Skip to content

Latest commit

 

History

History
73 lines (60 loc) · 3.84 KB

INSTALL_ASU.md

File metadata and controls

73 lines (60 loc) · 3.84 KB
Raw

General

Start with the general Enterprise new user setup, and do the steps up to and including anaconda installation. When you create your environment, be sure to use python 3. Give an appropriate name to the conda environment. Below we use epic as the example.

After the installation of anaconda do the following

$ user=`whoami`
$ env=epic

$ conda create -n ${env} python=3.6

Bifrost is tested with python 3.6, we have successfully installed with python 3.8 and 3.9. Choose your favorite!

It is advised to install the dependencies of LSL before attempting a pip install. While pip generally is good about grabbing package dependencies, the configuration of some packages requires dependencies to be installed before pip can check if the dependencies are installed.

While not explicitly mentioned, lsl depends on astropy for fits I/O. Inclusion of astropy is not a mistake.

To install dependencies for lsl excute the following lines. This may not be necessay depending on your environment but will help smooth the install.

$ conda activate ${env}
$ conda config --add channels conda-forge
$ conda install aipy scipy numpy fftw astropy healpy ephem pytz matplotlib
$ pip install lsl

Bifrost

$ cd ~/src
$ git clone https://github.com/epic-astronomy/bifrost.git
$ cd bifrost
# At the time of this commit, the most up-to-date and stable branch is ibverb-support
$ git checkout ibverb-support

There are several other branches, so you may need to check out something else if you are testing/developing something specific.

Note that the README.md in bifrost can be (and as of this commit is) out of date. Use the following line to install dependencies. (Do this in your epic env, not with sudo as the bifrost README indicates). The ctypes gen commit referenced here works for Ubuntu >18. In the future this may be another point of trouble. If this version doesn't work, you will want to file and issue on the ledatelescope bifrost repo. Or figure out a fix and send a PR :)

$ pip install contextlib2 pint git+https://github.com/olsonse/ctypesgen.git@9bd2d249aa4011c6383a10890ec6f203d7b7990f

Now we need to actually install bifrost. First set up configuration for intrepid. Copy the file LWA_EPIC/config/ASU_user.mk from this repository to your bifrost directory:

$ cp ~/src/LWA_EPIC/config/ASU_user.mk ~/src/bifrost/user.mk

Before proceeding to compiling, we need to ensure your PATH is setup to find the cuda libraries. Add the following lines to your .bashrc (or relevant shell initialization script):

export PATH=/usr/local/cuda-11.4/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-11.4/lib64:$LD_LIBRARY_PATH

Then source the file and reactivate your environment:

$ source ~/.bashrc
$ conda activate ${env}

Finally, install (note this must be done on intrepid with your environment activated):

$ make -j 32
$ make install INSTALL_LIB_DIR="/home/${user}/src/anaconda/envs/${env}/lib" INSTALL_INC_DIR="/home/${user}/src/anaconda/envs/${env}/include" PYINSTALLFLAGS="--prefix=/home/${user}/src/anaconda/envs/${env}"

Testing multiple branches

If you want to install and test multiple bifrost branches, it is encouraged to use different conda environments. Check the environment variables of the install directory to properly link with the libraries.

Test Run

Remember to check that you are logged onto intrepid and in your environment.

$ cd ~/src/LWA_EPIC/LWA
$ python LWA_bifrost.py --offline --tbnfile=/data5/LWA_SV_data/data_raw/TBN/Jupiter/058161_000086727 --imagesize 64 --imageres 1.79057 --nts 512 --channels 4 --accumulate 50 --ints_per_file 40

This will generate a series of .npz files with example images. Have a look at them to be sure they look like the sky!