $9$-long s-f's.

Now, every hand with an exactly $9$-long s-f can be represented as $(X,Y)$ in exactly one way. Meanwhile, every hand with an exactly $10$-long s-f can be represented as $(X,Y)$ in exactly two ways, and every hand with an exactly $11$-long s-f can be represented as $(X,Y)$ in exactly three ways, etc. Therefore:

$$5 \times {52 - 9 \choose 13 - 9} = N_9 + 2 N_{10} + 3 N_{11} + 4 N_{12} + 5 N_{13}$$

where $N_j$ is the no. of hands with an exactly $j$-long s-f.

By a completely analogous argument but this time considering $(W,Z)$ where $|W| = 10$ and the contents form a $10$-long s-f, and $|Z| = 3, W \cap Z = \emptyset$, we have

$$ 4 \times {52 - 10 \choose 13 - 10} = N_{10} + 2 N_{11} + 3 N_{12} + 4 N_{13}$$

The difference between the two gives the desired count, i.e. no. of hands with $9$-long (or longer) s-f:

$$N_9 + N_{10} + N_{11} + N_{12} + N_{13} = 5 \times {52 - 9 \choose 13 - 9} - 4 \times {52 - 10 \choose 13 - 10} = 571130 \,\,\,\,\,\,\square$$

$\endgroup$
2
  • $\begingroup$ I think you have written the coefficients in reverse order. How can there be multiple ways of representing a $13$ card s-f ? $\endgroup$ Commented Nov 24, 2020 at 9:29
  • $\begingroup$ A $13$-card s-f can be represented as a pair-of-sets $(X,Y)$ by picking any of the $5$ sub-sequences as $X$ and the rest as $Y$. Since $5 \times {43 \choose 4}$ counts such pairs-of-sets, any hand with a $13$-long s-f contributes a count of $5$ towards that total, hence $5 N_{13}$. $\endgroup$
    – antkam
    Commented Nov 24, 2020 at 14:28

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .