Real numbers , , and are chosen independently and at random from the interval for some positive integer . The probability that no two of , , and are within 1 unit of each other is greater than . What is the smallest possible value of ?
- A)
- B)
- C)
- D)
- E)
Answer
D
Key insight
Order x < y < z, then shift y down 1 and z down 2; the good region becomes all ordered triples in [0, n-2], giving probability ((n-2)/n)^3.
Solution
The point is uniform in a cube of side , so the probability is the volume of the good region divided by .
By symmetry the six orderings of are equally likely, so restrict to . This slice has volume , and within it the condition is just and (the pair is then automatically at least apart).
Remove the mandatory gaps: set , , . The conditions become
which is the ordered slice of a cube of side . Shifting does not change volume, so the good ordered volume is , and
We need . Since increases with , test the choices upward: gives , and gives .
The answer is .
Why this works
"No two within 1 unit" is the continuous version of choosing integers with required gaps: sort the values, subtract the forced spacing, and the constraint vanishes. Sorting costs a factor of that cancels, and translation preserves volume, so the probability is a pure scaling ratio . The transferable move is the gap-removing substitution, which works for any number of variables and any gap size.
Alternative approach
Sanity-check the method one dimension down. For two numbers in , the shift trick predicts probability ; drawing the square and removing the strip (area ) confirms it. The three-variable region is hard to draw, which is exactly why the ordering-plus-shift argument is preferable to direct volume computation.
The trap
Treating the three pairwise conditions as independent, or extrapolating the two-variable answer ((n-1)/n)^2, which gives a formula with n-1 instead of n-2.
Common mistakes
- Treating the three pairwise conditions as independent, or extrapolating the two-variable answer ((n-1)/n)^2, which gives a formula with n-1 instead of n-2.
- Dividing only the numerator by (giving ) or only the denominator, instead of noting the ordering factor cancels.
- Stopping at because "looks close," or forgetting the problem asks for strictly greater than .
Techniques
Bound the quantity above/below or estimate to pin it down · Substitute to simplify (u = x+1/x, shifting, scaling) · Exploit symmetry to reduce work or pair up objects