The paper is about measuring how well certain special numbers can be approximated by algebraic numbers, or equivalently, how small a polynomial with integer coefficients can get when evaluated at these special numbers. The specific number in question is the ratio of the derivative of the Bessel function J0 to J0 itself, evaluated at an algebraic point. Bessel functions arise throughout physics and engineering, and their values at algebraic arguments are known to be transcendental, meaning they cannot be roots of any polynomial with integer coefficients. A transcendence measure makes this more precise by giving a lower bound on how small such a polynomial can be, in terms of the polynomial's degree and the size of its coefficients.
Previous results, following from general theorems by Lang and Galochkin, gave a lower bound with an exponent of 4d squared times delta, where d is the degree of the algebraic input and delta is the degree of the polynomial being evaluated. The authors improve this bound by showing the exponent can be reduced by at least 2d times delta plus 1, which is a meaningful savings especially when d and delta are large. The result also extends beyond the Bessel function to a broader class of functions called E-functions, which are power series with algebraic coefficients that satisfy linear differential equations and have controlled coefficient growth. The key requirement is that the function and its derivative satisfy a certain algebraic independence condition.
The method builds on the classical Siegel-Shidlovskii approach, a powerful technique in transcendence theory that uses carefully constructed auxiliary functions and determinants. The authors sharpen the argument by optimizing the size of a key determinant that appears in this method, following a strategy they previously used to improve transcendence measures for the number e raised to algebraic powers. The improvement is not just a technical refinement for its own sake; getting tighter exponents in transcendence measures is a long-standing and difficult problem, and progress here reflects a genuine advance in understanding how transcendental these values really are.