Artifact (software + dataset) for "Phys: Probabilistic Physical Unit Assignment and Inconsistency Detection"

https://unl-nimbus-lab.github.io/phys

phys is a tool to detect physical unit inconsistencies in C++ that builds against the Robot Operating System. phys is a command-line tool that can be used with Docker.

For example, assume a developer is building a robot and wants to calculate the length of the hypotenuse of a triangle with sides x_err and y_err, but the developer accidentally uses a multiply symbol (*) instead of an addition symbol (+), and ends up adding x_err * x_err (meter-squared) to err_y (meter):

err_d = sqrt( err_x * err_x + err_y + err_y); (source)

This code compiles and but this bug can be difficult to detect.  It is coincidentally correct for values near 2. Yikes!  This kind of error is called a physical unit inconsistency, and phys can detect it without developer annotations.

-------------

THE BEST WAY TO START is by opening index.html from a web browser.

This repository contains the following files and folders:

• Dockerfile : a Docker install file for **phys**.
• INSTALL.txt  : Describes how to install **phys** using Docker and a minimal working example.
• LICENSE.txt : BSD-2-Clause license.
• README.md  :  This file. 
• STATUS  : Describes the ACM artifact badges sought for this artifact.
• data/  : Folder containing C++ files used to evaluate and test **phys**.
• docs/  : the help pages (HTML) referenced by index.html.
• index.html : THE BEST PLACE TO START.
• requirements.txt : List of python dependencies required by **phys**.  Referenced by the Dockerfile.
• src/ : The python source code for **phys**.

Credits

• CPPCheck
• libDAI
• NetworkX
• Phriky
• NLTK