show/hide this revision's text 5 Fixed mathematical typo

Two particles start out at random positions on a unit-circumference circle. Each has a random speed (distance per unit time) moving counterclockwise uniformly distributed within $[0,1]$. How long until they occupy the same position? In the example below, the red particle catches the green particle at $t=5.9$, i.e., nearly six times around the circle:
           Two Particles
The distribution of overtake-times is quite skewed, indicating perhaps the mean could be $\infty$. For example, in one simulation run, it took more than $3$ million times around the circle before one particle finally caught the other. So I don't trust the means I am seeing (about $25$).

What is the distribution of overtake-times?

I was initially studying $n$ particles on a circle, but $n=2$ seems already somewhat interesting...

Update (2Dec12). Alexandre Eremenko concisely established that the expected overtake-time (the mean) is indeed $\infty$. But I wonder what is the median, or the mode? Simulations suggest the median is about $1.58$ and the mode of rounded overtake-times is $1$, reflecting a distribution highly skewed toward rapid overtake. (The median is suspiciously close to $\pi/2$ ...)

Update (3Dec12). Fully answered now with Vaughn Climenhaga's derivation of the distribution, which shows that the median is $\frac 1 + \frac 1{\sqrt{3}} \approx 1.577$.

show/hide this revision's text 4 added 173 characters in body

Two particles start out at random positions on a unit-circumference circle. Each has a random speed (distance per unit time) moving counterclockwise uniformly distributed within $[0,1]$. How long until they occupy the same position? In the example below, the red particle catches the green particle at $t=5.9$, i.e., nearly six times around the circle:
           Two Particles
The distribution of overtake-times is quite skewed, indicating perhaps the mean could be $\infty$. For example, in one simulation run, it took more than $3$ million times around the circle before one particle finally caught the other. So I don't trust the means I am seeing (about $25$).

What is the distribution of overtake-times?

I was initially studying $n$ particles on a circle, but $n=2$ seems already somewhat interesting...

Update (2Dec12). Alexandre Eremenko concisely established that the expected overtake-time (the mean) is indeed $\infty$. But I wonder what is the median, or the mode? Simulations suggest the median is about $1.58$ and the mode of rounded overtake-times is $1$, reflecting a distribution highly skewed toward rapid overtake. (The median is suspiciously close to $\pi/2$ ...)

Update (3Dec12). Fully answered now with Vaughn Climenhaga's derivation of the distribution, which shows that the median is $\frac 1{\sqrt{3}} \approx 1.577$.

show/hide this revision's text 3 Update on Alexandre's mean, and questioning median & mode.

Two particles start out at random positions on a unit-circumference circle. Each has a random speed (distance per unit time) moving counterclockwise uniformly distributed within $[0,1]$. How long until they occupy the same position? In the example below, the red particle catches the green particle at $t=5.9$, i.e., nearly six times around the circle:
           Two Particles
The distribution of overtake-times is quite skewed, indicating perhaps the mean could be $\infty$. For example, in one simulation run, it took more than $3$ million times around the circle before one particle finally caught the other. So I don't trust the means I am seeing (about $25$).

What is the distribution of overtake-times?

I was initially studying $n$ particles on a circle, but $n=2$ seems already somewhat interesting...

Update (2Dec12). Alexandre Eremenko concisely established that the expected overtake-time (the mean) is indeed $\infty$. But I wonder what is the median, or the mode? Simulations suggest the median is about $1.58$ and the mode of rounded overtake-times is $1$, reflecting a distribution highly skewed toward rapid overtake. (The median is suspiciously close to $\pi/2$ ...)

show/hide this revision's text 2 added 49 characters in body
show/hide this revision's text 1