Generated by Codex with GPT 5.6 Sol High

Finding Primes by Eliminating Composites

Prime numbers look simple: each is divisible only by 1 and itself. Yet their locations become increasingly difficult to predict as numbers grow. Because every whole number can be factored into primes, understanding that irregular pattern is one of number theory’s central problems.

The oldest systematic search method is the Sieve of Eratosthenes. Starting with a list of numbers, it removes multiples of 2, then multiples of 3, then multiples of each successive prime. What remains is prime. Modern sieve theory extends that basic idea to much larger ranges, where mathematicians are usually not crossing off every number individually. Instead, they use information about which values have been eliminated to estimate how many primes a range must contain.

That elimination record is called Type I information. It can be extremely detailed without being enough to prove that a range contains a prime. Kevin Ford of the University of Illinois Urbana-Champaign notes that researchers can sometimes possess essentially the best Type I data available and still fail to detect any primes. Knowing a great deal about the composites does not automatically reveal what is left.

Two Kinds of Evidence, One Hard Boundary

Ford and University of Oxford mathematician James Maynard developed a method that combines Type I information with Type II information. The second category tracks overlaps among the eliminated values. Six, for example, appears both among multiples of 2 and among multiples of 3. Accounting for these repeated exclusions provides a complementary view of the same numerical range.

The two forms of evidence can be assigned different weights. By tuning those weights, mathematicians try to make the tightest possible estimate of how many primes must lie in a chosen interval. The useful analogy is a physical sieve: a mesh that is too fine catches everything, whereas one that is too coarse lets the desired objects pass through. The challenge is to identify exactly how strong the available information must be before primes can be guaranteed.

The new work does more than improve the tuning. It identifies precise limits beyond which no choice of weights can make this kind of prime count more accurate. Kaisa Matomäki of the University of Turku describes the result as specifying what counts as sufficiently good information for detecting primes. In other words, the method maps both the reach of current sieve techniques and the point where those techniques must fail.

Why a Limit Can Be Progress

A proof that a method cannot cross a particular boundary is not a defeat. It prevents mathematicians from spending effort on refinements that cannot succeed and clarifies what genuinely new ideas would be needed. Peter Sarnak of Princeton University argues that discovering what a sieve cannot achieve is fundamental to building a fuller theory of prime numbers.

The result, reported as a preprint when the article appeared, does not make prime distribution predictable or settle its major open problems. Its contribution is more structural: it shows how two important kinds of sieve information work together, establishes when they are enough, and marks their unavoidable blind spots. Ford hopes that sharper map will help researchers approach long-standing questions with a better sense of which routes remain open.