Chiustin

2020年3月15日星期日

2020年3月13日星期五

LAMMPS 水的force field

http://www.orsi.sems.qmul.ac.uk/downloads.html

2020年3月12日星期四

1. cora dataset介紹
cora有2708個數據點，每個數據點代表一篇科學論文，論文可以分成7個類別，個別為

(a) 規則學習 (Rule Learning)
(b) 強化學習 (Reinforcement Learning)
(c) 機率方法 (Probabilistic Methods)
(d) 理論 (Theory)
(e) 神經網路 (Neural Networks)
(f) 基因演算法(Genetic Algorithms)
(g) 案例式 (Case Based)

core有兩個檔案cora.cites和cora.content

cora.content總共有2708行，每一行分別是論文編號、論文的詞向量，一個有1433位的二進制和論文的類別

論文編號論文詞向量(1433位二進制) 論文類別
210XX 0 0..... 0 0 1 0 0 0 0 0 0 0 0 0 Neural_Networks
10723XX 0 0..... 0 0 0 0 0 1 0 0 0 0 0 0 Rule_Learning
11537XX 0 0..... 0 0 0 0 0 0 0 0 1 0 0 Reinforcement_Learning

cora.cites總共有5429行，每一行有兩個論文編號，第一個編號的論文先發表，第二個編號的論文引用第一個編號的論文

第一篇論文編號第二篇論文編號
35 1033
35 103482
35 103515

參考
https://blog.csdn.net/u010159842/article/details/103234942

https://github.com/tkipf/keras-gcn
https://blog.csdn.net/Eric_1993/article/details/102907104
https://luweikxy.gitbooks.io/machine-learning-notes/content/content/deep-learning/graph-neural-networks/graph-convolutional-networks/gcn-preliminary-understand.html
https://www.zhihu.com/question/54504471

2020年3月11日星期三

oplsaa轉到LAMMPS

tinker
https://dasher.wustl.edu/tinker/distribution/params/oplsaa.prm

oplsaa-erg_ff
https://bitbucket.org/comcon1/oplsaa-erg_ff/src/default/

Dihedral Coeffs
https://lammps.sandia.gov/threads/msg32415.html

Moltemplate OPLS-AA improper parameters
https://lammps.sandia.gov/threads/msg72033.html

2020年3月10日星期二

MD force field

http://erg.biophys.msu.ru/erg/tpp/index.py
http://zarbi.chem.yale.edu/ligpargen/
http://www.charmm-gui.org/?doc=input

2020年3月9日星期一

Lammps安裝過程

Updating Homebrew...
==> Auto-updated Homebrew!
Updated 2 taps (homebrew/core and homebrew/cask).
==> Updated Formulae
camlp5 lablgtk menhir ocaml ocaml-num ocamlsdl
inadyn libexosip mkvtoolnix ocaml-findlib ocamlbuild verilator
==> Deleted Formulae
camlp4
==> Updated Casks
homebrew/cask/amethyst homebrew/cask/freeplane homebrew/cask/mucommander homebrew/cask/tinymediamanager
homebrew/cask/base homebrew/cask/j homebrew/cask/mudlet homebrew/cask/veusz
homebrew/cask/blitz homebrew/cask/keystore-explorer homebrew/cask/numi homebrew/cask/xamarin-profiler
homebrew/cask/burn homebrew/cask/kode54-cog homebrew/cask/pinegrow homebrew/cask/yinxiangbiji
homebrew/cask/daisydisk homebrew/cask/mcgimp homebrew/cask/psychopy
homebrew/cask/electrocrud homebrew/cask/mellow homebrew/cask/refined-github-safari

