The almost semimonotone matrices (Q2302882)

A real square matrix \(A\) is called semimonotone if for each nonzero nonnegative vector \(x\), there exists an index \(k\) such that \(x_k>0\) and \((Ax)_k\) is nonnegative. \(A\) is called strictly semimonotone if for each nonzero nonnegative vector \(x\), there exists an index \(k\) such that both \(x_k\) and \((Ax)_k\) are positive. One of the reasons that this matrix class is important comes from the well-known result that \(A\) is (strictly) semimonotone if and only if the linear complementarity problem corresponding to \(A\) has a unique solution for all input vectors (\(q \geq 0) ~q >0\). A matrix \(A\) is called semipositive if there is a positive vector \(x\) such that \(Ax\) is positive; \(A\) is called weakly semipositive if there exists a nonzero nonnegative vector \(x\) such that \(Ax\) is nonnegative. The two matrix classes defined above, are related by the fact that, a matrix \(A\) is (strictly) semimonotone if and only if \(A\) and all its principal submatrices are (semipositive) weakly semipositive. An almost (strictly) semimonotone matrix \(A\) is a matrix which is not (strictly) semimonotone but whose proper principal submatrices are (strictly) semimonotone. It is known that a matrix \(A\) is almost (strictly) semimonotone if all the proper principal submatrices of \(A\) are (strictly) semimonotone and there exists a positive vector \(x\) such that the vector \(Ax\) is negative (\(Ax\) is nonpositive). The author proves some results on the classes of almost (strictly) semimonotone matrices of orders \(2 \times 2\) and \(3 \times 3\) and presents a characterization. Some interesting results for general almost semimonotone matrix are also shown. Here is an example: if \(A\) is almost semimonotone, then \(A\) has a simple negative eigenvalue, which is the smallest eigenvalue of \(A\) in absolute value, associated with a positive eigenvector. Further, there are no other nonnegative eigenvectors.