Paper Image

Simplifying neural network optimization across quantum phase transitions

Published on:

12 March 2024

Primary Category:

Disordered Systems and Neural Networks

Paper Authors:

Riccardo Rende,

Sebastian Goldt,

Federico Becca,

Luciano Loris Viteritti

Bullets

Key Details

Pretrain NNQS at expressive transition point, then fine-tune output layer

Greatly reduces computational cost to explore phase diagrams

Learns interpretable features near transitions

Accurately describes wide region around transition

Applicable to diverse quantum systems

AI generated summary

Simplifying neural network optimization across quantum phase transitions

This paper demonstrates a new technique to efficiently explore quantum phase diagrams with neural network quantum states (NNQS). By pretraining a NNQS near a phase transition, it captures the essential physics, allowing accurate fine-tuning across the full diagram by only adjusting the output layer. This reduces costs versus separate training everywhere.

Answers from this paper

Comments

No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up