==> Installing dependencies for lammps: hwloc, libevent, open-mpi, fftw and kim-api
==> Installing lammps dependency: hwloc
==> Downloading https://homebrew.bintray.com/bottles/hwloc-2.1.0.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/94/94e4e238c45da330b53fde9c622e74a2dfabd00a17f37fa1807b1d828452759d?__gda__=exp=1583746103~hmac=200ec34b4765b09c65c523c
######################################################################## 100.0%
==> Pouring hwloc-2.1.0.catalina.bottle.tar.gz
==> Caveats
Bash completion has been installed to:
/usr/local/etc/bash_completion.d
==> Summary
🍺 /usr/local/Cellar/hwloc/2.1.0: 881 files, 9.5MB
==> Installing lammps dependency: libevent
==> Downloading https://homebrew.bintray.com/bottles/libevent-2.1.11_1.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/9d/9d262f9ffb2268340a89c713826d8ca068bcac06c30baf49e6184ab4660d977a?__gda__=exp=1583746109~hmac=a9d25e6511b5ddd7eeaa86c
######################################################################## 100.0%
==> Pouring libevent-2.1.11_1.catalina.bottle.tar.gz
🍺 /usr/local/Cellar/libevent/2.1.11_1: 1,063 files, 5MB
==> Installing lammps dependency: open-mpi
==> Downloading https://homebrew.bintray.com/bottles/open-mpi-4.0.3.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/3b/3b143cf02a5345bb0d4df0777d3a34f806ff7fb66dc5d21993b8c4f218722ac7?__gda__=exp=1583746115~hmac=253bae48c6c40c68026dc37
######################################################################## 100.0%
==> Pouring open-mpi-4.0.3.catalina.bottle.tar.gz
🍺 /usr/local/Cellar/open-mpi/4.0.3: 752 files, 11.3MB
==> Installing lammps dependency: fftw
==> Downloading https://homebrew.bintray.com/bottles/fftw-3.3.8_1.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/27/271f7a5febd92b8757948c86fa24546aca9025087406056b7cf4036ca97b29e4?__gda__=exp=1583746121~hmac=4b173984e6ec8b297aa0ec8
######################################################################## 100.0%
==> Pouring fftw-3.3.8_1.catalina.bottle.tar.gz
🍺 /usr/local/Cellar/fftw/3.3.8_1: 73 files, 14.8MB
==> Installing lammps dependency: kim-api
==> Downloading https://homebrew.bintray.com/bottles/kim-api-2.1.3.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/ae/ae59f45bf8d539f1b633120cdb942c2a0a0f1880eee4ee91981f268792e1b1c4?__gda__=exp=1583746126~hmac=464da1ff6f45e08326d29d8
######################################################################## 100.0%
==> Pouring kim-api-2.1.3.catalina.bottle.tar.gz
Warning: kim-api dependency gcc was built with a different C++ standard
library (libstdc++ from clang). This may cause problems at runtime.
==> Caveats
Bash completion has been installed to:
/usr/local/etc/bash_completion.d

zsh completions and functions have been installed to:
/usr/local/share/zsh/site-functions

Emacs Lisp files have been installed to:
/usr/local/share/emacs/site-lisp/kim-api
==> Summary
🍺 /usr/local/Cellar/kim-api/2.1.3: 1,285 files, 21.6MB
==> Installing lammps
==> Downloading https://homebrew.bintray.com/bottles/lammps-2019-08-07.catalina.bottle.tar.gz
==> Downloading from https://akamai.bintray.com/60/604056c80bb3b36f0a1644388bff26f2bcdc3e2f2541247e9a7ce941b20e9bcc?__gda__=exp=1583746133~hmac=2bd684fb81d50cc717258b5
######################################################################## 100.0%
==> Pouring lammps-2019-08-07.catalina.bottle.tar.gz
Warning: lammps dependency gcc was built with a different C++ standard
library (libstdc++ from clang). This may cause problems at runtime.
🍺 /usr/local/Cellar/lammps/2019-08-07: 5,788 files, 335.5MB
==> Caveats
==> hwloc
Bash completion has been installed to:
/usr/local/etc/bash_completion.d
==> kim-api
Bash completion has been installed to:
/usr/local/etc/bash_completion.d

zsh completions and functions have been installed to:
/usr/local/share/zsh/site-functions

Emacs Lisp files have been installed to:
/usr/local/share/emacs/site-lisp/kim-api

Mac install tensorflow 2.1.0, Keras 2.3.1, scikit-learn, and RDkit

Tensorflow:
我選擇system安裝，不使用virtual environment

$ /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
$ export PATH="/usr/local/bin:/usr/local/sbin:$PATH"
$ brew update
$ brew install python
$ sudo pip3 install -U virtualenv
$ pip3 install --user --upgrade tensorflow

如果是python3.7.3版本的話要修改linecache.py

$ sudo vi /Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/linecache.py

第48行改成
for mod in list(sys.modules.values()): # for mod in sys.modules.values():

Keras:

$ brew install hdf5
$ brew install graphviz
$ sudo pip3 install pydot (一定要加sudo)
$ vi ~/.bash_profile
export PATH=/usr/local/bin:/usr/local/sbin:~/bin:$PATH
$ source ~/.bash_profile
$ sudo pip3 install keras (一定要加sudo)

scikit-learn:
$ pip3 install -U scikit-learn

RDkit:
$ brew tap rdkit/rdkit
$ brew install rdkit

If you need to have icu4c first in your PATH run:
echo 'export PATH="/usr/local/opt/icu4c/bin:$PATH"' >> ~/.zshrc
echo 'export PATH="/usr/local/opt/icu4c/sbin:$PATH"' >> ~/.zshrc

