1. What Is the Collatz Conjecture?
Start with any positive whole number. Then repeat this rule again and again.
- If the number is even, divide it by 2.
- If the number is odd, multiply by 3 and add 1.
Do you always reach 1 in the end, no matter which number you start from? That question is the Collatz conjecture.
Here is what happens if you start at 6.
Every number people have tried does reach 1. But nobody has shown that this is true for every positive whole number.
2. What Was Actually Studied?
These studies do not try to solve the Collatz conjecture. They use a computer to collect many number journeys and simply observe them.
- Draw the paths that numbers take, like a map.
- Check which places make a number's features easier to see.
- Combine where a number is with how it is moving at that spot.
- Sort numbers that move in similar ways into groups.
This research group did this in several separate ways. Each study used its own data, its own rulers, and its own way of counting.
Words you will meet below
- State space
- A map that lays out the possible states of a number.
- Coordinate
- A ruler used to look at the state of a number.
- Orbit (trajectory)
- The path one number has travelled.
- Boundary
- The place where a number moves from one range into the next.
3. Change the View, and Different Features Appear
This is the most important idea on the page.
The number journeys stay the same. But which ruler you use to measure them changes what stands out. Here are some of the rulers the studies used.
| Way of looking | What it means |
|---|---|
Distance to the exit of a section(exit_distance) |
How far the number's current spot is from the exit of the section it is in. |
A gauge of how much change is still left(remaining_K) |
A scale showing roughly how much change still lies ahead. |
How many times you could divide by 2 in that one step(transition_k) |
How big a single move was. |
The shape of a number seen through remainders(residue) |
What is left over after dividing — the number's local "shape". |
The beginning part of the path(prefix) |
Only what has happened from the start up to now. |
There are other differences in viewpoint too.
- Looking at which range a number moved from, and into.
- Looking at what kind of changes came just before.
- Reading a path from the start, or lining paths up from the end.
- Using one feature alone, or combining two or more.
None of these rulers changes the numbers themselves. They only change what is easy to see.
4. What the Observations Showed
Everything below is something that was observed. None of it is something that was proved.
- Number journeys looked different near a boundary than they did deep inside a region.
- Some features could not be separated neatly using only one piece of information.
- Combining "where the number is" with "how it is moving" sometimes made sorting easier.
- Paths had places that looked like branches, and also places where similar paths came back together.
- Looking at only a short piece of a path could tell things apart to some degree, but it could not fully reproduce the whole journey.
- Features worked out from the past part of a path showed a partial relationship with the short-term future part.
How large were these studies?
These figures come from separate experiments. They must not be added together, and they should not be compared as if they came from the same experiment.
- One study examined about 70,000 records.
- Within that study, 228 special records were observed.
- A different study followed 550 number paths.
5. The Journey, Sketched Roughly
6. A Comparison: Stations and Train Lines
A number's journey is easier to picture as a train map.
| In the world of numbers | On the train map |
|---|---|
| The state of a number | A station |
| One change | Travelling to the next station |
| The whole path a number took | A train line |
| A range where numbers gather | A big station |
| A place with many branches | A transfer station |
| Returning to a similar path | Lines merging together |
In the studies, looking at "which station you are at" together with "how you boarded there" sometimes made it easier to group similar cases.
7. Important Warnings
- This research is not a proof of the Collatz conjecture.
- It did not find a counterexample to the Collatz conjecture either.
- It is an observation of a finite amount of data, gathered by computer.
- The groupings and boundaries that were seen may not apply to every whole number.
- Seeing a relationship is not the same as understanding a cause.
Each study also used different data, different rulers, and different ways of counting. Numbers from different experiments cannot simply be compared with one another.
Three Key Points
- Change the way you look, and different features of a number's movement appear.
- Combining place with movement can sometimes make sorting easier.
- This was an observation of finite data, and not a proof.