开源量化框架Catalyst中文教程(2) -- 安装 官网教材

Install

To get started with Catalyst, you will need to install it in your computer. Like any other piece of software, Catalyst has a number of dependencies (other software on which it depends to run) that you will need to install, as well. We recommend using a software named Conda that will manage all these dependencies for you, and set up the environment needed to get you up and running as easily as possible. This is the recommended installation method for Windows, MacOS and Linux. See Installing with Conda.

What conda does is create a pre-configured environment, and inside that environment install Catalyst using pip, Python’s package manager. Thus, as an alternative installation method for MacOS and Linux, you can install Catalyst directly with pip (we recommend in combination with a virtual environment). See Installing with pip.

Alternatively you can install Catalyst using pipenv which is a mix of pip and virtualenv. See Installing with pipenv.

Regardless of the method, each operating system (OS), has its own prerequisites, make sure to review the corresponding sections for your system: Linux, MacOS and Windows.

Installing with conda

The preferred method to install Catalyst is via the conda package manager, which comes as part of Continuum Analytics’ Anaconda distribution.

The primary advantage of using Conda over pip is that conda natively understands the complex binary dependencies of packages like numpy and scipy. This means that conda can install Catalyst and its dependencies without requiring the use of a second tool to acquire Catalyst’s non-Python dependencies.

For Windows, you will first need to install the Microsoft Visual C++ Compiler for Python. Follow the instructions on the Windows section and come back here.

For instructions on how to install conda, see the Conda Installation Documentation. Alternatively, you can install MiniConda, which is a smaller footprint (fewer packages and smaller size) than its big brother Anaconda, but it still contains all the main packages needed. To install MiniConda, you can follow these steps:

1. Download MiniConda. Select either Python 3.6 (recommended) or Python 2.7 for your Operating System. The Enigma Data Marketplace will require Python3, that’s why we are recommending to opt for the newer version.
2. Install MiniConda. See the Installation Instructions if you need help.
3. Ensure the correct installation by running conda list in a Terminal window, which should print the list of packages installed with Conda.

For Windows, if you accepted the default installation options, you didn’t check an option to add Conda to the PATH, so trying to run conda from a regular Command Prompt will result in the following error: 'conda' is no recognized as an internal or external command, operatble program or batch file. That’s to be expected. You will nee to launch an Anaconda Prompt that was added at installation time to your list of programs available from the Start menu.

Once either Conda or MiniConda has been set up you can install Catalyst:

1. Download the proper .yml file matching your Conda installation from step #1 above. To download, simply click on the ‘Raw’ button and save the file locally to a folder you can remember. Make sure that the file gets saved with the .yml extension, and nothing like a .txt file or anything else.

   Linux or MacOS: Download the file python3.6-environment.yml (recommended) or python2.7-environment.yml

   Windows: Download the file python3.6-environment-windows.yml (recommended) or python2.7-environment.yml

2. Open a Terminal window and enter [cd/dir] into the directory where you saved the above .yml file.

3. Install using this file. This step can take about 5-10 minutes to install.

   Linux or MacOS Python 3.6:

   conda env create -f python3.6-environment.yml

   Linux or MacOS Python 2.7:

   conda env create -f python2.7-environment.yml

   Windows Python 3.6:

   conda env create -f python3.6-environment-windows.yml

   Windows Python 2.7:

   conda env create -f python2.7-environment.yml

4. Activate the environment (which you need to do every time you start a new session to run Catalyst):

   Linux or MacOS:

   source activate catalyst

   Windows:

   activate catalyst

5. Verify that Catalyst is install correctly:

   catalyst --version

   which should display the current version.

Congratulations! You now have Catalyst installed.

Troubleshooting conda Install

If the command conda env create -f python2.7-environment.yml in step 3 above failed for any reason, you can try setting up the environment manually with the following steps:

1. If the above installation failed, and you have a partially set up catalyst environment, remove it first. If you are starting from scratch, proceed to step #2:

   conda env remove --name catalyst

2. Create the environment:

   for python 2.7:

   conda create --name catalyst python=2.7 scipy zlib

or for python 3.6:

conda create --name catalyst python=3.6 scipy zlib

1. Activate the environment:

   Linux or MacOS:

   source activate catalyst

   Windows:

   activate catalyst

2. Install the Catalyst inside the environment:

   pip install enigma-catalyst matplotlib

3. Verify that Catalyst is installed correctly:

   catalyst --version

   which should display the current version.

Congratulations! You now have Catalyst properly installed.

Installing with pip

Installing Catalyst via pip is slightly more involved than the average Python package.

There are two reasons for the additional complexity:

1. Catalyst ships several C extensions that require access to the CPython C API. In order to build the C extensions, pip needs access to the CPython header files for your Python installation.
2. Catalyst depends on numpy, the core library for numerical array computing in Python. Numpy depends on having the LAPACK linear algebra routines available.

