An LLL algorithm with symmetries (Q6969885)

The article under review offers a profound and elegant generalization of the classical LLL algorithm by reframing lattice reduction in terms of arithmetic group actions on symmetric spaces. This conceptual leap to arbitrary split semisimple groups represents a significant theoretical advancement. The authors provide a coherent framework for defining reduced points in the symmetric space \(X_G\) and present a general reduction algorithm (Algorithm 2.10) that specializes to known cases. The detailed treatment of classical groups \(\mathrm{Sp}_{2g}\) and \(\mathrm{SO}_{2g}\), along with the exceptional group \(G_2\), constitutes a substantial contribution that will interest researchers in arithmetic geometry and representation theory.\N\NHowever, when evaluated as a contribution to algorithmic number theory, several limitations emerge. The claim of polynomial-time complexity, while intuitively appealing, lacks rigorous substantiation. The termination proof establishes an iteration bound but does not analyze the computational cost of critical operations, particularly the size-reduction step in the unipotent group \(N\), where commutator calculations in a ``good ordering'' of roots present significant complexity challenges.\N\NThe empirical validation remains preliminary. The provided examples for \(\mathrm{SO}_8\) and \(G_2\) demonstrate feasibility but lack performance metrics such as iteration counts, runtime analysis, or quantitative measures of output quality. Without benchmarks or scaling studies, the algorithm's practical efficiency remains uncertain.\N\NThe connection to arithmetic applications, particularly 2-descent for Jacobians via stably graded Lie algebras, is mathematically well-motivated but not fully realized computationally. The Section 4 example reduces a matrix associated with a curve but does not demonstrate how this reduction facilitates the computation of concrete arithmetic invariants. The acknowledgment that minimization algorithms for higher genus remain under development further limits current applicability.\N\NThe mathematical formalism, while rigorous, presents implementation challenges. The transition from the concrete geometry of \(\mathrm{SL}_g\) to the general group-theoretic framework is abrupt, and Algorithm 2.10 requires substantial expertise in reductive group structure theory for implementation.