Getting Started

The best way to learn about Gammapy is to read and play with the examples in the Gammapy Tutorial notebooks.

This section explains the steps to get set up for the Gammapy tutorials on your machine:

  1. Install Anaconda and the Gammapy environment
  2. Download tutorial notebooks and example datasets
  3. Check your setup and Gammapy environment
  4. Use Gammapy with Python, IPython or Jupyter

If you have used conda, Python, IPython and Jupyter before, you can just skim this page, and quickly copy & paste the commands to get set up.

Help!?

If you have any questions or issues, please ask for help on the Gammapy Slack, mailing list or on Github (whatever is easiest for you, see Gammapy contact)

Install

To install Gammapy we recommend that you install the Anaconda distribution from https://www.anaconda.com/download/. The following instructions and commands below need to have Anaconda installed.

Anaconda it’s free, works on Linux, MacOS and Windows. It installs in your home directory, no system privileges needed, and you can just delete it without problems if you don’t need or want it any more.

Besides the software included in the Anaconda distribution, it gives you the conda tool, which is a package and environment manager. We will use it to create a dedicated environment for the latest stable Gammapy version and a known good set of Gammapy dependencies (e.g. Python, Numpy and Astropy).

Once Anaconda has been installed, use the following commands to install and activate the gammapy-0.8 conda environment:

curl -O https://gammapy.org/download/install/gammapy-0.8-environment.yml
conda env create -f gammapy-0.8-environment.yml
conda activate gammapy-0.8

Congratulations! You are all set to start using Gammapy!

Note

Every time you open a new terminal window, you will have to activate the Gammapy conda environment again before you can use it.

conda activate gammapy-0.8

Download tutorials

You can now proceed to download the Gammapy tutorial notebooks and the example datasets used there (at the moment from CTA, H.E.S.S. and Fermi-LAT). The total size to download is about 100 MB.

gammapy download tutorials
cd gammapy-tutorials
export GAMMAPY_DATA=$PWD/datasets

You might want to put the definition of the $GAMMAPY_DATA environment variable in your shell profile setup file that is executed when you open a new terminal (for example $HOME/.bash_profile).

The datasets are curated and stable, the notebooks are still under development just like Gammapy itself, and thus stored in a sub-folder that contains the Gammapy version number.

The gammapy download command and versioning of the notebooks is new. If there are issues, note that you can just delete the folder any time using rm -r gammapy-tutorials and start over.

Also note that it’s of course possible to download just the notebooks, just the datasets, a single notebook or dataset, or a specific tutorial with the datasets it uses. See the help:

gammapy download --help

Check your setup

You might want to display some info about Gammapy installed. You can execute the following command, and it should print detailed information about your installation to the terminal:

gammapy info

If there is some issue, the following commands could help you to figure out your setup:

conda info
which python
which ipython
which jupyter
which gammapy
env | grep PATH
python -c 'import gammapy; print(gammapy); print(gammapy.__version__)'

You can also use the following commands to check which conda environment is active and which ones you have set up:

conda info
conda env list

If you’re new to conda, you could also print out the conda cheat sheet, which lists the common commands to install packages and work with environments.

Use Gammapy

Python

Gammapy is a Python package, so you can of course import and use it from Python:

$ python
Python 3.6.0 | packaged by conda-forge | (default, Feb 10 2017, 07:08:35)
[GCC 4.2.1 Compatible Apple LLVM 7.3.0 (clang-703.0.31)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from gammapy.stats import significance
>>> significance(n_on=10, mu_bkg=4.2, method='lima')
array([2.39791813])

IPython

IPython is nicer to use for interactive analysis:

$ ipython
Python 3.6.0 | packaged by conda-forge | (default, Feb 10 2017, 07:08:35)
Type 'copyright', 'credits' or 'license' for more information
IPython 6.5.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: from gammapy.stats import significance

In [2]: significance(n_on=10, mu_bkg=4.2, method='lima')
Out[2]: array([2.39791813])

For example you can use ? to look up help for any Gammapy function, class or method from IPython:

In [3]: significance?

Of course, you can also use the Gammapy online docs if you prefer, clicking in links (i.e. gammapy.stats.significance) or using search docs field in the upper left.

As an example, here’s how you can create gammapy.data.DataStore and gammapy.data.EventList objects and start exploring H.E.S.S. data:

>>> from gammapy.data import DataStore
>>> data_store = DataStore.from_dir('$GAMMAPY_DATA/hess-dl3-dr1/')
>>> events = data_store.obs(obs_id=23523).events
>>> print(events)
EventList info:
- Number of events: 7613
- Median energy: 0.953 TeV
- OBS_ID = 23523
>>> events.energy.mean()
<Quantity 4.418008 TeV>

Try to make your first plot using the peek helper method:

>>> events.peek()
>>> plt.savefig("events.png")

Python script

Another common way to use Gammapy is to write a Python script. Try it and put the following code into a file called example.py:

"""Example Python script using Gammapy"""
from gammapy.data import DataStore
data_store = DataStore.from_dir('$GAMMAPY_DATA/hess-dl3-dr1/')
events = data_store.obs(obs_id=23523).events
print(events.energy.mean())

You can run it with Python:

$ python example.py
4.418007850646973 TeV

If you want to continue with interactive data or results analysis after running some Python code, use IPython like this:

$ ipython -i example.py

Command line

As you have already seen, installing Gammapy gives you a gammapy command line tool with subcommands such as gammapy info or gammapy download.

We plan to add a high-level interface to Gammapy soon that will let you run Gammapy analyses via the command line interface, probably driven by a config file. This is not available yet, for now you have to use Gammapy as a Python package.

Jupyter notebooks

To learn more about Gammapy, and also for interactive data analysis in general, we recommend you use Jupyter notebooks. Assuming you have followed the steps above to install Gammapy and activate the conda environment, you can start JupyterLab like this:

$ jupyter lab

This should open up JupyterLab app in your web browser, where you can create new Jupyter notebooks or open up existing ones. If you have downloaded the tutorials with gammapy download tutorials, you can browse your gammapy-tutorials folder with Jupyterlab and execute them there.

If you haven’t used Jupyter before, try typing print("Hello Jupyter") in the first input cell, and use the keyboard shortcut SHIFT + ENTER to execute it.

Install issues

If you have problems and think you might not be using the right Python or importing Gammapy isn’t working or giving you the right version, checking your Python executable and import path might help you find the issue:

import sys
print(sys.executable)
print(sys.path)

To check which Gammapy you are using you can use this:

import gammapy
print(gammapy)
print(gammapy.__version__)

Now you should be all set and to use Gammapy. Let’s move on to the Tutorial notebooks.