Because LAPACK and the CPython headers are non-Python dependencies, the correctway to install them varies from platform to platform. If you’d rather use a single tool to install Python and non-Python dependencies, or if you’re already using Anaconda as your Python distribution, refer to the Installing with Conda section.

If you use Python for anything other than Catalyst, we strongly recommend that you install in a virtualenv. The Hitchhiker’s Guide to Python provides an excellent tutorial on virtualenv. Here’s a summarized version:

$ pip install virtualenv
$ virtualenv catalyst-venv
$ source ./catalyst-venv/bin/activate

Once you’ve installed the necessary additional dependencies for your system (GNU/Linux Requirements, MacOS Requirements or Windows Requirements) and have activated your virtualenv, you should be able to simply run

$ pip install enigma-catalyst matplotlib

Note that in the command above we install two different packages. The second one, matplotlib is a visualization library. While it’s not strictly required to run catalyst simulations or live trading, it comes in very handy to visualize the performance of your algorithms, and for this reason we recommend you install it, as well.

Troubleshooting pip Install

Issue:

Package enigma-catalyst cannot be found

Solution:

Make sure you have the most up-to-date version of pip installed, by running:

$ pip install --upgrade pip

On Windows, the recommended command is:

$ python -m pip install --upgrade pip

---

Issue:

Package enigma-catalyst cannot still be found, even after upgrading pip (see above), with an error similar to:

Downloading/unpacking enigma-catalyst
Could not find a version that satisfies the requirement enigma-catalyst
(from versions: 0.1.dev9, 0.2.dev2, 0.1.dev4, 0.1.dev5, 0.1.dev3,
0.2.dev1, 0.1.dev8, 0.1.dev6)
Cleaning up...
No distributions matching the version for enigma-catalyst

Solution:

In some systems (this error has been reported in Ubuntu), pip is configured to only find stable versions by default. Since Catalyst is in alpha version, pip cannot find a matching version that satisfies the installation requirements. The solution is to include the –pre flag to include pre-release and development versions:

$ pip install --pre enigma-catalyst

---

Issue:

Package enigma-catalyst fails to install because of outdated setuptools

Solution:

Upgrade to the most up-to-date setuptools package by running:

$ pip install --upgrade pip setuptools

---

Issue:

Missing required packages

Solution:

Download requirements.txt (click on the Raw button and Right click -> Save As…) and use it to install all the required dependencies by running:

$ pip install -r requirements.txt

---

Issue:

Installation fails with error: fatal error: Python.h: No such file or directory

Solution:

Some systems (this issue has been reported in Ubuntu) require python-dev for the proper build and installation of package dependencies. The solution is to install python-dev, which is independent of the virtual environment. In Ubuntu, you would need to run:

$ sudo apt-get install python-dev

---

Issue:

Missing TA_Lib

Solution:

Follow these instructions to install the TA_Lib Python wrapper (and if needed, its underlying C library as well).

Installing with pipenv

Installing Catalyst via pipenv is perhaps easier that installing it via pip itself but you need to install pipenv first via pip.

$ pip install pipenv

Once pipenv is installed you can proceed by creating a project folder and installing Catalyst on that project automagically as follows:

$ mkdir project
$ cd project
$ pipenv --two
$ pipenv install enigma-catalyst matplotlib

Until now the workflow compared to pip is almost identical, the difference is that you don’t need to load manually any virtualenv however you need to use the pipenv run prefix to run the catalyst command as follows:

$ pipenv run catalyst --version

If you want to know more about pipenv go to the pipenv github repo

GNU/Linux Requirements

On Debian-derived Linux distributions, you can acquire all the necessary binary dependencies from apt by running:

$ sudo apt-get install libatlas-base-dev python-dev gfortran pkg-config libfreetype6-dev

On recent RHEL-derived derived Linux distributions (e.g. Fedora), the following should be sufficient to acquire the necessary additional dependencies:

$ sudo dnf install atlas-devel gcc-c++ gcc-gfortran libgfortran python-devel redhat-rep-config

On Arch Linux, you can acquire the additional dependencies via pacman:

$ pacman -S lapack gcc gcc-fortran pkg-config

Amazon Linux AMI Notes

The packages pip and setuptools that come shipped by default are very outdated. Thus, you first need to run:

$ pip install --upgrade pip setuptools

The default installation is also missing the C and C++ compilers, which you install by:

$ sudo yum install gcc gcc-c++

Then you should follow the regular installation instructions outlined at the beginning of this page.

MacOS Requirements

The version of Python shipped with MacOS by default is generally out of date, and has a number of quirks because it’s used directly by the operating system. For these reasons, many developers choose to install and use a separate Python installation. The Hitchhiker’s Guide to Python provides an excellent guide to Installing Python on MacOS, which explains how to install Python with the Homebrew manager.

