Projective coefficient heights of univariate polynomials #
The height of a polynomial is the native projective height of its coefficient vector. It is unchanged by nonzero scalar multiplication; constants and the zero polynomial have multiplicative height one. The normalization is relative to the admissible family of absolute values on the coefficient field, rather than its degree-root normalization.
The product estimate uses Mathlib's Height.mulHeight_linearMap_apply_le on coefficient
convolution. Its archimedean factor counts the coefficients of the smaller-degree factor;
the nonarchimedean estimate has no such factor. These estimates are used to control
coefficients of auxiliary polynomials in Diophantine approximation.
References #
- Lukas Pottmeyer, Diophantine Approximation, §3.2, Definition 3.2.1 and Lemma 3.2.2
(17 January 2022). Those notes use absolute heights; here we use the relative height
supplied by
Height.AdmissibleAbsValues.
The relative projective multiplicative height of the coefficients of a polynomial.
Instances For
The relative projective logarithmic height of the coefficients of a polynomial.
Instances For
Polynomial logarithmic height agrees with the native finitely supported height.
Polynomial height is the height of the tuple indexed by its nonzero coefficients.
Any finite coefficient tuple containing the support gives the polynomial height.
The finite coefficient-tuple description of logarithmic polynomial height.
Any finite coefficient tuple extending past the degree gives the polynomial height.
The logarithmic coefficient-tuple description when the tuple extends past the degree.
Nonzero scalar multiplication does not change projective coefficient height.
The coefficient height of the linear polynomial with root a is the scalar height of a.
The logarithmic coefficient height of the linear polynomial with root a is the
scalar logarithmic height of a.
The height of a product is bounded by the product of coefficient heights, with the archimedean convolution factor given by the smaller degree plus one.
The logarithmic form of the coefficient convolution estimate.