A new approach to the Lvov-Kaplansky conjecture through gradings (Q6173930)

The Lvov-Kaplansky conjecture states that if \(f(x_1,\ldots,x_m)\) is a multilinear polynomial in the free associative algebra \(K\langle X\rangle\) over a field \(K\), then its image on the \(n\times n\) matrix algebra \(M_n(K)\) is \(\{0\}\), \(K\), \(sl_n(K)\) or \(M_n(K)\). The answer to the conjecture is known only for \(m = 2\) or \(n = 2\) (under some restrictions on the base field \(K\)), with partial results for \(n = 3\) and \(m = 3\). A survey on the topic can be found in [\textit{A. Kanel-Belov} et al., SIGMA, Symmetry Integrability Geom. Methods Appl. 16, Paper 071, 61 p. (2020; Zbl 1459.16012)]. A weaker version of the Lvov-Kaplansky conjecture is the Mesyan conjecture: If the multilinear polynomial \(f(x_1,\ldots,x_m)\) is nonzero on \(M_n(K)\), \(m\geq 1\), \(n\geq 2\) and \(m\leq 2n - 1\), then the image of \(f\) on \(M_n(K)\) contains \(sl_n(K)\). \par In the present paper the authors assume that \(M_n(K)\) is endowed with the elementary \({\mathbb Z}_n\)-grading defined by the property that the matrix units \(E_{ij}\) are of degree \(j-i\) (mod \(n\)). But the approach in the paper is different to the usual approach of studying gradings of algebras and their graded polynomial identities. The authors use gradings and images of multilinear polynomials to obtain results about ordinary polynomial identities and central polynomials and to present an equivalent statement to the Lvov-Kaplansky conjecture. \par The first result of the paper gives that the multilinear polynomial \(f(x_1,\ldots,x_m)\) is a polynomial identity for \(M_n(K)\) if and only if \(f\) satisfies the condition (S0) If \(a_1,\ldots,a_m \in M_n(K)\) are homogeneous matrices satisfying \(\displaystyle \sum_{i=1}^m\deg(a_i) = 0\) then \(f(a_1,\dots,a_m) = 0\). Then the authors consider the following conditions on the multilinear polynomial \(f(x_1,\ldots,x_m)\): (S1) If \(a_1,\ldots,a_m \in M_n(K)\) are homogeneous matrices satisfying \(\displaystyle \sum_{i=1}^m\deg(a_i) \not= 0\) then \(f(a_1,\dots,a_m) = 0\). (S2) If \(a_1,\ldots,a_m \in M_n(K)\) are homogeneous matrices satisfying \(\displaystyle \sum_{i=1}^m\deg(a_i) = 0\) then \(\text{tr}(f(a_1,\dots,a_m)) = 0\). The results of the paper give for \(f(x_1,\ldots,x_m)\) and \(M_n(K)\) that: \begin{itemize} \item \(f\) satisfies (S1) and (S2) if and only if \(f\) is a polynomial identity; \item \(f\) satisfies (S1) and does not satisfy (S2) if and only if the image of \(f\) is \(K\). \item \(f\) satisfies (S2) and does not satisfy (S1) if and only if the linear span of the image of \(f\) is \(sl_n(K)\). \item \(f\) satisfies neither (S1) nor (S2), if and only if the linear span of the image of \(f\) is \(M_n(K)\). \end{itemize} Now the equivalent form of the Lvov-Kaplansky conjecture is: The Lvov-Kaplansky is true for \(M_n(K)\) if and only if for any \(m\geq 1\) and for any multilinear polynomial \(f\in K\langle x_1,\ldots,x_m\rangle\) the following assertions hold: \begin{itemize} \item If \(f\) does not satisfy (S1) and satisfies (S2), then \(\text{Im}(f) = sl_n(K)\). \item If \(f\) does not satisfy neither (S1) nor (S2), then \(\text{Im}(f) = M_n(K)\). \end{itemize} Similarly, the Mesyan conjecture is true if the following assertion holds: \begin{itemize} \item If \(f\) does not satisfy (S1) then \(\text{Im}(f)\supseteq sl_n(K)\). \end{itemize} As an application of their approach the authors give alternative proofs for several results by Kanel-Belov, Malev and Rowen.