Installation

PyTransit can be installed from PyPI

pip install pytransit

using conda

conda install conda-forge::pytransit

or from source

git clone https://github.com/hpparvi/PyTransit.git
cd PyTransit
pip install .

Add -e to the last command for an editable install if you intend to modify the code.

Requirements

PyTransit needs Python 3 with NumPy, SciPy, pandas, Numba, Astropy, and Matplotlib. These are all installed automatically.

Optional dependencies

Two features depend on packages that are not installed automatically.

PyOpenCL enables the GPU models described in The OpenCL backend. It also needs a working OpenCL runtime for your device, which is installed separately and depends on your hardware and operating system.

pip install pyopencl

LDTk enables LDTkLDModel, which computes limb darkening profiles from stellar atmosphere models rather than from an analytic law. See Limb darkening.

pip install ldtk

Both imports are guarded, so PyTransit works without them; only the corresponding features are unavailable.

Verifying the installation

from numpy import pi, linspace
from pytransit import RoadRunnerModel

tm = RoadRunnerModel('quadratic')
tm.set_data(linspace(-0.1, 0.1, 1000))
flux = tm.evaluate(k=0.1, ldc=[0.2, 0.1], t0=0.0, p=1.0, a=3.0, i=0.5*pi)

print(flux.min())     # ~0.989

The first evaluation triggers Numba’s just-in-time compilation and takes a few seconds. Subsequent evaluations are fast.