Theorem 1

For all 0 < i < m-1, we have

(a) .

(b) for .

(c) and for j>0.

(d) , for , and .

Proof

To prove (d), we note that when the distance , there are two ways that the next distance can be j. The first way occurs if the dealer's secret card has face value m-1-j. This occurs with probability . The second way occurs if the dealer's secret card has face value m, thereby overshooting the player's closest secret card by one, and the face value of is j+1, making . This second event occurs with probability . Thus we have shown (d) for all j < m-1. To finish (d), note that the only way and is when both face values of and are m. This occurs with probability .

The first row of M is the absorbing state, hence (c). To prove (b) we consult Figures 1 and 2. We see that the crucial difference between and is the penultimate card to , call it star. the shaded boxes contain the only possible occurences of the other . If we imagine an arc drawn between any two adjacent , then we have two cases. The first case occurs when the star card remains vacant -- no occupy star. Then the same number of arcs which occured in the event now occur in the event. In the second case, star is occupied by some , and now the total number of arcs is 1 more than there were in the event, giving us a probability of . In summary , thus proving (b).

Note that when i>j, the dealer's secret card could have a face value of i-j, allowing . This event occurs with probability and cannot happen when , whence the different structure of the upper triangular portion of M. In fact the same reasoning as in the proof of (b) allows us to subtract this one event from those occuring in , thus subtracting a probability of at the end of the process. This proves (a).