NERSC Python Installation - LSSTDESC/ComputingInfrastructure GitHub Wiki
DM stack
Install miniconda based on the version installed via newinstall.sh
bash Miniconda3-4.5.12-Linux-x86_64.sh -b -p /global/common/software/lsst/common/miniconda/py3-4.5.12
Using the conda list available in the scipipe_conda_env repo
Additional Items
- pip install camb
- pip install fast3tree
- pip install fitsio
- pip install healpy
- pip install https://bitbucket.org/yymao/helpers/get/v0.3.2.tar.gz (v0.3.1 for Run1.2 releases)
- conda install markupsafe
- conda install nose
- pip install parsl==0.7.2
- pip install pymssql
-
- export PYMSSQL_BUILD_WITH_BUNDLED_FREETDS=1 (avoid conda-forge) http://www.pymssql.org/en/stable/freetds.html
- conda install scikit-learn
- pip install tables
- pip install TreeCorr==4.0.1 (3.3.7 DC2 Run1.2 env)
- pip install https://github.com/LSSTDESC/descqa/archive/v2.0.0-0.7.0.tar.gz
- pip install https://github.com/LSSTDESC/gcr-catalogs/archive/v0.14.4.tar.gz
- pip install GCR==0.8.8
- pip install corner (ref: http://corner.readthedocs.io/en/latest/install.html)
- pip install pyccl==1.0.0 (make sure gcc for the stack is enabled)
- pip install https://github.com/yymao/FoFCatalogMatching/archive/v0.1.0.tar.gz
- pip install https://github.com/msimet/Stile/archive/v0.1.tar.gz (currently 0.1)
- https://github.com/LSSTDESC/CatalogMatcher python setup.py install
- pip install scikit-image
- pip install https://github.com/LSSTDESC/desc-dc2-dm-data/archive/v0.3.0.tar.gz
- pip install emcee
- pip install psycopg2== 2.8.2 (latest) 2.7.5 Run1.2 PostgreSQL API, stick with 2.7.5 to support NERSC version of PostgreSQL)
- pip install extinction (for sncosmo) https://extinction.readthedocs.io/en/latest/
- pip install seaborn https://seaborn.pydata.org/installing.html (for example notebooks)
Note we also need ipython and jupyter, but those are installed as dependencies of parsl
MWV request:
Here are pip-installed Python libraries that are being used in the data access and visualization:
bokeh dask dask.distributed datashader fastparquet google-cloud-bigquery holoviews pyarrow ipywidgets
It would be nice, but not necessary to install fastparquet
and pyarrow
with snappy compression.
Environments
- root
- stack - test environment which is a clone of the root used to test installation of new packages
- parsl-latest - clone copy of root with latest parsl release
- clmassmod - test env for the clmassmod group