An ENHSP binary containing satisficing configurations from Novelty Heuristics, Multi-Queue Search, and Portfolios for Numeric Planning at SOCS-24. See reference for the bib entry.
The binary is compiled with OpenJDK 17. If you are using Ubuntu, you can install the required Java Runtime Environment with
sudo apt install openjdk-17-jre
If you are using a different system, any JRE version >=17
is likely to work.
The input is a PDDL domain file <domain>
, a PDDL problem file <problem>
, and an ENHSP configuration <config>
. The command line instruction is
java -jar enhsp.jar -o <domain> -f <problem> -planner <config>
where the top new configurations ordered by total coverage in the paper are
sat-mq3h3n
sat-mq3h
sat-mq3n
sat-hiqb2add
sat-hiqb2mrphj
sat-hmd
Note that performance of a configuration depends on the domain, and lower ranked configurations could perform better for your use case.
We plan to integrate the source code with the original ENHSP source code in the near or far future.
@inproceedings{chen:thiebaux:socs2024,
author = {Dillon Ze Chen and Sylvie Thi{\'{e}}baux},
title = {Novelty Heuristics, Multi-Queue Search, and Portfolios for Numeric Planning},
booktitle = {17th International Symposium on Combinatorial Search (SoCS)},
year = {2024},
}