A frog sitting at the point begins a sequence of jumps, where each jump is parallel to one of the coordinate axes and has length , and the direction of each jump (up, down, right, or left) is chosen independently at random. The sequence ends when the frog reaches a side of the square with vertices and . What is the probability that the sequence of jumps ends on a vertical side of the square?
- A)
- B)
- C)
- D)
- E)
Answer
B
Key insight
Use symmetry to collapse the nine interior points to three unknowns (center is 1/2 by symmetry), then solve a small linear system for the start point.
Solution
Let be the probability of ending on a vertical side ( or ) from the interior point . By the square's symmetries, (vertical and horizontal look the same from the center), , and .
Set , , . Conditioning on the first jump:
- From : left hits (win), right goes to , up or down goes to a point with value :
- From : left wins, down hits (lose), up goes to , right to :
- From : down loses, up goes to , left or right goes to a point with value :
Substitute into the second equation: , so , i.e. .
Then , so and .
The answer is .
Why this works
A random walk that stops at a boundary is a state machine: the probability from each interior point is the average of the probabilities of its four neighbors, with boundary points fixed at or . Symmetry is what makes it tractable: nine unknowns become three, and the center's value is free.
The trap
Assuming the frog is equally likely to hit either kind of side because the start is one step from the left edge and two from the top and bottom, or setting up states without using symmetry and drowning in equations.
Common mistakes
- Assuming the frog is equally likely to hit either kind of side because the start is one step from the left edge and two from the top and bottom, or setting up states without using symmetry and drowning in equations.
- Forgetting that the frog can wander back toward the start; a one- or two-step tree analysis undercounts and gives a value below the true .
Techniques
Define states/recurrence and iterate · Exploit symmetry to reduce work or pair up objects