Python Tutorial

FormationTemps.jl can be called from Python using juliacall. This requires a working Julia installation (v1.12+) and Python 3.12+ with uv (recommended) or pip. We strongly advise against Conda.

Prerequisites

• Julia 1.12+: install from julialang.org or via juliaup.
• Python 3.12+ with uv (recommended) or pip.

Installation

From a Local Clone

If you have cloned the FormationTemps.jl repository:

cd FormationTemps.jl
uv sync
uv run python deps/setup.py

This installs FormationTemps.jl as a development package, meaning changes you make to the Julia source code are immediately reflected without reinstalling.

From the Julia Package Registry

If you want to use the released version of FormationTemps.jl without cloning the repository:

pip install juliacall juliapkg numpy matplotlib

Then set the following environment variables in your shell profile (e.g. ~/.bashrc, ~/.zshrc) so they persist across sessions:

export PYTHON_JULIAPKG_EXE=$(which julia)
export JULIA_NUM_THREADS=1

Finally, run the following once in Python to register the Julia dependency:

import juliapkg
juliapkg.require_julia("~1.12")
juliapkg.add("FormationTemps", "03bcd87b-2230-4045-a5fa-95a5fcdd1ff8", version="^1")
juliapkg.resolve()

This pulls the latest compatible version of FormationTemps.jl from the Julia General registry.

Basic Usage

Once installed, FormationTemps.jl can be loaded in Python via:

from juliacall import Main as jl
jl.seval("using Korg")
jl.seval("using FormationTemps")
FT = jl.FormationTemps
Korg = jl.Korg

A complete example computing a formation temperature spectrum (following the Basic Tutorial) is shown below:

import os
import matplotlib
# matplotlib.use("Agg")

from juliacall import Main as jl
import numpy as np
import matplotlib.pyplot as plt

# load FormationTemps
jl.seval("using Korg")
jl.seval("using FormationTemps")
FT = jl.FormationTemps
Korg = jl.Korg

# apply project style
plt.style.use(os.path.join(str(FT.moddir), "fig.mplstyle"))

# read the linelist (ships with the package)
# slicing done on Julia side to preserve 1-based indexing
linelist = jl.seval(
    'Korg.read_linelist(joinpath(FormationTemps.datdir, "Sun_VALD.lin"))[16000:16100]'
)

# set stellar parameters (velocities in m/s)
star = FT.StellarProps(Teff=5777.0, logg=4.44, Fe_H=0.0,
                       vsini=2100.0, v_macro=3400.0, v_micro=850.0)

# compute formation temperatures (use_gpu=False for portability)
result = FT.calc_formation_temp(star, linelist, use_gpu=False, convolve=True, u1=0.43, u2=0.31)

# extract results into numpy arrays
wavs = np.asarray(result.wavs)
flux = np.asarray(result.flux)
temps = np.asarray(result.form_temps)

# plot
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(9.6, 6.4), sharex=True)
ax1.plot(wavs, flux, c="k")
ax1.set_ylabel(r"{\rm Normalized Flux}")
ax2.plot(wavs, temps, c="k")
ax2.set_xlabel(r"{\rm Vacuum Wavelength [\AA]}")
ax2.set_ylabel(r"{\rm Formation Temperature [K]}")
fig.tight_layout()

# write out the file
import os
outpath = os.path.join(os.getcwd(), "formation_temps.png")
fig.savefig(outpath, dpi=150, bbox_inches="tight")
print(f"Saved {outpath}")
plt.close()

Troubleshooting

export PYTHON_JULIAPKG_EXE=$(which julia)

import ssl; print(ssl.OPENSSL_VERSION)

rm -rf .venv
uv sync --python /path/to/python3

export JULIA_NUM_THREADS=1

Tips