No label defined (Q4432389)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1996442
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 1996442 |
Statements
2000
0 references
multiple Laurent series
0 references
number of distinct Laurent expansions
0 references
polynomial amoebas
0 references
estimate for the number of components of the amoeba complement
0 references
Newton polyhedron
0 references
vertex polynomial
0 references
0.8556920289993286
0 references
0.8116330504417419
0 references
Any function \(h\), which is holomorphic in some open polyannulus, may be represented as a multiple Laurent series. An important special case is when \(h\) is taken to be a rational function \(h = g/f\). It is clear that the number of distinct Laurent expansions for \(h\) is the same as for the simpler fraction \(h=1/f.\) But how many are there? The number of distinct Laurent series expansions of \(1/f\) is equal to the number of connected components of \(\mathbb R^n\setminus\mathcal A_f\), where \(\mathcal A_f\) denotes the amoeba of \(f\): \(\mathcal A_f = \{(\log| z_1| ,\dots,\log| z_n| )\); \(f(z)=0, z\in(\mathbb C\setminus\{0\})^n\}.\) There is a lower and an upper estimate for the number of components of the amoeba complement, and they are both phrased in terms of the Newton polyhedron \(\mathcal N_f\) associated with the polynomial \(f\). If \(f\) is given by \(\sum_{\alpha\in A} a_\alpha z^\alpha\), for some finite set \(A\subset\mathbb Z^n\subset\mathbb R^n\), then its Newton polyhedron \(\mathcal N_f\) is by definition the convex hull in \(\mathbb R^n\) of \(A\). Let \(m_f\) denote the number of components \(\mathbb R^n\setminus\mathcal A_f\). Also, let \(v_f\) be the number of vertices of \(\mathcal N_f\), while \(i_f\) is the cardinality of \(\mathcal N_f\cap \mathbb Z^n\). The inequalities then hold: \(v_f\leq m_f\leq i_f\). The upper bound is sharp. It is not known if the lower bound is also sharp, but numerous examples seem to indicate that this is the case. Let us call a Laurent polynomial \(f\) a vertex polynomial if it only contains monomials \(z^\alpha\) such that \(\alpha\) are vertices of the Newton polytope \(\mathcal N_f\). The author advances the Conjecture A: For any vertex polynomial \(f\), one has \(m_f = v_f\). Given a vertex polynomial \(f(z) =\sum_\alpha a_\alpha z^\alpha\), define its spine to be the piecewise linear subset \(\mathcal S_f\subset\mathbb R^n\) given by \(\{(\log| z_1| ,\dots,\log| z_n| )\); \(| a_\alpha z^\alpha| =| a_\beta z^\beta| =\max_\gamma | a_\gamma z^\gamma|\) , or some \(\alpha\neq\beta\}\). In other words, \(\log| z| \) belongs to the spine precisely if there is no term in \(f(z)\) with strictly largest modulus. It is not hard to see that \(\mathbb R\setminus \mathcal S_f\) consists of \(v_f\) different connected components, none of which is bounded. Conjecture B: For any vertex polynomial \(f\) one has \(\mathcal S_f\subset A_f\) and these two sets have the same homotopy type. The analogous statement is false for a general polynomial \(f\), if one uses the above recipe for the spine \(\mathcal S_f\). This is easily seen already in one variable, and a multidimensional counterexample occurs in Example 4 of the paper. A list of amoebas computed explicitly is given. For the entire collection see [Zbl 1019.00009].
0 references