Some lower bounds in the B. and M. Shapiro conjecture for flag varieties (Q650115): Difference between revisions

Schubert calculus allows to count geometric objects subject to certain incidence relations. Its rigorous foundation was established by intersection theory. For example, consider \(2d-2\) lines in general position in \(\mathbb{P}^d(\mathbb{C})\). The Schubert Calculus allows to prove that the number of \(2\)-subspaces of \(\mathbb{P}^d(\mathbb{C})\) meeting all \(2d-2\) lines is exactly the \(d\)-th Catalan number. From the point of view of intersection theory, such a problem is solved by counting intersection multiplicities of Schubert varieties in the Grassmannian. If all the \(2d-2\) lines are reals one can ask how many of the previous \(2\)-subspaces are real. Problem like this are considerably more complicated. For this particular problem, in [\textit{F. Sottile}, Exp. Math. 9, No. 2, 161--182 (2000; Zbl 0997.14016)] was proved that for some choice of the \(2d-2\) lines all these 2-codimension subspaces are real. Boris and Michael Shapiro conjectured that this happen when all the given lines are tangent to the rational normal curve \(\gamma(t)=[1,t,t^{2},\dots,t^d]\) at distinct real points. This conjecture has been proved in [\textit{A. Eremenko} and \textit{A. Gabrielov}, Ann. Math. (2) 155, No. 1, 105--129 (2002; Zbl 0997.14015)]. In [\textit{E. Mukhin, V. Tarasov} and \textit{A. Varchenko}, Ann. Math. (2) 170, No. 2, 863--881 (2009; Zbl 1213.14101)], a more general version of the previous conjecture for higher dimensional subspaces was proved. Then B. and M. Shapiro suggested an extension of their conjecture to flags, but this last conjecture fails (see the previous work of Sottile for some counterexample). However, experiments suggest that the conjecture might hold whenever a certain monotonicity condition is met (see [\textit{J. Ruffo, Y. Sivan, E. Soprunova} and \textit{F. Sottile}, Exp. Math. 15, No. 2, 199--221 (2006; Zbl 1111.14049)]). The authors are interested in a special case of the previous generalized conjecture, which (assuming the monotonicity condition) was proved in [\textit{A. Eremenko, A. Gabrielov, M. Shapiro} and \textit{A. Vainshtein}, Proc. Am. Math. Soc. 134, No. 4, 949--957 (2006; Zbl 1110.14052)]. More precisely, they are interested to the following problem: Let \(y_1,\dots,y_{2d-3},r,s\) be distinct real points. For \(1\leq i\leq 2d-3\), let \(T_i\) be the line tangent to \(\gamma\) at \(y_i\) and let \(T_{2d-2}\) be the line through the points \(\gamma(r)\) and \(\gamma(s)\). Among the codimension 2-subspaces of \(\mathbb{P}^d(\mathbb{C})\) meeting all the previous lines, how many are real? In this case the monotonicity condition means that the interval \(I\) with endpoints \(r\) and \(s\) contains all or none of the \(y_1,\dots,y_{2d-3}\). In this case (or more generally when the special case of the conjecture holds) all the points are real. By the cited result of Sottile there are cases where none points are real. In this article, a lower bound to the number of real solutions of the previous problem is given (when the monotonicity condition does not hold). The authors use a point of view similar to the cited Eremenko and Gabrielov proof of the first B. and M. Shapiro conjecture. Eremenko and Gabrielov have proved that, if the critical points of a rational function are all real, then the function is a real rational function up to composing with a fractional linear transformation. The authors of this work study the following problem (equivalent to the previous one): Let \(y_1,\dots,y_{2d-3},r,s,I\) as before and suppose that the monotonicity condition is not satisfied, i.e. \(I\) contains some but not all the \(y_i\). How many equivalence classes of real rational functions \(f\) of degree \(d\) having critical points at \(y_1,\dots,y_{2d-3}\) and satisfying \(f(r)=f(s)\) are there?