Computing tensor rank decomposition is currently a quite hot topic in the numerical linear algebra community and involve solving systems of polynomial equations. In short, the tensor rank decomposition can be seen as an SVD decomposition but for "matrices" with more than 2 indices.