1. Critical points
-
Minimal and critical points
Suppose f(x) is a (multivariate) Laurent polynomial, and we are looking at its Laurent series expansion corresponding to a component C of the complement of the amoeba A_f. If f =\sum_{\omega} a_\omega x^{\omega}, a_\omega \in \mathbb Q_{\geq 0} (that is f is combinatorial), and p is a minimal point that is a minimizer of the height function -r \cdot \log|x|, then p is critical. However this is not necessarily true when f is not combinatorial.Problem 1.1.
[Steve Melczer] When do we have minimal points that are critical? -
Problem 1.2.
[Steve Melczer] If p minimizes -r \cdot x locally on the contour, then is it critical?-
Remark. [Steve Melczer] Yuliy Baryshnikov pointed out that this question is false in this level of generality. A counter-example is any contour that has cusps.
-
-
Let f(x_1,\dots,x_d) be a polynomial. Suppose p is a point such that f(p)=\frac{\partial f}{\partial x_j}(p)=0 for j=1,\dots,d. Let V_f denote the variety.
Problem 1.3.
[Robin Pemantle] Is V_f = \bigcup_{i=1}^m H_i, where H_i are smooth hypersurfaces in a small ball containing p? -
Problem 1.4.
[Robin Pemantle] Can computer algebra detect this? -
Symmetric multilinear functions
Let Q(z) be a multivariate Laurent polynomial, z=(z_1,\dots,z_d), and let r \in \mathbb C \mathbb P^{d-1}. In the smooth case, z is a critical point in direction r if \left(z_1 \frac{\partial Q}{\partial z_1},\dots, z_1 \frac{\partial Q}{\partial z_d}\right)=r in \mathbb C \mathbb P^{d-1}. In general, if you look at the maximal smooth piece of the variety V_Q that contains z, then this binary relation is (r is in the log normal space to V_Q at z).Problem 1.5.
[Robin Pemantle] Let r=[1:\cdots:1], Q be a symmetric multilinear function and \delta(x)=Q(x,\dots,x), \rho a minimum modulus root of \delta(x). Then (\rho,\dots,\rho) is a minimal point and a critical point in direction r. Can this be generalized to find critical points in other directions? -
Problem 1.6.
[Robin Pemantle] Can the multilinearity of Q be removed, but keeping r=[1:\cdots:1]? -
Critical points at infinity
Consider a multivariate Laurent polynomial Q(z), z=(z_1,\dots,z_d) and r \in \mathbb C \mathbb P^{d-1}. Let x^{(1)},x^{(2)},\dots be a sequence of points in the variety V_Q, where x^{(j)} is a critical point in the direction r^{(j)}. Let X denote the toric variety of the Newton polytope of Q. Let (x,r) be any subsequential limit of (x^{(j)},r^{(j)}) in X \times \mathbb C \mathbb P^{d-1}.Problem 1.7.
[Robin Pemantle] If x \in X is a point in the torus orbit corresponding to a face \Gamma of the Newton polytope of Q, then is r parallel to the face?
Cite this as: AimPL: Analytic combinatorics in several variables, available at http://aimpl.org/combinseveral.