we give a new theoretical tool to solve sparse system with finitely many solutions. It is based on toric varieties and basic linear algebra; eigenvalues,eigenvectorsand coefficient matrices. We adapt Eigenvalue theorem and Eigenvector theorem to work with a canonical rectangular matrix (the first Koszul map) and prove that these new theorems serve to solve overdetermined sparse systems and to count the expected number of solutions.