这个回答解释了如何使用Poetry依赖管理创建Python 3、Jupyter 1和ipykernel 5工作流。诗歌使创建一个虚拟环境的Jupyter笔记本很容易。我强烈建议不要运行python3命令。安装全局依赖项的Python工作流程会让你陷入依赖项地狱。

以下是对干净、可靠的Poetry工作流程的总结:

安装依赖诗词添加熊猫jupyter ipykernel 在虚拟环境中用诗壳打开一个壳 打开Jupyter notebook，访问与Jupyter notebook相关的所有虚拟环境

还有干净的Conda工作流。注意这个帖子里的很多答案——它们会让你走上一条会给你带来很多痛苦和折磨的道路。

如果你使用的是anaconda发行版，这对我来说是有效的(在macintosh上):

$ conda create -n py3k python=3 anaconda

$ source activate py3k

$ ipython kernelspec install-self

最后一个命令需要注意:

(py3k)Monas-MacBook-Pro:cs799 mona$ ipython kernelspec install-self
[TerminalIPythonApp] WARNING | Subcommand `ipython kernelspec` is deprecated and will be removed in future versions.
[TerminalIPythonApp] WARNING | You likely want to use `jupyter kernelspec` in the future
[InstallNativeKernelSpec] WARNING | `jupyter kernelspec install-self` is DEPRECATED as of 4.0. You probably want `ipython kernel install` to install the IPython kernelspec.
[InstallNativeKernelSpec] Installed kernelspec python3 in /usr/local/share/jupyter/kernels/python3
(py3k)Monas-MacBook-Pro:cs799 mona$ ipython kernel install 
Installed kernelspec python3 in /usr/local/share/jupyter/kernels/python3

按照上述步骤在OSX Yosemite中进行测试，并输入jupter notebook并在浏览器中创建一个新的notebook，您将看到以下截图:

在Ubuntu 14.04上，我不得不使用之前答案的组合。

首先，安装pip3 安装python-pip3

然后用pip3安装jupyter Pip3安装jupyter

然后使用ipython3安装内核 Ipython3内核安装

在ElementaryOS Freya(基于Ubuntu 14.04)上，其他答案都没有立即对我起作用;我得到了

[TerminalIPythonApp]警告|文件不存在:'kernelspec'

quickbug在Matt的回答中描述的错误。我首先要做的是:

Sudo apt-get安装pip3

安装ipython[所有]

这时你就可以运行Matt建议的命令了;即:ipython kernelspec install-self和ipython3 kernelspec install-self

现在，当我启动ipython notebook，然后打开一个notebook时，我能够从kernel菜单中选择Python 3内核。

将多个内核安装到单个虚拟环境(venv)

这些答案中的大多数(如果不是全部的话)假设您乐于在全局范围内安装包。这个答案适合你，如果你:

使用*NIX机器 不喜欢全局安装包 不要使用anaconda <->，你很乐意从命令行运行jupyter服务器 想要知道内核安装“在哪里”。

(注意:这个答案在python3-jupyter安装中添加了一个python2内核，但在概念上很容易交换。)

Prerequisites You're in the dir from which you'll run the jupyter server and save files python2 is installed on your machine python3 is installed on your machine virtualenv is installed on your machine Create a python3 venv and install jupyter Create a fresh python3 venv: python3 -m venv .venv Activate the venv: . .venv/bin/activate Install jupyterlab: pip install jupyterlab. This will create locally all the essential infrastructure for running notebooks. Note: by installing jupyterlab here, you also generate default 'kernel specs' (see below) in $PWD/.venv/share/jupyter/kernels/python3/. If you want to install and run jupyter elsewhere, and only use this venv for organizing all your kernels, then you only need: pip install ipykernel You can now run jupyter lab with jupyter lab (and go to your browser to the url displayed in the console). So far, you'll only see one kernel option called 'Python 3'. (This name is determined by the display_name entry in your kernel.json file.) Add a python2 kernel Quit jupyter (or start another shell in the same dir): ctrl-c Deactivate your python3 venv: deactivate Create a new venv in the same dir for python2: virtualenv -p python2 .venv2 Activate your python2 venv: . .venv2/bin/activate Install the ipykernel module: pip install ipykernel. This will also generate default kernel specs for this python2 venv in .venv2/share/jupyter/kernels/python2 Export these kernel specs to your python3 venv: python -m ipykernel install --prefix=$PWD/.venv. This basically just copies the dir $PWD/.venv2/share/jupyter/kernels/python2 to $PWD/.venv/share/jupyter/kernels/ Switch back to your python3 venv and/or rerun/re-examine your jupyter server: deactivate; . .venv/bin/activate; jupyter lab. If all went well, you'll see a Python 2 option in your list of kernels. You can test that they're running real python2/python3 interpreters by their handling of a simple print 'Hellow world' vs print('Hellow world') command. Note: you don't need to create a separate venv for python2 if you're happy to install ipykernel and reference the python2-kernel specs from a global space, but I prefer having all of my dependencies in one local dir

博士TL;

Optionally install an R kernel. This is instructive to develop a sense of what a kernel 'is'. From the same dir, install the R IRkernel package: R -e "install.packages('IRkernel',repos='https://cran.mtu.edu/')". (This will install to your standard R-packages location; for home-brewed-installed R on a Mac, this will look like /usr/local/Cellar/r/3.5.2_2/lib/R/library/IRkernel.) The IRkernel package comes with a function to export its kernel specs, so run: R -e "IRkernel::installspec(prefix=paste(getwd(),'/.venv',sep=''))". If you now look in $PWD/.venv/share/jupyter/kernels/ you'll find an ir directory with kernel.json file that looks something like this:

{
  "argv": ["/usr/local/Cellar/r/3.5.2_2/lib/R/bin/R", "--slave", "-e", "IRkernel::main()", "--args", "{connection_file}"],
  "display_name": "R",
  "language": "R"
}

总之，内核只是从内核调用特定于语言的可执行文件。Json文件，jupyter在…/share/jupyter/kernels目录和列表在本例中，调用R来运行IRkernel::main()函数，该函数将向Jupiter服务器来回发送消息。类似地，python2内核只是使用ipykernel_launcher模块调用python2解释器，如.venv/share/jupyter/kernels/python2/kernel中所示。json等。

如果您想一下子运行所有这些指令，这里有一个脚本。

解决方案在官方文档中有详细的说明:https://ipython.readthedocs.org/en/latest/install/kernel_install.html

我尝试了第一种方法。因为我已经安装了ipykernel，简单地运行python3 -m ipykernel install——user就解决了这个问题。