tools_wrapped

Installing the tools wrapped by 4GP

Depending what tasks you envisage doing with 4GP, you will almost certainly need to install some of the software packages it provides wrappers for - for example, the Cannon, Turbospectrum or 4FS.

Note that you do not need to install tools that you are not going to use.

Installation paths

When you install these tools, 4GP needs to be able to find them.

By far the simplest approach is if you install each tool in the same directory where you keep your working copies of 4most-4gp and 4most-4gp-scripts. Thus, after installing various tools, this directory might look as follows:

dcf21@astrolabe:~/iwg7_pipeline$ ls
4most-4gp                  downloads         idl_packages    rvspecfit
4most-4gp-scripts          forwardModelling  interpol_marcs  sme
4most-iwg7-pipeline-tests  fromBengt         OpSys           TheCannon
4MOST_testspectra          fromKeith         pepsi           turbospectrum-15.1
AnniesLasso                hot_stars         pyphot          virtualenv

This is the default place where 4GP looks to find each tool, and if they are installed in this way then you will not need to explicitly tell it where to look.

You are permitted to install these tools in a different location, but then you will need to pass configuration parameters to a number of 4GP's modules (usually called binary_path) to tell it where to find each tool.

Installing the Cannon

At the time of writing, there are various branches of the Cannon (also known as Annie's Lasso), and each has its own different API. To avoid confusion, we have our own 4GP GitHub repository containing the version of the Cannon that we use. This is a fork of a version released by Andy Casey in 2007.

Within this repository, there are branches named after each release of 4GP, e.g. release-2019-03-01-1. These contain the versions of the Cannon expected by each release of 4GP with the same name. The master branch always represents the most recent stable release.

git clone https://github.com/dcf21/AnniesLasso.git
cd AnniesLasso
python setup.py install

Do not use the dev branch in this repository. It contains the latest release of the Cannon by Andy Casey, which has some bugs which are unresolved at the time of writing. It produces worse results than before.

Installing pyphot

The pyphot source code can be obtained from a GitHub repository. As with the Cannon, we maintain our own fork of the repository with branches labelled release-2018-01-12-1, etc, to indicate versions which are compatible with each release of 4GP.

git clone https://github.com/dcf21/pyphot.git
cd pyphot
python setup.py install

Installing Turbospectrum

If you want to synthesize spectra using Turbospectrum, the following commands will download and install it for you.

Download the code, as follows:

# Fetch the Turbospectrum source code
wget http://www.pages-perso-bertrand-plez.univ-montp2.fr/DATA/Turbospectrum-v15.1.tar.gz

# Fetch the interpol source code, which is used to interpolate MARCS models
wget http://marcs.astro.uu.se/documents/auxiliary/interpol_marcs.tar.gz

Proceed to build the tools as follows:

# Unpack Turbospectrum and build it from source
tar xvfz Turbospectrum-v15.1.tar.gz
mv EXPORT-15.1 turbospectrum-15.1
cd turbospectrum-15.1/exec-gf-v15.1/
make

# Unpack interpol and build it from source
cd ../..
tar xvfz interpol_marcs.tar.gz
cd interpol_marcs
gfortran interpol_modeles.f -o interpol_modeles

For more information, read the README.md file in the fourgp_specsynth directory.

You will also need to obtain a copy of the MARCS grid of model atmospheres, and a suitable line list for the 4MOST spectral range. Your best bet is to ask us to supply you with this data, which is stored in the following directory on a computer in Lund:

rsync -av dcf21@astrolabe.astro.lu.se:iwg7_pipeline/fromBengt .

Installing 4FS

If you want to use 4FS, build it as follows:

git checkout https://dcf21@gitlab.4most.eu/tdwelly/OpSys.git
sudo apt-get install libhealpix-cxx-dev libchealpix-dev libcfitsio3-dev python-healpy
cd OpSys
make

Note that the 4MOST GitLab account is password protected, so you will need to get your own account before you will be able to check out the code.