GTLS (GPU Transit Least Squares)

A GPU algorithm for speeding up periodic transit detection based on TLS.

Installation

Requirements

A CUDA-capable GPU is required.

Install

This program is based on cupy. Due to there are many versions of cupy, we cannot specify the version of cupy in the package. So, you need to install cupy manually first.

Please refer to the official document, and install the version of cupy that is suitable for your environment.

After installing cupy, you can install this program by running the following command:

pip3 install gputls

Usage

#Assume that you have a time series data: time, flux
from gtls import gtls
model = gtls(time, flux)
period, duration, depth, T0, SDE = model.power()

You can also use the test script in the repository:

python3 generatedTest.py

For now, There are no detailed documents. Please refer to the TLS first, since the usage of this program is almost the same as TLS.

License

The GTLS(GPU Transit Least Squares) algorithm is adapted from the TLS(Transit Least Squares) algorithm by Michael Hippke & René Heller (2019).

The TLS is an open source software with MIT license. The copyright of the TLS algorithm is held by its authors.

The GTLS is also an open source software with MIT license.