Erdős's reverse Littlewood–Offord conjecture

Let x1,,xnx_1,\dotsc,x_n be unit complex numbers, and let ε1,,εn{1,+1}\varepsilon_1,\dotsc,\varepsilon_n\in\{-1,+1\}. Erdős's conjecture. The number of sums

i=1nεixi\sum_{i=1}^n \varepsilon_i x_i

with

i=1nεixi1\left|\sum_{i=1}^n \varepsilon_i x_i\right|\leq 1

is greater than c2n/nc2^n/n for some absolute constant c>0c>0. This conjecture is false as stated: taking an odd number of copies of (1,0)(1,0) and of (0,1)(0,1) makes every sum have norm at least 2\sqrt{2}, so the unit-radius claim cannot hold in general.

Sources & referencesView supporting material

Primary source

Lawrence Hollom, Julien Portier and Victor Souza, “Double-jump phase transition for the reverse Littlewood–Offord problem”, arXiv:2503.24202 (2025).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.