ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Vector Balancing

Let X,YX,Y be norms on Rd\mathbb{R}^d. Given vectors v1,...,vn∈Rdv_1,...,v_n\in \mathbb{R}^d satisfying ∣∣vi∣∣X≤1||v_i||_X\le 1 for all ii. We want to find signs ϵi∈{−1,1}\epsilon_i \in \{-1,1\} to minimize ∣∣∑i=1nϵivi∣∣Y|| \sum_{i=1}^n\epsilon_iv_i ||_Y. Call this the discrepancy of the vectors. Define αn(X,Y,d)\alpha_n(X,Y,d) as the maximum of this discrepancy over all choice of nn vectors satisfying ∣∣vi∣∣X≤1||v_i||_X\le 1 and α(X,Y,d)=supnαn(X,Y,d)\alpha(X,Y,d) =sup_n \alpha_n(X,Y,d). Banaszczyk proved α(X,Y,d)≥d(vol(BX)vol(BY))1/d\alpha(X,Y,d) \ge \sqrt{d}(\frac{vol(B_X)}{vol(B_Y)})^{1/d} where BX,BYB_X,B_Y are the unit balls of the corresponding norms.

Is this lower bound also an upper bound for α(X,Y,d)\alpha(X,Y,d) upto log⁡d\log d factors?

Record

Added
Links

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.