Searcharxiv⌕ Search

arXiv subjects

Joshua Yi

Publications and source records attributed to Joshua Yi.

2 recordsLinked to original sources

Time Direction Entanglement Entropy of Free Fermion Ground States

While the entanglement entropy of a spatial subregion has been extensively studied, a properly defined entanglement entropy of quantum states in a time direction interval remains underexplored. We investigate the scaling behaviors of the recently defined time direction entanglement entropy (TDEE) for free fermion quantum states, which is computed from the reduced density matrix of a sub-Hilbert space that minimally supports all the time-evolved operators in the time direction interval. Based on the Fisher--Hartwig conjecture for Toeplitz matrices, we develop an analytical framework for understanding the TDEE in a time interval with discrete time step $τ$. For disorderless one-dimensional (1D) systems (gapless fermions or band insulators), in different regimes of gap, bandwidth and $τ$, we find the TDEE can exhibit area law (for nonzero gap), logarithmic law or volume law in the length of time. For 1D Anderson insulators, we numerically find the TDEE shows a Page-curve like behavior, growing at short time, and decaying in power-law at long time exceeding the crossover time determined by localization. In the continuum time limit $τ\rightarrow 0$, the TDEE of disorderless gapless fermions, band insulators and Anderson insulators exhibit logarithmic law, area law and sub-area law, respectively. We also show that the sub-Hilbert space of a time-direction interval stabilizes in the $τ\rightarrow 0$ limit, which is understandable as a subspace of the Hilbert space of the Lieb-Robinson causal diamond.

quant-ph↗

Universal Vector Neural Machine Translation With Effective Attention

Neural Machine Translation (NMT) leverages one or more trained neural networks for the translation of phrases. Sutskever introduced a sequence to sequence based encoder-decoder model which became the standard for NMT based systems. Attention mechanisms were later introduced to address the issues with the translation of long sentences and improving overall accuracy. In this paper, we propose a singular model for Neural Machine Translation based on encoder-decoder models. Most translation models are trained as one model for one translation. We introduce a neutral/universal model representation that can be used to predict more than one language depending on the source and a provided target. Secondly, we introduce an attention model by adding an overall learning vector to the multiplicative model. With these two changes, by using the novel universal model the number of models needed for multiple language translation applications are reduced.

cs.CL↗