Each of the sides and the diagonals of a regular pentagon are randomly and independently colored red or blue with equal probability. What is the probability that there will be a triangle whose vertices are among the vertices of the pentagon such that all of its sides have the same color?
- A)
- B)
- C)
- D)
- E)
Answer
D
Key insight
With no one-color triangle, every corner has 2 red and 2 blue lines, forcing the red lines into a five-sided loop: only 12 of 1024.
Solution
There are segments, each red or blue, so colorings, all equally likely. "At least one one-color triangle" is very common, so count the rare opposite: colorings with no one-color triangle.
1. No corner can have three segments of the same color. Each corner has segments. Suppose corner had three red segments, going to , , .
- If were red, triangle would be all red. So is blue.
- The same way, and are blue.
- But then triangle is all blue.
Either way there is a one-color triangle. The same happens with three blue. So in a good coloring every corner has exactly red and blue segments.
2. What do the red segments look like? Every corner touches exactly red segments. Start at a corner and follow red segments: you enter each corner on one red segment and leave on the other, so you trace a closed loop. A loop needs at least corners, and corners cannot be split into two loops of or more. So the red segments form one loop through all corners.
3. Does every such loop work? The blue segments are then the other segments, which also form a loop through all corners. A five-corner loop contains no triangle. So every red loop gives a good coloring.
4. How many five-corner loops are there? Line up the corners in a loop order: start at corner and arrange the other after it, ways. Each loop was counted twice (once going each direction), so there are loops.
5. Probability.
The answer is .
Why this works
A small rule at each corner (no three segments of one color) turned out to fix the whole picture: two interlocking five-sided loops, one red and one blue. When "at least one" is almost certain, it is usually much easier to count the rare cases where it fails and subtract from .
The trap
Trying to count the colorings that do have a one-color triangle directly, with overlaps among the 10 triangles, which is long and easy to get wrong.
Concise solution
There are colorings of the ten segments. Count the colorings with no single-color triangle.
Fix a vertex ; four segments leave it. If three of them, say to , were red, then to avoid red triangles , , the segments would all be blue, forming a blue triangle. So no vertex has three edges of one color: every vertex has exactly red and blue edges.
The red edges then form a graph on vertices in which every vertex has degree . Such a graph is a union of cycles covering all vertices; since a cycle needs at least vertices, cannot be split, so the red graph is a single -cycle. Its complement (the blue edges) is then also a -cycle, and a -cycle contains no triangle, so every such coloring works.
The number of -cycles on labeled vertices is (arrange the vertices around a cycle, up to rotation and reflection).
Therefore the probability of at least one monochromatic triangle is
The answer is .
Review in the AoPS books
- Introduction to Counting & Probability, Chapter 7: Introduction to Probability
- Introduction to Counting & Probability, Chapter 2: Basic Counting Techniques (§2.5)
- Intermediate Counting & Probability, Chapter 5: The Pigeonhole Principle
- Introduction to Counting & Probability, Chapter 8: Basic Probability Techniques
Techniques
Count the complement and subtract from the total · Pigeonhole principle