Optimization and Identification of Lattice Quantizers
Artikel i vetenskaplig tidskrift, 2025
Lattices with minimal normalized second moments are designed using a new numerical optimization algorithm. Starting from a random lower-triangular generator matrix and applying stochastic gradient descent, all elements are updated towards the negative gradient, which makes it the most efficient algorithm proposed so far for this purpose. A graphical illustration of the theta series, called theta image, is introduced and shown to be a powerful tool for converting numerical lattice representations into their underlying exact forms. As a proof of concept, optimized lattices are designed in dimensions up to 16. In all dimensions, the algorithm converges to either the previously best known lattice or a better one. The dual of the 15-dimensional laminated lattice is conjectured to be optimal in its dimension and its exact normalized second moment is computed.
mean square error
lattice design
numerical optimization
moment of inertia
theta series
Algorithm
stochastic gradient descent
theta image
vector quantization
normalized second moment
laminated lattice
quantization constant
lattice quantization
Voronoi region
Författare
Erik Agrell
Max-Planck-Gesellschaft
Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk
Daniel Pook-Kolb
Leibniz Universität Hannover
Bruce Allen
Leibniz Universität Hannover
IEEE Transactions on Information Theory
0018-9448 (ISSN) 1557-9654 (eISSN)
Vol. In PressÄmneskategorier (SSIF 2025)
Sannolikhetsteori och statistik
Den kondenserade materiens fysik
DOI
10.1109/TIT.2025.3565218