Lost on an Octahedron

Alice and Bob stand on opposite vertices of a regular octahedron. At the beginning of every minute,

  • each chooses an adjacent vertex uniformly at random and moves towards it;
  • each moves at a constant rate of 1 edge per minute;
  • they will stop when they meet up.

The expected value of the number of minutes until they meet up is equal to p q \frac{p}{q} , where p p and q q are coprime positive integers.

Find the value of p + q . p + q.


Note : It is possible that they would meet at a vertex or at the midpoint of an edge.


The answer is 62.

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.

1 solution

Steven Yuan
Jun 7, 2017

Let us consider two states:

In state #1, Alice and Bob are on opposite vertices of the octahedron, i.e. they are separated by a distance of 2 edge lengths. In state #2, Alice and Bob are on adjacent vertices of the octahedron i.e. they are separated by a distance of 1 edge length.

Let T 1 T_1 denote the expected amount of time for Alice and Bob to meet each other given that they are in state #1, and define T 2 T_2 similarly for state #2. Our goal is to find the value of T 1 , T_1, since Alice and Bob start out on opposite vertices of the octahedron.

We can find that, after Alice and Bob choose a vertex, it takes 1 2 \dfrac{1}{2} minutes for them to meet up if they meet up in the middle of an edge, 1 1 minute if they meet up at a vertex, T 1 + 1 T_1 + 1 minutes if they end up in state #1, and T 2 + 1 T_2 + 1 minutes if they end up in state #2.

When Alice and Bob are in state #1, there is a 1 4 \dfrac{1}{4} probability that they meet up at a vertex; a 1 4 \dfrac{1}{4} probability that they will end up back in state #1; and a 1 2 \dfrac{1}{2} probability that they will end up in state #2. Thus,

T 1 = 1 4 ( 1 ) + 1 4 ( T 1 + 1 ) + 1 2 ( T 2 + 1 ) . T_1 = \dfrac{1}{4}(1) + \dfrac{1}{4}(T_1 + 1) + \dfrac{1}{2}(T_2 + 1).

When Alice and Bob are in state #2, there is a 1 16 \dfrac{1}{16} probability that they meet up in the middle of an edge; a 1 8 \dfrac{1}{8} probability they will meet up at a vertex; a 11 16 \dfrac{11}{16} probability that they will end up back in state #2; and a 1 8 \dfrac{1}{8} probability that they will end up in state #1. Thus,

T 2 = 1 16 ( 1 2 ) + 1 8 ( 1 ) + 11 16 ( T 2 + 1 ) + 1 8 ( T 1 + 1 ) . T_2 = \dfrac{1}{16} \left (\dfrac{1}{2} \right) + \dfrac{1}{8} (1) + \dfrac{11}{16}(T_2 + 1) + \dfrac{1}{8}(T_1 + 1).

This gives us a system of equations for T 1 T_1 and T 2 . T_2. Solving for T 1 T_1 yields T 1 = 51 11 , T_1 = \dfrac{51}{11}, so p + q = 62 . p+q = \boxed{62}.

All that time working it out and failing, I had been forgetting that meeting on the edge would account for only half a minute, tiny detail, all the errors!

Brendan Hunt - 3 years, 11 months ago

Log in to reply

made the same stupid mistake!!!!

Steven Linnell - 3 years, 11 months ago

What category do these type of problems fall into? I am asking because I had studied probability extensively in my under grad. program but I have never come across such a problem

Eppepalli Rohith - 3 years, 11 months ago

Log in to reply

Try hunting for information about Markov Chains. This problem has four states: Opposite, Adjacent, Meet at Vertex, Meet Halfway. The last two are absorbing.

Mark Hennings - 3 years, 11 months ago

Look at random walks. With discrete problems like these the problem is to work out how to get from A to B by reference to where you are at a given point in time. One example being a fly walking randomly along the vertices of a cube reaching the opposite side. Once you have defined all the possible states - in this case A and B are either opposite to or next to each other while the hypothetical fly is either 3,2, or 1 steps from the goal. But it all boils down to simultaneous equations in these types of problems.

Malcolm Rich - 3 years, 11 months ago

I have spent a literal hour on this problem.

And when I figured it out, I was screaming.

Thank you for the nice problem, it was a nice time struggling with logics. xD

Boi (보이) - 3 years, 11 months ago

Log in to reply

Haha, thanks! It's nice to hear that you were challenged by this but still managed to figure it out! :-)

Steven Yuan - 3 years, 11 months ago

There a possibility though, however small, that they will keep moving in the same direction separated by one move and never meet. Given 2 randomly acting people, I can't see how can you calculate with certainty, the number of minutes it will take until they meet. Isn't the best you can do, to approximate the probability that they will meet after X minutes?

Mattias Bardenheuer - 3 years, 11 months ago

Log in to reply

But the probability that they never meet is 0 0 . If p 1 p_1 is the probability that they never meet, given that they are in State 1, and p 2 p_2 is the probability that they never meet, given that they are in State 2, then p 1 = 1 4 p 1 + 1 2 p 2 p 2 = 11 16 p 1 + 1 8 p 2 \begin{aligned} p_1 & = \tfrac14p_1 + \tfrac12p_2 \\ p_2 & = \tfrac{11}{16}p_1 + \tfrac18p_2 \end{aligned} and hence p 1 = p 2 = 0 p_1 = p_2 = 0 .

This is just like saying that if you keep tossing a fair coin, then you are certain to get a Head sometime. The probability of an infinite number of Tails is 0 0 .

Mark Hennings - 3 years, 11 months ago

I got the correct fraction and screwed up the last step. Wondering why adding 51 and 11 is so difficult :(

Richard Desper - 3 years, 11 months ago

Very nicely written solution to an equally intriguing problem. Kudos!

Vinayak Verma - 3 years, 11 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...