Consider this sequence of numbers: 5, 7, 9. Can you spot the pattern? Here’s another with the same pattern: 15, 19, 23. One more: 232, 235, 238.

“Three equally spaced things,” says Raghu Meka, a computer scientist at UCLA. “That’s probably the simplest pattern you can imagine.”

Yet for almost a century, mathematicians in the field of combinatorics have been puzzling out how to know whether an endless list of numbers contains such a sequence, called an arithmetic progression. In other words, is there a way to be mathematically certain that a set contains a sequence of three or more evenly spaced numbers, even if you don’t know much about how the numbers in the set were selected or what the progression might be?

Progress on the question has been slow, even plodding. But last year, Meka and Zander Kelley, a Ph.D. computer science student at the University of Illinois Urbana-Champaign, surprised mathematicians by making an exponential leap. The researchers are outsiders in combinatorics, which is concerned with counting configurations of numbers, points or other mathematical objects. And the duo didn’t set out to tackle the mystery of arithmetic progressions.

Kelley and Meka were instead investigating abstract games in computer science. The pair sought a mathematical tool that might help them understand the best way to win a particular type of game over and over again. “I’m super-interested in a collection of techniques that fall under this umbrella called structure versus randomness,” Kelley says. Some of the earliest progress on arithmetic progressions relied on such techniques, which is what led Kelley and Meka to dive into the topic.

The mystery of whether arithmetic progressions will show up is just one of many mathematical questions related to order versus disorder in sets of objects. Understanding order — and when and where patterns must emerge — is a recurring theme in many branches of math and computer science.

Another example of order in objects says that any group of six people must contain either a group of at least three mutual acquaintances (all three know each other) or a group of at least three complete strangers (no one knows another). Research has shown that it doesn’t matter who they are, where they are from or how they were selected. There’s something powerful, maybe almost spooky, about the fact that we can say this — and make other similar claims about structure in sets — with mathematical certainty.

Solving the mystery of arithmetic progressions might open doors to investigating more complex relationships among numbers in a set — gaps that change in more elaborate ways, for instance. “These are more sophisticated versions of the same theorems,” says Bryna Kra, a mathematician at Northwestern University in Evanston, Ill. “Typically, once you see arithmetic progressions … you see other patterns.”

After publishing their work on arithmetic progressions, Kelley and Meka, with Shachar Lovett of the University of California, San Diego, imported techniques from their investigations of arithmetic progressions into a different context. The researchers solved a question in communication complexity, a subfield of theoretical computer science concerned with transmitting data efficiently between parties who have only partial information.

What’s more, knowing that certain mathematical structures have to appear in certain situations can be useful in real-world communication networks and for image compression.

Potential applications aside, researchers who study arithmetic progressions — or other facets of purely theoretical mathematics — are often motivated more by sheer curiosity than any practical payoff. The fact that questions about such simple patterns and when they appear remain largely unanswered is, for many, reason enough to pursue them.

## What are arithmetic progressions?

Let’s take a moment to get our hands on some sets of numbers and the arithmetic progressions those sets contain, starting with the prime numbers, perennial favorites of math enthusiasts. A prime number is any whole number divisible only by itself and by 1; the first 10 primes are 2, 3, 5, 7, 11, 13, 17, 19, 23 and 29. Within those numbers, we can find a few arithmetic progressions. The numbers 3, 5 and 7 form a three-term arithmetic progression with a gap of two. But the numbers in a progression don’t have to follow each other immediately within the larger set: The numbers 5, 11, 17, 23 and 29 form a five-term arithmetic progression with a gap of six.