Move to a new website

For anyone wander on this site, I have moved to a new domain and website https://www.guangshi.io The new site is built using eleventy and hosted on Netlify. It is more like a personal website instead of just a blog.

Use Tikhonov regularization to solve Fredholm integral equation

This article shows an algorithm to solve Fredholm integral equation of the first kind. Background A Fredholm integral equation of the first kind is written as, $latex f(x)=\int_{a}^{b}K(x,s)p(s)\mathrm{d}s. $ The problem is to find $latex p(s)$, given that $latex f(x)$ and $latex K(x,s)$ are known. This equation occurs quite often in many different areas. Discretization-based … Continue reading Use Tikhonov regularization to solve Fredholm integral equation

Use Multidimensional LSTM network to learn linear and non-linear mapping

This note is about the effectivenss of using multidimensional LSTM network to learn matrix operations, such as linear mapping as well as non-linear mapping. Recently I am trying to solve a research problem related to mapping between two matrices. And came up the idea of applying neural network to the problem. The Recurrent Neural Network … Continue reading Use Multidimensional LSTM network to learn linear and non-linear mapping

Generate gaussian chain/random walk using normal modes

Long time ago, I wrote about how to use Pivot algorithm to generate equilibrium conformations of a random walk, either self-avoiding or not. The volume exclusion of a self-avoiding chain make it non-trivial to generate conformations. Gaussian chain, on the other hand, is very easy and trivial to generate. In addition to the efficient pivot … Continue reading Generate gaussian chain/random walk using normal modes

Simulating Brownian Dynamics (overdamped Langevin Dynamics) using LAMMPS

I have migrated to my new domain and website here ⇒ https://www.guangshi.io/ LAMMPS is a very powerful Molecular Dynamics simulation software I use in my daily research. In our research group, we mainly run Langevin Dynamics (LD) or Brownian Dynamics (BD) simualtion. However, for some reason, LAMMPS doesn't provide a way to do Brownian Dynamics … Continue reading Simulating Brownian Dynamics (overdamped Langevin Dynamics) using LAMMPS

Big Bend徒步露营攻略

5月不是最适合去Big Bend露营的季节, 沙漠在每天最热的时候温度会高达40度, 再加上缺水, 所以很少人会选择夏季来Big Bend. 来Austin之后, 偶然看到一张South rim落日后的照片, 紫色的天空, 脚下是一眼望不到头的Chiwawa沙漠, 看到这张照片后就下定决心一定要去一趟. 无奈一直没有合适的时间, 再加上去Big Bend最适宜的季节 (冬季) 很难订到campsite, 一年多了也没有去成. 5月份学期结束, 我们决定干脆说走就走, 现在来看我俩很庆幸有了这次稍显准备仓促的旅行, 要不然真不知道什么时候能去成了. big bend NP位于德州与墨西哥交界处, 由一条叫Rio Grande的河分界, 整个公园属于Chiwawa沙漠的一部分. 由于位置偏远, (我们从austin开过去要8个小时, 圣安东尼奥算是离公园最近的大城市也得6个小时), 所以是美国到访人数最少的几个国家公园之一 (全美一共59个国家公园big bend到访人数排名42). 这里最受欢迎的活动是徒步和露营, 整个公园的trail一共长180英里(290公里), 由于游客稀少, 所以在这里的露营和徒步会给你非常难忘的体验. Big Bend另外一个出名的是这里的夜空, 这里有美国最黑的夜晚 (因为附近实在太荒了, 几乎没有任何光污染), 遗憾的是我们没能好好欣赏这里的银河. 推荐如果有机会一定要熬到后半夜看看银河. 这里是公园官方关于观星的页面 从五月到八月沙漠地区白天会有极端干燥高温天气, 对于没有什么野外生存经验的人, 由于暴晒和缺水, 夏天在Big Bend的沙漠平原两天以上的徒步和露营是存在生命危险的. 所以我们决定只在High Chisos活动. High chisos是位于公园中央地区的一片拔地而起的山脉, … Continue reading Big Bend徒步露营攻略

Automated python script to access ipython notebook on a remote server

My research work involves a lot of using of IPython Notebook. I usually do it on an office MAC. However I also very often need to access it from home. After a brief searching, I found these three wonderful articles on this topic. Remote hosting of IPython Notebooks Accessing IPython Notebook remotely over an SSH … Continue reading Automated python script to access ipython notebook on a remote server