Worked solution: Gelfond–Schneider transcendence proof via auxiliary functions (1934)
Complex analysis has a classical tool, Jensen's formula, that relates the value of a holomorphic function at the origin to how large it grows on a big circle and where its zeros sit inside that circle. Because our has zeros of high order at every one of the points , this formula essentially says: since vanishes so much inside the circle, its value (or the value of its -th derivative) at the origin has to be very small — smaller, in fact, than any fixed power of can compensate for, once the radius of the circle is chosen just right.
Carrying out the estimate carefully shows that decays roughly like for some constant depending only on — a genuinely tiny quantity once is large.
Jensen's formula states that for holomorphic on a disk of radius with zeros inside (and ), ; when vanishes to order at the formula generalizes to relate to the same quantities (Siu, 'Jensen's Formula'). Applied to , but restricted only to the points (each contributing a term , using that has order zeros there) gives an inequality, since these are not literally all the zeros of in the disk (Siu, inequality labeled ).
Bounding the growth of on the circle : since and have order (i.e. ), and has coefficients and degree of size , one gets (Siu, growth estimate via the Appendix's derivative bounds ). Choosing balances the two terms and gives .
Combining both sides of Jensen's formula (the left side contributing from the known zeros, the right side bounded by plus , since the 'size' of the algebraic number costs an extra factor of to account for its conjugates) yields the final analytic bound: for an explicit constant depending on once is chosen large enough relative to (Siu, proof of Main Theorem, final comparison of growth orders).
- Jensen's formula
- A formula in complex analysis expressing for a holomorphic function in terms of the locations of its zeros inside a disk and the average of on the boundary circle — the key tool for turning 'many zeros' into a numerical upper bound on a function's value.