It is possible to define the determinant of a tensor. We think of a tensor as a collection of numbers but this collection easily extends to a proper multilinear map. If $T:\{1,....,n\}^m\to \mathbb C$ then one can define $$\operatorname{Det}T:=\sum_{\sigma_2,...,\sigma_m\in S_n}\left(\left[\prod_{i=2}^m\operatorname{Sign}\sigma_i\right]\left[\prod_{k=1}^n T(k,\sigma_2(k),...,\sigma_m(k))\right]\right).$$

This determinant satisfies some rules that are similar to the normal determinant. For example, if we add a slice to a different slice then the determinant remains the same (for matrices, if we add one column to another one then the determinant does not change). If we view map $T$ as a tensor then the value of the determinant is also invariant under a change of basis (if the transition matrix has determinant one).

For fixed $m$, the naive algorithm to determine the number runs in $O(n!^{m-1})$ (ie not very efficiently). It would be very helpful to have an efficient algorithm to make some small calculations, or even knowing the computational complexity of calculating this number would be very helpful (preferably polynomial, of course). If there is any literature on this determinant (I have not been able to find any on this) then this would also be hugely welcomed.