Dual rook polynomials (Q1377870)

The classical rook polynomial \(P_{\mathfrak B}(x)\) of a board \(\mathfrak B\) (which can be considered as a finite subset of cells of an infinite chessboard) is \(P_{\mathfrak B}(x)=\sum _{k\geq0} ^{}R_k(\mathfrak B)x^k\), where \(R_k(\mathfrak B)\) is the number of ways to place \(k\) nonattacking rooks on the board \(\mathfrak B\). The property that rooks are nonattacking can be rephrased by saying that each row and column must contain at most one rook. What the author considers in the paper is in a way dual: With \(\widetilde R_k(\mathfrak B)\) denoting the number of ways to place \(k\) rooks on \(\mathfrak B\) such that each row and column contains at least one rook, he calls the polynomial \(\widetilde P_{\mathfrak B}(x)=\sum _{k\geq0} ^{}\widetilde R_k(\mathfrak B)x^k\) the dual rook polynomial of \(\mathfrak B\). Apparently, these polynomials occured (implicitly) for the first time in [\textit{P. Z. Wang, E. S. Lee} and \textit{S. K. Tan}, J. Math. Anal. Appl. 160, No. 2, 500-504 (1991; Zbl 0765.05005)] (with \(\mathfrak B\) being a rectangular board) in the context of fuzzy statistics. The main result of the paper under review (which extends the result in the aforementioned paper) is that if \(\mathfrak B\) is a skew Ferrers board then \(\widetilde P_{\mathfrak B}(-1)\) is always \(0,1\), or \(-1\). Clearly, this tells that the number of ways of placing an even number of rooks on a skew Ferrers board such that each row and column contains at least one rook differs from the number of ways of placing an odd number of rooks on a skew Ferrers board having the same property by \(0,1\), or \(-1\). The last section of the paper contains a number of intriguing conjectures. Recall that the sequence \((R_k(\mathfrak B))_{k\geq0}\) of ordinary rook numbers of any board \(\mathfrak B\) is (strongly) log-concave. Apparently, the same is true for the dual rook numbers \((\widetilde R_k(\mathfrak B))_{k\geq0}\), although the author is not able to prove it. In the case of ordinary rook numbers this follows from a special case of a result of \textit{O. J. Heilmann} and \textit{E. H. Lieb} [Commun. Math. Phys. 25, 190-232 (1972; Zbl 0228.05131)] which says that all the zeroes of the ordinary rook polynomial \(P_{\mathfrak B}(x)\) are real and nonpositive. While this is, in general, false for the dual rook polynomial \(\widetilde P_{\mathfrak B}(x)\), what seems to be true is that all the zeroes of \(\widetilde P_{\mathfrak B}(x)\) have nonpositive real part.