Towards Randomized Testing of $q$-Monomials in Multivariate Polynomials (1302.5898v3)

Abstract: Given any fixed integer $q\ge 2$, a $q$-monomial is of the format $\displaystyle x^{s_1}{i_1}x^{s_2}{i_2}...x_{i_t}^{s_t}$ such that $1\le s_j \le q-1$, $1\le j \le t$. $q$-monomials are natural generalizations of multilinear monomials. Recent research on testing multilinear monomials and $q$-monomails for prime $q$ in multivariate polynomials relies on the property that $Z_q$ is a field when $q\ge 2 $ is prime. When $q>2$ is not prime, it remains open whether the problem of testing $q$-monomials can be solved in some compatible complexity. In this paper, we present a randomized $O^*(7.15^k)$ algorithm for testing $q$-monomials of degree $k$ that are found in a multivariate polynomial that is represented by a tree-like circuit with a polynomial size, thus giving a positive, affirming answer to the above question. Our algorithm works regardless of the primality of $q$ and improves upon the time complexity of the previously known algorithm for testing $q$-monomials for prime $q>7$.