In order to cure a certain disease, two new kinds of medicine, A and B, have been developed. We hope to know which one is more effective. Animal experiments have been carried out for this purpose. The experimental scheme is as follows: Two white mice are randomly selected in each round to compare the efficacy. For two white mice, one is randomly given medicine A and the other is given medicine B. After the results of one round of treatment are obtained, the next round of experiment will be arranged. When 4 more mice are cured with one kind of medicine than the other, the experiment is finished and that kind of medicine is considered to be more effective .
For the convenience of describing the problem, for each round of experiment:
If the mouse treated with A medicine were cured and the mouse treated with B medicine were not cured, A medicine would get 1 point and B medicine would get -1 point.
If the mouse treated with B medicine were cured and the mouse treated with A medicine were not cured, A medicine would get -1 point and B medicine would get 1 point.
If both were cured or not cured, both two kinds of medicine would get 0 point.
The cure rates of A and B medicine are represented as and respectively.
Initially, we give 4 points to both medicine A and B, and for each round of experiment, we record the points of both medicine. Let denotes the probability of "A is considered more effective than B under the condition that the current point of A is ".
If , , the value of can be expressed as , where are positive coprime integers.
What is the value of ?
Note: In the original problem, the author explicitly gave the formula for , but in fact it can be derived and I think it boring to plug the values into the formula, so I deleted it. Can you recover it?
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
Enjoyed the question
Solution:
First note that for each mouse:
P(A increases by 1) = 1 0 1 ,
P(A decreases by 1) = 1 0 4 ,
P(A stays the same) = 1 0 5 .
So, ignoring mice where the results of A and B were the same:
P(A increases by 1) = 5 1 , P(A decreases by 1) = 5 4 .
Hence:
For i = 1 , . . . , 7 , p i = 5 1 p i + 1 + 5 4 p i − 1 . Looking for a solution of the form p n = λ n , we get:
5 λ = λ 2 + 4 , so λ = 1 , 4 , and p i = A 4 i + B .
Boundary conditions:
p 0 = 0 ⟹ B = − A
p 8 = 1 ⟹ A ( 4 8 − 1 ) = 1 , so A = 4 8 − 1 1 .
p 4 = 4 8 − 1 4 4 − 1 = 4 4 + 1 1 = b a .
So b − a = 4 4 = 2 8 = 2 5 6 .
-----------------
Indeed any 1D random walk where the ratio between moving up 1 and down 1 is a : b , and p 0 = 0 , is of the form p i = A ( a i − b i ) .
So the boundary condition on the top of p N = 1 , gives us p i = a N − b N a i − b i .