Table of Contents
Fetching ...

Brualdi-Goldwasser-Michael problem for maximum permanents of {\rm(0,1)}-matrices

Tingzeng Wu, Xiangshuai Dong, Huazhong Lü

TL;DR

The paper resolves the Brualdi-Goldwasser-Michael problem for maximum permanents of $n\times n$ (0,1)-matrices with a fixed number of zeros by deriving a closed-form upper bound $\\mu(n,\\tau)$ and characterizing extremal configurations. Using a factorial-product maximization argument, it proves that the maximum is achieved by block-structured matrices $K_{n,\\sigma}$ when $\sigma-kn\\equiv 0 \pmod{k+1}$ and $(k+1)n-\\sigma\equiv 0 \pmod{k}$, with the explicit value $\\mu(n,\\tau)=((k+1)!)^{(\\sigma-kn)/(k+1)}(k!)^{((k+1)n-\\sigma)/k}$. For the range $2n+1 \le\\sigma \le 3n$ (equivalently $n^{2}-3n \le\\tau \le n^{2}-2n-1$) a detailed inductive classification identifies the extremal matrices and provides the exact permanents, involving specific families of block-structured matrices determined by $\\sigma-2n$ mod 3 and parity of $3n-\\sigma$. Maple-assisted verification anchors the base case $n=8$, after which the induction yields a complete description in this regime. The results illuminate the interplay between combinatorial matrix structure and permanent optimization and advance the understanding of permanents near the high-density region of ones.

Abstract

Let $\mathscr{U}(n,τ)$ be the set of all {\rm(0,1)}-matrices of order $n$ with exactly $τ$ 0's. Brualdi et al. investigated the maximum permanents of all matrices in $\mathscr{U}(n,τ)$(R.A. Brualdi, J.L. Goldwasser, T.S. Michael, Maximum permanents of matrices of zeros and ones, J. Combin. Theory Ser. A 47 (1988) 207--245.). And they put forward an open problem to characterize the maximum permanents among all matrices in $\mathscr{U}(n,τ)$. In this paper, we focus on the problem. And we characterize the maximum permanents of all matrices in $\mathscr{U}(n,τ)$ when $n^{2}-3n\leqτ\leq n^{2}-2n-1$. Furthermore, we also prove the maximum permanents of all matrices in $\mathscr{U}(n,τ)$ when $σ-kn\equiv0 (mod~k+1)$ and $(k+1)n-σ\equiv0(mod~k)$, where $σ=n^{2}-τ$, $kn\leqσ\leq (k+1)n$ and $k$ is integer.

Brualdi-Goldwasser-Michael problem for maximum permanents of {\rm(0,1)}-matrices

TL;DR

The paper resolves the Brualdi-Goldwasser-Michael problem for maximum permanents of (0,1)-matrices with a fixed number of zeros by deriving a closed-form upper bound and characterizing extremal configurations. Using a factorial-product maximization argument, it proves that the maximum is achieved by block-structured matrices when and , with the explicit value . For the range (equivalently ) a detailed inductive classification identifies the extremal matrices and provides the exact permanents, involving specific families of block-structured matrices determined by mod 3 and parity of . Maple-assisted verification anchors the base case , after which the induction yields a complete description in this regime. The results illuminate the interplay between combinatorial matrix structure and permanent optimization and advance the understanding of permanents near the high-density region of ones.

Abstract

Let be the set of all {\rm(0,1)}-matrices of order with exactly 0's. Brualdi et al. investigated the maximum permanents of all matrices in (R.A. Brualdi, J.L. Goldwasser, T.S. Michael, Maximum permanents of matrices of zeros and ones, J. Combin. Theory Ser. A 47 (1988) 207--245.). And they put forward an open problem to characterize the maximum permanents among all matrices in . In this paper, we focus on the problem. And we characterize the maximum permanents of all matrices in when . Furthermore, we also prove the maximum permanents of all matrices in when and , where , and is integer.

Paper Structure

This paper contains 8 sections, 14 theorems, 72 equations, 2 tables.

Key Result

Theorem 1.4

(bru2) Let $n$ and $\tau$ be integers with $n\geq3$ and $n^{2}-2n\leq\tau\leq n^{2}-n$. Then $u(n,\tau)= 2^{\lfloor\frac{\sigma-n}{2} \rfloor}$. Moreover, for $A\in\mathscr{U}(n,\tau)$, we have ${\rm per}A=\mu(n,\tau)$ if and only if $A$ is combinatorially equivalent to one of the following matrices

Theorems & Definitions (28)

  • Conjecture 1.1
  • Conjecture 1.2
  • Theorem 1.4
  • Theorem 1.5
  • Theorem 1.6
  • Theorem 1.7
  • Theorem 1.8
  • Theorem 1.9
  • Lemma 2.1
  • Claim 1
  • ...and 18 more