vip-hci 1.1.2

Package for astronomical high-contrast image processing.

---------------------------------------------------
        oooooo     oooo ooooo ooooooooo.
         `888.     .8'  `888' `888   `Y88.
          `888.   .8'    888   888   .d88'
           `888. .8'     888   888ooo88P'
            `888.8'      888   888
             `888'       888   888
              `8'       o888o o888o
---------------------------------------------------
         Vortex Image Processing package
---------------------------------------------------

Introduction

VIP is a python package for angular, reference star and spectral differential imaging for exoplanet and disk high-contrast imaging. VIP is compatible with Python 3.7, 3.8 and 3.9 (Python 2 compatibility dropped with VIP 0.9.9).

The goal of VIP is to integrate open-source, efficient, easy-to-use and well-documented implementations of high-contrast image processing algorithms to the interested scientific community. The main repository of VIP resides on GitHub, the standard for scientific open source code distribution, using Git as a version control system.

VIP started as the effort of Carlos Alberto Gomez Gonzalez, a former PhD student of the VORTEX team (ULiege, Belgium). VIP’s development has been led by Dr. Gomez with contributions made by collaborators from several teams (take a look at the contributors tab on VIP’s GitHub repository). It is now maintained by Dr. Valentin Christiaens. Most of VIP’s functionalities are mature but it doesn’t mean it’s free from bugs. The code is continuously evolving and therefore feedback/contributions are greatly appreciated. If you want to report a bug or suggest a functionality please create an issue on GitHub. Pull requests are very welcomed!

Documentation

The documentation for VIP can be found here: http://vip.readthedocs.io.

Jupyter notebook tutorial

Tutorials, in the form of Jupyter notebooks, showcasing VIP’s usage and other resources such as test datasets are available in the VIP-extras repository. Alternatively, you can execute this repository on Binder (in the tutorials directory). The first (quick-start) notebook can be visualized online with nbviewer. If you are new to the Jupyter notebook application check out the beginner’s guide.

TL;DR setup guide

$ pip install vip_hci

Installation and dependencies

The benefits of using a Python package manager (distribution), such as (ana)conda or Canopy, are many. Mainly, it brings easy and robust package management and avoids messing up with your system’s default python. An alternative is to use package managers like apt-get for Ubuntu or Homebrew/MacPorts/Fink for macOS. We recommend using Miniconda.

VIP depends on existing packages from the Python ecosystem, such as numpy, scipy, matplotlib, pandas, astropy, scikit-learn, scikit-image, photutils and others. There are different ways of installing VIP suitable for different scenarios.

Using pip

The easiest way to install VIP is through the Python Package Index, aka PyPI, with the pip package manager. Simply run:

$ pip install vip_hci

With pip you can easily uninstall, upgrade or install a specific version of VIP. For upgrading the package run:

$ pip install --upgrade vip_hci

Alternatively, you can use pip install and point to the GitHub repo:

$ pip install git+https://github.com/vortex-exoplanet/VIP.git

Using the setup.py file

You can download VIP from its GitHub repository as a zip file. A setup.py file (setuptools) is included in the root folder of VIP. Enter the package’s root folder and run:

$ python setup.py install

Using Git

If you plan to contribute or experiment with the code you need to make a fork of the repository (click on the fork button in the top right corner) and clone it:

$ git clone https://github.com/<replace-by-your-username>/VIP.git

If you do not create a fork, you can still benefit from the git syncing functionalities by cloning the repository (but will not be able to contribute):

$ git clone https://github.com/vortex-exoplanet/VIP.git

Before installing the package, it is highly recommended to create a dedicated conda environment to not mess up with the package versions in your base environment. This can be done easily with (replace vipenv by the name you want for your environment):

$ conda create -n vipenv python=3.9 ipython

Note: installing ipython while creating the environment with the above line will avoid a commonly reported issue which stems from trying to import VIP from within a base python2.7 ipython console.

To install VIP, simply cd into the VIP directory and run the setup file in ‘develop’ mode:

$ cd VIP
$ python setup.py develop

If cloned from your fork, make sure to link your VIP directory to the upstream source, to be able to easily update your local copy when a new version comes out or a bug is fixed:

$ git add remote upstream https://github.com/vortex-exoplanet/VIP.git

If you plan to develop VIP or use it intensively, it is highly recommended to also install the optional dependencies listed below.

Optional dependencies

The following dependencies are not automatically installed upon installation of VIP but may significantly improve your experience:

• VIP contains a class vip_hci.vip_ds9 that enables, through pyds9, the interaction with a DS9 window (displaying numpy arrays, controlling the display options, etc). To enable this feature, pyds9 must be installed from the latest development version: pip install git+git://github.com/ericmandel/pyds9.git#egg=pyds9

• Also, you can install the Intel Math Kernel Library (mkl) optimizations (provided that you have a recent version of conda) or openblas libraries. Either of them can be installed with conda install.

• For maximum speed on image operations, such as scaling, shifts or rotations (the latter being a typical bottleneck in most post-processing routines), one can install OpenCV (pip install opencv-python). To use it, just set imlib=’opencv’ (combined e.g. with interpolation=’lanczos4’) instead of the default imlib=’vip-fft’ when using any routine involving image operations. While it can yield up to 3 orders of magnitude speed improvement, note that flux conservation will be worse than using FFT-based methods.

• VIP offers the possibility of computing SVDs on GPU by using CuPy (starting from version 0.8.0) or PyTorch (from version 0.9.2). These remain as optional requirements, to be installed by the user, as well as a proper CUDA environment (and a decent GPU card).

• Finally, bad pixel correction routines can be optimised with Numba, which converts some Python code, particularly NumPy, into fast machine code. A factor up to ~50x times speed improvement can be obtained on large images compared to NumPy. Numba can be installed with conda install numba.

Loading VIP

Finally, start Python (or IPython or a Jupyter notebook if you prefer) and check that you are able to import VIP:

import vip_hci as vip

If everything went fine with the installation, you will see a welcome message. Now you can start finding exoplanets!

Image conventions

By default, VIP routines are compatible with either even- or odd-dimension input frames. For VIP routines that require the star to be centered in the input images (e.g. post-processing routines involving (de)rotation or scaling), the code will assume that it is placed on (zero-based indexing):

• size/2-0.5 for odd-size input images;

• size/2 for even-size input images;

i.e. exactly on a pixel in either cases. The VIP recentering routines will place the star centroid at one of these locations accordingly.

Mailing list

Please subscribe to our mailing list if you want to be informed of VIP’s latest developments (new versions and/or updates).

Attribution

Please cite Gomez Gonzalez et al. (2017) whenever you publish data reduced with VIP . Astrophysics Source Code Library reference [ascl:1603.003]. In addition, please cite the relevant publication(s) for the algorithms you use within VIP (usually mentioned in the documentation, e.g. Marois et al. 2006 for median-ADI).

Note: The specfit module, previously part of VIP, has now been moved to a separate GitHub repository.