-
Notifications
You must be signed in to change notification settings - Fork 83
Getting started
Easiest way to get a good base Python installation is to install the Enthought Python Distribution (free for academics!). You will need to sign up for account using your university email and then wait for two emails. The first email contains an account verification link. The second email contains detailed instructions about how to install and update you EPD. The rub: visit the EPD repository page (making sure that you are logged in to your Enthought account!) and click the Repository Login button. This should take you to a list of EPD installers. Select either the 32 or 64-bit versions depending on your system.
Once you have downloaded and installed the EPD, open a command prompt (i.e., terminal window) and type python at the prompt. If the installation has been successful, you should see some information about the version of the EPD that you are running, the version of Python that you are running, and be taken to a Python prompt: >>>. Quit Python by typing control-z, and then type ipython at the prompt. You should be taken to the IPython prompt and given some information about the version of IPython that you are using. If you do not have version 0.13.1, the you will want to update.
To update to the most recent version of IPython, first quit IPython by typing either quit or control-z. Then at the command prompt type:
enpkg ipython
This will use the Enthought package installer to download and build the most recent version of IPython (currently 0.13.1). Do the same for Pandas:
enpkg pandas
should update you to the most recent version of Pandas. Still need to update versions of NumPy and SciPy and install PyMacLab. Before we can do that we need to download distribute and then pip. Instructions on obtaining and installing distribute and pip can be found on this blog (ignore the stuff about powershell and vitualenv for now). More info on how to install distribute and pip.
Currently most recent version of NumPy available using enpkg is 1.6.2. Most recent, stable version of NumPy is 1.7.0.
Currently most recent version of SciPy available using enpkg is 0.10.1. Most recent, stable version of SciPy is 0.11.0.
Unless you have a MatLab license, you will need to install octave in order use Dynare (you can still use Dynare++ from the command line to solve models, but estimation capabilities are limited to non-existent in Dynare++). The versions of Octave that are compatible with Dynare can be downloaded from the Dynare website.
In order to run Octave via the the command prompt each session, so your computer can find it within its know list of applications, you need to add the folder containing the .exe Octave application files to the system path. [To add a directory to the computer's system path in Windows] (http://www.mathworks.co.uk/support/solutions/en/data/1-15ZLK/index.html) simply...
- Open "My Computer"
- Click on "System Properties"
- Open Advanced system settings"
- In the "Advanced" tab, select "Environment Variables", and choose to edit "Path" listed under "System variables".
- At the end of this path, insert a semi-colon followed by the directory where Octave executables are saved. Default seems to be:
;c:\octave\3.6.2_gcc4.6.2_20120609\bin. - Close all open command prompts and open a new terminal window. Type 'octave' at the prompt. If Octave loads, then congratulations! If not...check to make sure you followed steps 1-6 exactly (i.e., no typos, file path to Octave is correct, etc.)
Download and install the most recent stable release of Dynare (which will include Dynare++ as well!) from the Dynare website.
To add Dynare to Octave PATH in Windows, find the following directory:
C:\Octave\3.6.2_gcc4.6.2_20120609\share\octave\site\m\startup
and open the file octaverc using a text editor. Insert the line addpath c:\dynare\4.3.2\matlab; after the last line starting EXEC_Path. Save and overwrite existing octaverc file.
Although the above seems to work, I am suspicious because our solution differs substantially from the official instructions from Dynare on setting the Octave path. I was able to find the mentioned .octaverc file without issue on my Mac. The official octave user guide claims that you can save your path variable as follows. Start Octave and type the following at the prompt:
addpath c:\dynare\4.3.2\matlab
followed by:
savepath()
Omitting the filename means that the path variable will be saved to ~/.octaverc. This should do the trick and might be superior...or might not!
Once you have installed both Octave and Dynare and set the above path variables correctly, then you will want to check that you Dynare installation is working. Easiest way to check the installation is to run one of the example model files included with Dynare by following these instructions.
Will want to have a proper text editor (Notepad is not an option!). Is Emacs the solution? Maybe Notepad++ would be better. Perhaps an IDE like Spyder?
Version control is an important component of doing reproducible, computational research. Git is an excellent, distributed (i.e., good for collaborative research!) version control system. You can download both the bash shell version and a Windows GUI version from the Git downloads website.