English

Kippenhahn's Conjecture Revisited

Functional Analysis 2026-03-11 v1

Abstract

In 1951 paper \cite{Ki} Kippenhahn conjectured that if the characteristic polynomial \ PA(x1,x2,x3)=\mboxdet(x1A1+x2A2x3I)P_A(x_1,x_2,x_3)=\mbox{det}(x_1A_1+x_2A_2-x_3I), \ where A1A_1 and A2A_2 are n×nn\times n Hermitian matrices, has a repeated factor in the polynomial ring \C[x1,x2,x3]\C[x_1,x_2,x_3], then the pair (A1,A2)(A_1,A_2) is unitary equivalent to a direct sum (C1C2, D1D2)(C_1\oplus C_2, \ D_1\oplus D_2) where Ci,DiMni(\C)C_i, D_i\in M_{n_i}(\C) for some 1ni<n, n1+n2=n,i=1,21\leq n_i<n, \ n_1+n_2=n, i=1,2. Kippenhahn verified the conjecture whenever the degree of the minimal polynomial of x1A1+x2A2x_1A_1 + x_2A_2 is 1 or 2. In subsequent works \cite{Sh1,Sh2} Shapiro obtained a number of results which supported the conjecture. In particular, she showed that it held if n5n \leq 5. In 1983 Laffey \cite{La} showed that, in general, Kippenhahn's conjecture was not true by constructing a counterexample for n=8n=8. Since then additional counterexamples were worked out (see \cite{Wa} for example). Some positive results in this direction including the quantum version of the conjecture can be found in \cite{F1, F2, KVo1, Law}. In this paper we use methods of recently developed local spectral analysis to give some necessary and sufficient conditions for the affirmative answer to Kippenhahn's conjecture in terms of the characteristic polynomials of certain elements of the algebra generated by the matrices in the tuple.

Keywords

Cite

@article{arxiv.2603.09915,
  title  = {Kippenhahn's Conjecture Revisited},
  author = {Michael Stessin},
  journal= {arXiv preprint arXiv:2603.09915},
  year   = {2026}
}
R2 v1 2026-07-01T11:13:24.038Z