Images of multilinear polynomials in the algebra of finitary matrices contain trace zero matrices (Q2040550)

Let \(F\) be an infinite field and let \(F\langle X_1, \dots, X_n\rangle\) denote the free associative algebra, freely generated by the set \(X=\{ X_1, \dots, X_n\}\). The elements of \(F\langle X_1, \dots, X_n \rangle\) are called noncommutative polynomials. A polynomial \(f=f(x_1, \dots, x_n)\) is called multilinear if \(f\) can be written as \[ f(x_1, \dots, x_n)=\sum_{\sigma\in S_n}\lambda_\sigma X_{\sigma(1)}\cdots X_{\sigma(n)}, \] where \(S_n\) denotes the symmetric group and \(\lambda_\sigma\in F\). For a given \(F\)-algebra \(A\), a polynomial \(f(x_1,\dots,x_n)\in F\langle X_1, \dots, X_n\rangle\) defines a map (also denoted by \(f\)) \[ \begin{array}{cccc} f: & A^n & \longrightarrow & A \\ & (a_1,\dots,a_n) & \mapsto & f(a_1,\dots,a_n) \\ \end{array} \] The image of such map \(f\) is called the image of the polynomial \(f\) on \(A\). The most important and long-standing problem related to images of polynomials is the so called Lvov-Kaplansky conjecture. It states that the image of a multilinear polynomial \(f\) on \(M_d(F)\) is a vector space, which is equivalent to the image of \(f\) being one of the following: \(\{0\}\), \(F\) (viewed as the set of scalar matrices), \(sl_d(F)\) (the set of trace zero matrices) or \(M_d(F)\). Solutions to such problem are known only for \(d=2\) [\textit{A. Kanel-Belov} et al., Proc. Am. Math. Soc. 140, No. 2, 465--478 (2012; Zbl 1241.16017)] if \(F\) is quadratically closed. The case \(d=3\) has interesting advances, but not a solution up to now. The infinite-dimensional analogue of such conjecture (i.e., for \(A=\mathrm{End}(V)\), where \(V\) is an infinite-dimensional vector space over \(F\)) was proved by the author in [\textit{D. Vitas}, J. Algebra 565, 255--281 (2021; Zbl 1459.16027)], and states that an arbitrary nonzero multilinear polynomial is surjective on \(A\). In the paper under review, the author studies the similar question for the algebra \(M_\infty (F)\), of finitary matrices over \(F\), i.e., countably infinite matrices with only finitely many nonzero entries. The main theorem of the paper implies that if \(F\) is an infinite field and \(d\in \mathbb{N}\), then for every nonzero multilinear polynomial \(f\), there exists an \(s\in\mathbb{N}\) such that the image of \(f\) on \(M_s(F)\) contains \(sl_d(F)\). The above inclusion should be understand as \[M_d(F)=\begin{pmatrix} M_d(F) & 0 \\ 0 & 0 \end{pmatrix}\subseteq M_s(F).\] This result is similar to the statement of the so called Mesyan Conjecture [\textit{Z. Mesyan}, Linear Multilinear Algebra 61, No. 11, 1487--1495 (2013; Zbl 1290.15011)], a weaker form of the Lvov-Kaplansky Conjecture, which states that if \(d\geq n-1\) the image of a multilinear polynomial of degree \(n\) on \(M_d(F)\) contains \(sl_d(F)\). Such conjecture was proved for \(d=3\) by Mesyan and for \(d=4\) by Buzinsky and Winstanley. It is worth mentioning that the case \(d=4\) contains an error in its proof, but the result is still valid. An immediate consequence of the above theorem is that the analog of the Mesyan conjecture for \(M_\infty(F)\) is true. Namely, if \(F\) is an infinite field and \(f\) is a nonzero multilinear polynomial, then the image of \(f\) on \(M_\infty(F)\) contains \(sl_\infty(F)\) (the set of trace zero finitary matrices). Also, the reverse inclusion holds if and only if \(f\) is a sum of commutators. The paper uses similar methods and ideas as the ones in [\textit{D. Vitas}, J. Algebra 565, 255--281 (2021; Zbl 1459.16027)].