========================================================================
                           PbP.s - version 2.0
                      Essential User Instructions 
========================================================================
   This code is provided for research and experimental purposes only

           Alfonso Gerevini, Alessandro Saetti and Mauro Vallati              
                         University of Brescia 
                 Via Branze 38, I-25123, Brescia, Italy
                  {gerevini,saetti,mauro.vallati}@ing.unibs.it 
         
				June 2012


This version of PbP is optimized for runtime metric.


PbP is a Planner using a set of planning algorithms and different 
types of knowledge.

1. Installation
-------------------------------
Requirements:

PbP contains several different planners. For this reason you need:
C / C++
Python
Awk
Bison / Flex
Ruby
Java (1.6)
Objective Caml >= 3.06

Parts of this learner require a Sun java 1.6 JVM in the path, as well as the existence of ruby. This tool uses the HAL framework to coordinate experiment runs, and is packaged in this submission as a library. Source code is available from http://hal.cs.ubc.ca/


The PbP planner and its incorporated tecnhiques have been compiled on Debian-based 32bit linux distribution.

2. Running PbP
--------------------
From inside the PbP directory run:

Learning stage:
learner -o <domain file> -t <training folder> -k <DCK folder> [ -s time (in seconds, e.g. 900.00) ]


First option specifies the path to the domain 
definition file. Second option specifies the path to the folder with the set of 
training problems. Third option indicates the output folder for the learned 
Domain-specific Control Knowledge (DCK). Fourth option indicates the CPU-time limits for configurating the portfolio. The default value is 900 seconds.

Testing stage:
planner -o <domain file> -f <problem file> -p <plan file> [-k <DCK folder>] [-s time (in seconds, e.g. 900.00) ]

The <DCK folder> option is optional and will 
only be specified when evaluating the effectiveness of the learned knowledge. 
When this option is not specified, the planner should run in knowledge-free mode. 
The <plan file> option indicates the filename in which the result plan should be stored. The time indicates with configured portfolio as to be used for the testing phase. Default value is 900 seconds. For exploiting a portfolio, you must have configured it before for the given CPU-time limit.

3. Description
--------------
PbP uses the following set of planning algorithms:

LAMA, Richter & Westphal, 2010
Fast Downward, Helmert, 2006
Sgplan, Chen, Wah & Hsu, 2006
YAHSP, Vidal, 2004
LPG-td, Gerevini, Saetti & Serina, 2005
Metric-FF, Hoffmann & Nebel, 2001
MacroFF, Botea, Enzenberger, Müller & Schaeffer, 2005
Marvin, Coles & Smith, 2007

For extracting knowledge, it uses:

Wizard, Newton et al. 2007
MacroFF, Botea, Enzenberger, Müller & Schaeffer, 2005
ParamILS, Hutter, Hoos, Leyton-Brown & Stützle, 2009.



Disclaimer
----------

By releasing this code we imply no warranty as to its reliability and its use is entirely at
your own risk.
