Skip to content

Calling sxs from Julia#

Julia makes it extraordinarily simple to call Python packages from within Julia.

Note

This page describes calling the sxs Python package from Julia. Instructions for calling Julia components from Python are found elsewhere.

Optional preliminary step: A new environment#

Julia also makes it much easier to work with "environments" than Python does. If you want to create one from working with sxs in Julia, just create a new directory, cd into it, and run 

julia --project=.
This will create a file named Project.toml in that directory containing this new environment. Every time you need to work with this environment, you can tell Julia to use it by adding --project= followed by the path to that directory. Alternatively, if Julia is already running, just hit ] to enter the package manager, and run 
activate /path/to/new/directory

Getting Julia working with Python#

Once Julia is running (with the project you want), just run 

import Pkg
Pkg.add(["PythonCall", "CondaPkg"])
using PythonCall
import CondaPkg
CondaPkg.add("numba", channel="numba")
CondaPkg.add("llvmlite", channel="numba")
CondaPkg.add("sxs", channel="conda-forge")
Once everything is installed, the main difference from Python is that you have to import the module as 
sxs = pyimport("sxs")
Now, you can call functions from the sxs package nearly identically to how they are called in Python. For example, the simple call to load a waveform from the README of this package is 
waveform = sxs.load("SXS:BBH:0123/Lev/rhOverM", extrapolation_order=2)
That exact line works just in Julia like it does in Python. (Note that strings must use double quotes in Julia, and Python's True and False are lowercased in Julia.) And accessing members and methods of Python objects is just as natural. For example, waveform.t and waveform.data return the same data as in Python.

See the PythonCall documentation for more details on how to call Python from Julia, and any conversions that may need to be done. Note that pyconvert(Any, x) will usually convert x to its most natural Julia equivalent.