A new perspective on the powers of two descent for discrete logarithms in finite fields

TL;DR

This paper offers a new geometric proof for the powers of two descent method in discrete logarithm computations, simplifying the analysis and potentially advancing toward quasi-polynomial time algorithms in finite fields.

Contribution

It provides a unified geometric argument for the powers of two descent method, strengthening the theoretical foundation and removing the need for subgroup analysis.

Findings

Stronger proof of the powers of two descent correctness

Unified geometric approach to analyze the method

Potential implications for quasi-polynomial time algorithms

Abstract

A new proof is given for the correctness of the powers of two descent method for computing discrete logarithms. The result is slightly stronger than the original work, but more importantly we provide a unified geometric argument, eliminating the need to analyse all possible subgroups of $\mathrm{PGL}_2(\mathbb F_q)$. Our approach sheds new light on the role of $\mathrm{PGL}_2$, in the hope to eventually lead to a complete proof that discrete logarithms can be computed in quasi-polynomial time in finite fields of fixed characteristic.