show · lf.newton_polygon all knowls · up · search:

Let KK be a nonarchimedean local field with discrete valuation vKv_K, and fK[x]f \in K[x], given by f(x)=a0+a1x++anxn, f(x) = a_0 + a_1 x + \cdots + a_nx^n, and such that a0an0a_0a_n \neq 0.

Then the Newton polygon of ff is the lower convex hull of the set of points (i,vK(ai))(i,v_K(a_i)), if we ignore all points with ai=0a_i=0.

In the context of isogeny classes of abelian varieties over finite fields, the constant term a0=1a_0=1 always, and therefore the Newton polygon begins at (0,0)(0,0). In this context, the LMFDB accepts two methods to enter Newton polygons:

  • The first is to give a list of slopes [s1,s2,s3,][s_1,s_2,s_3, \ldots]. This denotes the Newton polygon that, starting at (0,0)(0,0), is given by a line segment of width 1 with slope s1s_1, followed by a line segment of width 1 with slope s2s_2, etc. Note that it is possible for a slope to appear multiple times if the Newton polygon has the same slope for a width of more than 1. Note also that because the Newton polygon is a convex hull, the slopes sis_i will necessarily be increasing.

  • The second is to give a list of points, called "breaks," [(x1,y1),(x2,y2),(x3,y3),][(x_1,y_1),(x_2,y_2), (x_3,y_3), \ldots], for 0<x1<x2<x3<0 < x_1 < x_2 < x_3 < \ldots. This denotes the Newton polygon that, starting at (0,0)(0,0), has constant slope from (0,0)(0,0) to (x1,y1)(x_1,y_1), then has constant (but different) slope from (x1,y1)(x_1,y_1) to (x2,y2)(x_2,y_2), etc. Note that by the definition of the Newton polygon, each xix_i is a positive integer.

Authors:
Knowl status:
  • Review status: reviewed
  • Last edited by Andrew Sutherland on 2020-10-24 17:11:34
Referred to by:
History: (expand/hide all) Differences (show/hide)