AbstractLogical cryptanalysis has been introduced by Massacci and Marraro as a general framework for encoding properties of crypto-algorithms into SAT problems, with the aim of generating SAT benchmarks that are controllable and that share the properties of real-world problems and randomly generated problems.In this paper, spurred by the proposal of Cook and Mitchell to encode the factorization of large integers as a SAT problem, we propose the SAT encoding of another aspect of RSA, namely finding (i.e. faking) an RSA signature for a given message without factoring the modulus.Given a small public exponent e, a modulus n and a message m, we can generate a SAT formula whose models correspond to the eth roots of m modulo n, without encoding t...