Artics
    16 posts
From RNN to ChatGPT
This note roughly introduces the development in natural language processing during the past decades. The RNN, LSTM, attention, and Transformer would be covered. Then, the technologies behind the recent popular ChatGPT will be shared, including GPT-series models, Prompts, and how to train language model fitting human intent
对一般刚体的定点转动的统计力学分析
在统计力学中, 我们通常会关注正则动量与坐标. 因为由它们确定的相空间体积元在正则变换下不变. 但在我们实际进行计算时, 却并非总是正则动量与坐标的表达最为方便. 一个例子就是刚体力学中, 我们会偏爱使用转动惯量和惯量主轴的方法来讨论问题. 然而作为角动量, 它们彼此之间并不对易, 因而在处理统计力学积分时往往带来困惑: 我们能使用它们作为被积变量吗? 进一步的, 用刚体转动模型理解多原子分子热容时同样会遇到这个问题. 本文旨在对这个问题给出解答: 在讨论不依赖转动构型的力学量时, 使用主轴角动量作为动力学变量是完全可行的. 变量替换的 Jacobian 只是转动构型的函数.
Introduction to Measurement Based Quantum Computing
In this note, we will generally discuss the basic conceptions of measurement-based quantum computing(MBQC). We focus on the workflow of MBQC and how to transpile a circuit-based model to MBQC. The parallization for quantum computing and other optimizations are not fully covered but the references are offered. Finally, we discuss the development of MBQC in theoretical perspective.