2025-01-23 19:53:30 -08:00
|
|
|
#import "@local/handout:0.1.0": *
|
|
|
|
|
2025-01-24 17:30:38 -08:00
|
|
|
#show: handout.with(
|
2025-01-23 19:53:30 -08:00
|
|
|
title: [Somewhat Random Numbers],
|
|
|
|
by: "Mark",
|
|
|
|
)
|
|
|
|
|
|
|
|
#problem()
|
|
|
|
Alice generates 100 random numbers uniformly from $[0,1]$. \
|
|
|
|
Bob generates 101 random numbers from $[0, 1]$, but deletes the lowest result.
|
|
|
|
|
|
|
|
#v(2mm)
|
|
|
|
|
|
|
|
Say we have both of the resulting arrays, but do not know who generated each one. \
|
|
|
|
We would like to guess which of the two was generated by Bob. \
|
|
|
|
What is the optimal strategy, and what is its probability of guessing correctly?
|
|
|
|
|
|
|
|
|
|
|
|
#solution([
|
|
|
|
Looking at the mean seems like a good idea, but there's a better way: \
|
|
|
|
Assign the array with the smaller _minimum_ to Alice.
|
|
|
|
|
|
|
|
#v(3mm)
|
|
|
|
|
|
|
|
To compute the probability, generate 201 numbers. \
|
|
|
|
Assign the first 100 to Alice and the rest to Bob. \
|
|
|
|
Look at the lowest two numbers (of these 201, *before* Bob drops his lowest).
|
|
|
|
|
|
|
|
#v(8mm)
|
|
|
|
|
|
|
|
We'll use the following notation: \
|
|
|
|
`AB` means the lowest was owned by Alice, and the second-lowest, by Bob.
|
|
|
|
|
|
|
|
#v(2mm)
|
|
|
|
|
|
|
|
Probabilities are as follows: \
|
|
|
|
- `AA`: $100\/201 times 99\/200 approx 0.246$
|
|
|
|
- `AB`: $100\/201 times 101\/200 approx 0.251$
|
|
|
|
- `BA`: $101\/201 times 100\/200 approx 0.251$ // spell:disable-line
|
|
|
|
- `BB`: $101\/201 times 100\/200 approx 0.251$
|
|
|
|
|
|
|
|
#v(4mm)
|
|
|
|
Now, Bob drops his lowest number. \
|
|
|
|
We'll cross out the number he drops and box the new lowest number (i.e, the one we observe):
|
|
|
|
- #{
|
|
|
|
(
|
|
|
|
box(`A`, stroke: ored, inset: 1pt)
|
|
|
|
+ box(`A`, inset: 1pt)
|
|
|
|
+ box([: $approx 0.246$], inset: (top: 1pt, bottom: 1pt))
|
|
|
|
)
|
|
|
|
}
|
|
|
|
- #{
|
|
|
|
(
|
|
|
|
box(`A`, stroke: ored, inset: 1pt)
|
|
|
|
+ box(strike(`B`), inset: 1pt)
|
|
|
|
+ box([: $approx 0.251$], inset: (top: 1pt, bottom: 1pt))
|
|
|
|
)
|
|
|
|
}
|
|
|
|
- #{
|
|
|
|
(
|
|
|
|
box(strike(`B`), inset: 1pt)
|
|
|
|
+ box(`A`, stroke: ored, inset: 1pt)
|
|
|
|
+ box([: $approx 0.251$], inset: (top: 1pt, bottom: 1pt))
|
|
|
|
)
|
|
|
|
}
|
|
|
|
- #{
|
|
|
|
(
|
|
|
|
box(strike(`B`), inset: 1pt)
|
|
|
|
+ box(`B`, stroke: ored, inset: 1pt)
|
|
|
|
+ box([: $approx 0.251$], inset: (top: 1pt, bottom: 1pt))
|
|
|
|
)
|
|
|
|
}
|
|
|
|
#v(8mm)
|
|
|
|
Alice has the smallest number in 3 of 4 cases, which have a total probability of $approx 0.749$.
|
|
|
|
])
|