RP-DeLP_solver

Introduction

This README explains how to install the RP-DeLP solver in your local machine as well as the dependencies required. There's a brief explanation of the directory structure and what main scripts do.

In case you don't want to intall the solver in your computer, you can use the RP-DeLP web version which is hosted at http://arinf.udl.cat/rp-delp.

For further explanations or any dectected issue please contact some of the authors:

• guitart.francesc@gmail.com
• ramon@diei.udl.cat
• tracy@diei.udl.cat

Directory overview:

• solvers: This directory contains max-ideal and multiple outputs solvers, as well as auxiliar libraries.
• solvers/debug: this directory contains encodings generated during solving. If it does not exist will be created automatically.
• web: This directory contains scripts for formatting input files. Originally was devoted to read from a web-based application, but now it also converts pdlp files to xml.
• web/user_submitted_programs: this directory contains xml transformations of input files. If it does not exist will be created automatically.

Dependencies:

1. python 2.7

2. sympy dpll installation (example of installation in Fedora):
   #yum install sympy

3. Configure rpdelp_web dependency:
   3.1. Add rpdelp_web.py to pythonpath, example (replace INSTALLATION_PATH for the propper path):
   $export PYTHONPATH=$PYTHONPATH:(INSTALLATION_PATH)/web/app/

4. Install minisat and clingo:
   4.1. minisat (version 2.2.0): http://minisat.se/MiniSat.html
   4.2. clingo (version 3.0.5): http://sourceforge.net/projects/potassco/files/clingo/3.0.5/
   (Note: there should be two links (called clingo and minisat) in your $PATH)

Examples:

• There are some pdlp program examples in:
  https://github.com/f-guitart/RP-DeLP_examples

• These examples are explained in the following paper:
  Web Based System for Weighted Defeasible Argumentation.
  Teresa Alsinet, Ramón Béjar, Francesc Guitart, Lluis Godo.
  CLIMA 2013: 155-171

• Execution arguments:
  The solver is called using solvers/solver.py script. Here is a list of execution arguments:

  • -h, --help show this help message and exit
  • -f FILE, --file=FILE reads FILE as input
  • -x FILE, --xml=FILE reads XML FILE as input
  • -o OUTPUT, --output=OUTPUT output computing options
  • -s SOLVER, --solver=SOLVER solver to use
  • -d, --debug write debug to PATH (not working)

• Examples of usage:

  • $python solver.py -f examples/program1.pdlp -s minisat -o max-ideal
  • $python solver.py -f examples/program1.pdlp -s clingo -o max-ideal
  • $python solver.py -f examples/program1.pdlp -s minisat -o multiple
  • $python solver.py -f examples/program1.pdlp -s clingo -o multiple

Output format:

After the execution of the script a JSON file should de created. This file will be named using the following pattern <original_name>-<output_option>-<solver-option>.json. The JSON file contains the following fields:

• file (file solved)
• solver (solver used)
• output (chosen semantics)
• n_outputs (number of outputs)
• timeout (timeout value, not used)
• consistent (true if the strict part of the program is consistent, false otherwise)
• blocked_list (arguments blocked)
• warrant_list (arguments warranted)

The warrant and blocked lists are ordered list containing the arguments warranted and blocked for each output.

Arguments are formatted as follows:

• Blocked:
  • Conclusion
  • Reason of blocking (cycle or conflict)
  • Conflicting literals or literals involved in the cycle
  • Support
  • Strength
• Warranted:
  • Conclusion
  • Support
  • Strength