physical_event_primitives

Repository for computational, behavioral, and imaging studies of physical event perception.

Setup

Requirements: singularity

If you are working on a Linux computer, download singularity and you should be good to go! :

1. git clone the repository
2. cd physical_event_primitives
3. git submodule update --init (initialize the Blender egg importer add-on)
4. ./setup.sh pull true true (pulls the container from Box.com, sets up Conda and Julia environments). Run ./setup.sh build true true if you want to build the container for some reason.

If you are working on a Mac, the situation is more complicated.

1. Follow instructions here: https://sylabs.io/guides/3.5/admin-guide/installation.html#mac to download vagrant, vagrant-manager, and virtual box. You need to have Homebrew to install this.
2. Make sure to create the virtual machine inside of the physical_event_primitives directory, or in a directory that encloses it.
3. When you want to build or pull the singularity image, go into the directory with the Vagrantfile and run vagrant ssh. Then cd into the physical_event_primitives directory which is located within /vagrant/. You can now run the above commands as if you were working on a Linux!

Mac users may also need to cp default.conf user.conf and change the Julia depot line to julia_depot:/home/vagrant/.julia/.

Interacting with the image

Run ./run.sh <command> to execute commands with the image, e.g. to launch Julia REPL ./run.sh julia.

If using Milgram and need a virtual display (e.g. rendering with Blender), run ./run.sh xvfb-run -a <command>.

Disclaimer: Much of this code is taken directly from http://geometry.cs.ucl.ac.uk/projects/2019/causal-graphs/ with tweaks to fit our specific situation

Contents

blender/ - Scripts used to export animations to Blender

core/ - Core package; contains all the algorithms

demos/ - Demos to play with & contains the main generate.py script for generating videos

gui/ - Graphical modules -- mostly irrelevant here

scenarios/ - Config files of different possible scenarios

Usage

You can run the script using ./run.sh demos/generate.py