Assuming you’ve installed Python with Homebrew, you’ll also likely need the following brew packages:

$ brew install freetype pkg-config gcc openssl

MacOS + virtualenv/conda + matplotlib

The first time that you try to run an algorithm that loads the matplotlib library, you may get the following error:

RuntimeError: Python is not installed as a framework. The Mac OS X backend
will not be able to function correctly if Python is not installed as a
framework. See the Python documentation for more information on installing
Python as a framework on Mac OS X. Please either reinstall Python as a
framework, or try one of the other backends. If you are using (Ana)Conda
please install python.app and replace the use of 'python' with 'pythonw'.
See 'Working with Matplotlib on OSX' in the Matplotlib FAQ for more
information.

This is a matplotlib-specific error, that will go away once you run the following command:

$ echo "backend: TkAgg" > ~/.matplotlib/matplotlibrc

in order to override the default MacOS backend for your system, which may not be accessible from inside the virtual or conda environment. This will allow Catalyst to open matplotlib charts from within a virtual environment, which is useful for displaying the performance of your backtests. To learn more about matplotlib backends, please refer to the matplotlib backend documentation.

Windows Requirements

In Windows, you will first need to install the Microsoft Visual C++ Compiler, which is different depending on the version of Python that you plan to use:

• Python 3.5, 3.6: Visual C++ 2015 Build Tools, which installs Visual C++ version 14.0. This is the recommended version
• Python 2.7: Microsoft Visual C++ Compiler for Python 2.7, which installs version Visual C++ version 9.0

This package contains the compiler and the set of system headers necessary for producing binary wheels for Python packages. If it’s not already in your system, download it and install it before proceeding to the next step. If you need additional help, or are looking for other versions of Visual C++ for Windows (only advanced users), follow this link.

Once you have the above compiler installed, the easiest and best supported way to install Catalyst in Windows is to use Conda. If you didn’t any problems installing the compiler, jump to the Conda section, otherwise keep on reading to troubleshoot the C++ compiler installtion.

Some problems we have encountered installing the Visual C++ Compiler mentioned above are as follows:

• The system administrator has set policies to prevent this installation.

  In some systems, there is a default Windows Software Restriction policy that prevents the installation of some software packages like this one. You’ll have to change the Registry to circumvent this:

  • Click Start, and search for regedit and launch the Registry Editor
  • Navigate to the following folder: HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows\Installer
  • If the last folder does not exist, create it by right-clicking on the parent folder and choosing -> New -> Key and typing Installer
  • If there is an entry for DisableMSI, set the Value data to 0.
  • If there is no such entry, click on the Edit menu -> New -> DWORD (32-bit) Value and enter DisableMSI as the Name (and by default you get 0 as the Value Data)

 

• The installer has encountered an unexpected error installing this package. This may indicate a problem with this package. The error code is 2503.

  We have observed this when trying to install a package without enough administrator permissions. Even when you are logged in as an Administrator, you have to explictily install this package with administrator privileges:

  • Click Start and find CMD or Command Prompt
  • Right click on it and choose Run as administrator
  • cd into the folder where you downloaded VCForPython27.msi
  • Run msiexec /i VCForPython27.msi

Updating Catalyst

Catalyst is currently in alpha and in under very active development. We release new minor versions every few days in response to the thorough battle testing that our user community puts Catalyst in. As a result, you should expect to update Catalyst frequently. Once installed, Catalyst can easily be updated as a pip package regardless of the environment used for installation. Make sure you activate your environment first as you did in your first install, and then execute:

$ pip uninstall enigma-catalyst
$ pip install enigma-catalyst

Alternatively, you could update Catalyst issuing the following command:

$ pip install -U enigma-catalyst

but this command will also upgrade all the Catalyst dependencies to the latest versions available, and may have unexpected side effects if a newer version of a dependency inadvertently breaks some functionality that Catalyst relies on. Thus, the first method is the recommended one.

Getting Help

If after following the instructions above, and going through the Troubleshooting sections, you still experience problems installing Catalyst, you can seek additional help through the following channels:

• Join our Catalyst Forum, and browse a variety of topics and conversations around common issues that others face when using Catalyst, and how to resolve them. And join the conversation!
• Join our Discord community, and head over the #catalyst_dev channel where many other users (as well as the project developers) hang out, and can assist you with your particular issue. The more descriptive and the more information you can provide, the easiest will be for others to help you out.
• Report the problem you are experiencing on our GitHub repository following the guidelines provided therein. Before you do so, take a moment to browse through all previous reported issues in the likely case that someone else experienced that same issue before, and you get a hint on how to solve it.

https://www.enigma.co/catalyst/install.html