For compilers to find icu4c you may need to set:
export LDFLAGS="-L/usr/local/opt/icu4c/lib"
export CPPFLAGS="-I/usr/local/opt/icu4c/include"

==> openblas
openblas is keg-only, which means it was not symlinked into /usr/local,
because macOS provides BLAS and LAPACK in the Accelerate framework.

For compilers to find openblas you may need to set:
export LDFLAGS="-L/usr/local/opt/openblas/lib"
export CPPFLAGS="-I/usr/local/opt/openblas/include"

==> rdkit
You may need to add RDBASE to your environment variables.
For Bash, put something like this in your $HOME/.bashrc:
export RDBASE=/usr/local/share/RDKit

參考
https://github.com/rdkit/homebrew-rdkit

Mac執行tensorflow發生RuntimeError: dictionary changed size during iteration

問題：
Homebrew的python3版本是3.7.3
在執行tensorflow時出現以下

.
.
.
File "/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/linecache.py", line 48, in getlines
for mod in sys.modules.values():

RuntimeError: dictionary changed size during iteration

解決：
$ sudo vi /Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/linecache.py

第48行改成
for mod in list(sys.modules.values()): # for mod in sys.modules.values():

參考
http://www.kaierlong.me/blog/post/kaierlong/tensorflow-2.0-%E9%94%99%E8%AF%AF-RuntimeError-dictionary-changed-size-during-iteration

2020年3月8日星期日

Free AI, ML, Deep Learning Video Lectures

https://www.marktechpost.com/free-ai-ml-deep-learning-video-lectures/?fbclid=IwAR15uKRk-9qptjRnxOcq_ujEYYBy-dWlHYiaoaUt5OUL3x4bc9kPrvwlicc

2020年3月4日星期三

在 Ubuntu Linux 18.04 安裝 NVIDIA 驅動程式, CUDA, cuDNN, PyTorch, TorchVision, TensorFlow, Keras, XGBoost 等人工智慧軟體套件

https://tw.openrobot.org/article/index?sn=11701&fbclid=IwAR1kphLLIRkyS5RcxQ1EUo1IFdUPtihwqXYzRlepRY_zHZjIU-i4NlI41_4

2020年3月3日星期二

廣義卷積神經網絡與任意連續數據上李群的等方差

卷積層的平移等方差使卷積神經網絡能夠很好地泛化圖像問題。雖然平移等方差為圖像提供了強大的歸納偏置(inductive bias)，但我們通常還希望與其他變換（例如旋轉）等方差，尤其是對於非圖像數據。我們提出了一種通用的方法來構造一個卷積層，該卷積層與帶有滿射的(surjective )指數映射(exponential map)的任何指定李群(Lie group)的變換等價。將等方差合併到新組中僅需要實現組指數映射和對數映射即可實現快速原型製作。為了展示我們方法的簡單性和通用性，我們將相同的模型體系結構應用於圖像，球棒分子數據和哈密頓動力學系統。對於漢密爾頓系統，我們模型的等方差尤其具有影響力，從而可以精確地守恆線性和角動量。

https://arxiv.org/pdf/2002.12880.pdf

A Deep Generative Model for Fragment-Based Molecule Generation

In this paper, the authors discuss about the molecule generation, which is a challenging problem in cheminformatics. The two types of deep generative approaches generally used are encoding molecular graphs as strings of text, and learns their corresponding character based language model while another approach operates directly on the molecular graph. But the above approaches have two limitations, like the generation of invalid and duplicate molecules.

To overcome the limitations of the model, the authors of this paper proposed a language model for small molecular structures called fragments, loosely inspired by the well known paradigm of Fragment-Based Drug Design. In simple language, they proposed to generate molecules fragment by fragment, instead of atom by atom.

The authors of this paper experimentally show that their model largely outperforms other language model-based competitors, reaching state-of-the-art performances typical of graph-based approaches.

In terms of the methods used in this paper, the main approach encompasses three steps:

1.Break molecules into sequences of fragments
2.Encode them as SMILES words
3.Learn their corresponding language model

https://arxiv.org/pdf/2002.12826.pdf

https://github.com/marcopodda/fragment-based-dgm