DEFT

This repository contains code for paper "Causal Discovery via Vertical Partitioning of Data Features" -- under review

Instruction to create runtime

Following these steps:

1. Make sure you have conda ready
2. Using conda, create an R-supported python environment

   conda create -n rpy r-essentials r-base python=3.10

3. Activate the environment, and install the following package:

   pip install GPy
   pip install igragh
   pip install cdt bnlearn
   pip install causal-learn gcastle CausalDisco

4. Pytorch is automatically installed when you install bnlearn, but it can be of the wrong cuda version if you are using a GPU engine. So, reinstall it from the Pytorch official website if needed.
5. You will find in the legacy folder, the zip file prepare-r.r. You either let the content be

   install.packages("BiocManager", repos="http://cran.us.r-project.org")
   BiocManager::install('pcalg')

   or (try this first and the above second if this one fails)

   install.packages('pcalg_2.7-12.tar.gz', repos = NULL, type="source")

   And then run

   Rscript prepare-r.r

   Make sure to comfirm that the code is successfully processed as followed in the terminal:

    ** installing vignettes
    ** testing if installed package can be loaded from temporary location
    ** checking absolute paths in shared objects and dynamic libraries
    ** testing if installed package can be loaded from final location
    ** testing if installed package keeps a record of temporary installation path
    * DONE (pcalg)

Instruction to run code

bash run_bash.sh

The bash file is reading-friendly. Make sure the data exists in data/ and the folder res/ is created in advance.