Abstract:
It is proved that the classical algorithm for computing
the Newton–Puiseux expansion of roots of a polynomial using the
Newton polygon's method has a polynomial complexity in the model
of computation when sizes of coefficients and constant fields are
taking into account. As a consequence we get polynomial-time algorithms
for factoring polynomials over fields of zero characteristic
of formal power series in one variable. Further, there
are constructed polynomial-time algorithms for computing uniformizing
elements of local rings of points of algebraic curves,
indices of ramification and inertia, the geometrical genus of a
curve, the normalization of an algebraic curve.
By definition we set that a series is computed in time polynomial
in $A_1,\dots,A_m$ iff for each $i$ its $i$-th partial sum is
computed in time polynomial in $A_1,\dots,A_m$ and $i$.
The estimation of length of coefficients of the Newton-Peuiseux expansion is central in the paper.
It is based mainly on the lemma 2.1 which affords to reduce the general problem to
the some analog of Hensel's lifting with only one step in the
Newton–Peiseux expansion till the stabilization, i.e. separation
of roots of the polynomial in the Newton polygon.