Title: Refined multiplicative tensor product of matrix factorizations

Abstract: An algorithm for matrix factorization of polynomials was proposed in \cite{fomatati2022tensor} and it was shown that this algorithm produces better results than the standard method for factoring polynomials on the class of summand-reducible polynomials. In this paper, we improve this algorithm by refining the construction of one of its two main ingredients, namely the multiplicative tensor product $\widetilde{\otimes}$ of matrix factorizations to obtain another different bifunctorial operation that we call the reduced multiplicative tensor product of matrix factorizations denoted by $\overline{\otimes}$. In fact, we observe that in the algorithm for matrix factorization of polynomials developed in \cite{fomatati2022tensor}, if we replace $\widetilde{\otimes}$ by $\overline{\otimes}$, we obtain better results on the class of summand-reducible polynomials in the sense that the refined algorithm produces matrix factors which are of smaller sizes.