English

Recursive eigen extrusion: Expanding eigenbasis conjecture

General Mathematics 2025-11-28 v1

Abstract

Consider nn linearly independent vectors in Cn\mathbb{C}^n which form columns of a matrix AA. The recursive evaluation of eigen directions (normalized eigenvectors) of AA is the solution of an eigenvalue problem of the form AiXi=XiΛiA_iX_i=X_i\Lambda_i with i=0,1,2i=0,1,2 \dots; and here Λi\Lambda_i is the diagonal matrix of eigenvalues and columns of XiX_i are the eigenvectors. Note that Ai+1=ϕ(Xi)A_{i+1}=\phi(X_i) where ϕ\phi normalizes all eigenvectors to unit L2\mathcal{L}_2 norm such that all diagonal elements [ϕ(X)ϕ(X)]jj=1[\phi(X)^\dagger\phi(X)]_{jj}=1. It is to be proven that for any matrix AoA_o and n7n \leq 7, the limiting set of matrices AiA_i with ii \to \infty is the set of unitary matrices U(n)U(n) with XiXiIX_i^\dagger X_i \to I. Interestingly, this problem also represents a recursive map that maximizes some average distance among a set of nn points on the unit nn-sphere. We first formally pose this conjecture, present extensive numerical results highlighting it, and prove it for special cases.

Keywords

Cite

@article{arxiv.1907.12039,
  title  = {Recursive eigen extrusion: Expanding eigenbasis conjecture},
  author = {M Hariprasad},
  journal= {arXiv preprint arXiv:1907.12039},
  year   = {2025}
}