A note on Wronskians and the ABC theorem in function fields of prime characteristic (Q1283499): Difference between revisions

Voloch and Brownawell and Masser proved the ``ABC-theorem in several variables'' for function fields of characteristic 0. They used nonvanishing properties of certain Wronskians. Their work was extended by the author to function fields of positive characteristic [J. Number Theory 58, 139-157 (1996; Zbl 0866.11066)]. In the present paper she further generalizes her results. To be precise, she proves a ``truncated second main theorem'' for function fields of characteristic \(p\geq 0\). Her main tool is again a nonvanishing result for Wronskians suitably adapted to the characteristic \(p\)-situation. Let \(C\) be an irreducible nonsingular projective algebraic curve of genus \(g\) over an algebraically closed field \(k\) of characteristic \(p> 0\). Let \(K\) be the function field of \(C\). Define the fields \(K_m:=K^{p^m}\). Suppose \(p^a\leq n<p^{a+1}\) and \(m\geq a+1\). Let \(x_0\ldots x_n\) be elements of \(K\). Denote by \(l_c+1\) the dimension of the vector space spanned by \(x_0\ldots x_n\) over \(K_c\) for \(0\leq c\leq m\). Let \(0=c_0<c_1<\cdots <c_u\leq m\) be the indices with \(0=l_{c_0}<l_{c_1}<\cdots <l_{c_u}=n\) and \(l_j=l_{c_i}\) for \(c_i\leq j <c_{i+1}\), \(i=0\ldots u-1\). For every point \(P\in C\), let \(v_P\) be the discrete valuation with \(v_P(t)=1\) where \(t\) is a uniformizing parameter at \(P\). Define the height \[ h({\mathbf x})= -\sum_{P\in C(K)}\min (v_P(x_0)\ldots v_P(x_n)), \] where \({\mathbf x}=(x_0\ldots x_n)\). Let \(L_1\ldots L_q\) (\(q\geq n+1\)) be linear forms in \(n+1\) variables with coefficients in \(k\) and in general position and define \[ \lambda_{i,P}({\mathbf x})= v_P(L_i(x_0\ldots x_n))-\min_{0\leq j\leq n}\big( v_P(x_j)\big). \] Let \(S\) be a finite subset of \(C(K)\). Then the truncated second main theorem reads as follows. Let \(x_0\ldots x_n\) be elements of \(K\), linearly independent over \(K_m\) and put \({\mathbf x}=(x_0\ldots x_n)\). Then there are integers \(0=e_0<e_1<\cdots <e_n\) with \[ e_{l_{i-1}+s}\leq \min (sp^{c_i-1},p^{c_i}-1) \quad \text{for }1\leq s\leq l_{c_i}-l_{c_{i-1}} \] such that \[ (q-n-1)h({\mathbf x})\leq\sum_{i=1}^q\sum_{P\not\in S} \min_{1\leq i\leq q}\{e_n,\lambda_{i,P}({\mathbf x})\}+(e_1+\cdots +e_n)\max (0,2g-2+|S|). \] Finally, the author observes that if \(K\) has characteristic \(0\), the same estimate holds with \(e_i=i\).