SIAM Journal on Matrix Analysis and Applications · 2013 · 10 citations · 12 references
Natural SciencesMathematicsCombinatoricsDisturbance EcologyRepresentation TheoryEigenvalue ProblemsMatrix ComputationMatrix StructureEigenvalue BoundsEigenvalues
Among hollow, symmetric $n$-by-$n$ nonnegative matrices, it is shown that any number $k$, $2\leq k\leq n-1$ of nonpositive eigenvalues is possible. However, as $n$ grows, small numbers of nonpositive eigenvalues become increasingly rare. In particular, if there is only a given finite number of distinct off-diagonal entries, the minimum number of nonpositive eigenvalues (among $n$-by-$n$ hollow, symmetric, nonnegative matrices) grows with $n$, and this remains so if just the ratio of the smallest positive off-diagonal entry to the largest is bounded away from zero. Nonetheless, every $n$-by-$n$ hollow, symmetric, nonnegative matrix with positive off-diagonal entries and with two nonpositive eigenvalues may be embedded in an $(n+1)$-by-$(n+1)$ one as a principal submatrix. This statement does not hold without the requirement of positive off-diagonal entries. Our proof recognizes some special, anti-Morishima structure of Schur complements which may be of independent interest. Relations to the independence number of a graph are also presented.
12
Frank Plumpton Ramsey · Proceedings of the London Mathematical Society · 1930 · 1.9K citations
Reducibility Among Combinatorial Problems
Kichard Rarp · 2012 · 1.4K citations
THE LAPLACIAN SPECTRUM OF GRAPHS y
Bojan Mohar · 1991 · 1.1K citations
Qualitative Problems in Matrix Theory
John Maybe, J. P. Quirk · SIAM Review · 1969 · 142 citations