← Back to arXiv
arXivCombinatoricsarXiv:2608.10015

Kneserized Anticoncentration and Reverse Absorption for Graham's Rearrangement Conjecture

The paper tackles a problem in combinatorial number theory called Graham's rearrangement conjecture. The conjecture asks whether, given any collection of nonzero elements from a cyclic group of whole numbers modulo some integer, you can always arrange those elements in some order such that all the running partial sums are distinct and nonzero. This has been proven for prime-sized groups, but composite-sized groups are harder because their structure is more complicated. The authors extend the result to new families of composite groups by developing new mathematical tools suited to the messiness that composites introduce.

The first main tool is an anticoncentration estimate, which roughly measures how spread out a random subset sum tends to be across possible values. For prime-sized groups, such estimates work cleanly, but for composite groups a "periodic loss" appears, meaning the sums tend to cluster around subgroup-like structures rather than spreading uniformly. The authors use a classical result from additive combinatorics called Kneser's theorem to handle this clustering and still extract enough control over the distribution of subset sums to push through the ordering argument. This lets them prove the conjecture for groups whose size is a prime multiplied by any fixed integer, provided the prime is large enough.

The second main contribution is a technique called reverse absorption, which handles more general composite groups whose size is a product of several prime powers that are all roughly the same size. The key insight is a dichotomy: either the subset sums spread out nicely, in which case the anticoncentration estimates apply directly, or almost all of the elements bunch up inside a smaller subgroup or one of its cosets, giving the set a rigid structure that can be exploited differently. By iterating this dichotomy and peeling away layers of structure, the authors handle increasingly complex composite groups and prove the conjecture whenever the number of prime factors is bounded and the primes involved are all large and close to one another.

Read original →