How many trials would you need to roll a die so that the probability of getting one of each number is greater than half?
My Solution:
We want it so that after $n$ rolls, we have at least one of each number. We can use the stars and bars method here, where we have $n$ stars and $5$ dividers.
For example, the following sequence represents scoring exactly one of each number, except the number "6", which is scored three times.
$$\star | \star | \star| \star|\star| \star\star \star$$
However, to guarantee that we always have one of each type, we need the following two conditions.
The bars cannot be together, because if they were then that means that there is a number that was not rolled.
No bars can be at the ends.
Now, there are $n$ stars, which means $n-1$ gaps excluding the end-gaps. Place the five dividers there, so $\binom{n-1}{5}$. Since this is a probability, we can divide by the total number of outcomes, which is $6^n$. Hence, the probability is the quotient, and then we can use whatever program to give approximate solutions.
Am I even close here? My gut feeling is telling me to look at a binomial distribution but again this could be way off.
There is a fairly simple recursive way to do the problem...
Let $p_i(n)$ be the probability that you have thrown exactly $i$ different values in $n$ tosses. Then $$p_i(n+1)=\frac {i}6\times p_i(n)+\frac {7-i}6\times p_{i-1}(n)$$
That expression makes it relatively easy to compute $p_6(n)$ numerically and we get $$p_6(13)=0.513858194$$
as the first value greater than $.5$
I don't immediately see a quick analytic way to solve it.