Effect-Abstraction Based Relaxation for Linear Numeric Planning

Authors: Dongxu Li, Enrico Scala, Patrik Haslum, Sergiy Bogomolov

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 6 Experiments In this section, we study the practical implications of the effect abstraction implemented by the hadd abs schema. We implemented hadd abs in the ENHSP planner3, and compared it to other heuristics in a greedy best-first search (f(n) = h(n) with ties broken in favor of lower g-values).
Researcher Affiliation Academia Dongxu Li1, Enrico Scala2, Patrik Haslum1,3, Sergiy Bogomolov1 1 The Australian National University 2 Fondazione Bruno Kessler (Italy) 3 CSIRO Data61
Pseudocode Yes Algorithm 1: AIBR-based-Decomposition
Open Source Code Yes We implemented hadd abs in the ENHSP planner3, and compared it to other heuristics in a greedy best-first search (f(n) = h(n) with ties broken in favor of lower g-values). 3https://gitlab.com/enricos83/ENHSP-Public
Open Datasets Yes Domains. We extended simple numeric planning domains from the literature [Scala et al., 2016a; Franc es and Geffner, 2015; Piacentini et al., 2018]
Dataset Splits No The paper evaluates performance on various planning problem domains (e.g., FO-COUNT, FO-SAILING, FO-FARMLAND, TPP-METRIC) and refers to 'instances' within these domains. However, it does not specify explicit training, validation, and test dataset splits with percentages or counts, which are typical for machine learning evaluations.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, or memory) used for running the experiments.
Software Dependencies No The paper mentions implementing 'hadd abs in the ENHSP planner' but does not specify version numbers for ENHSP or any other software dependencies.
Experiment Setup Yes We implemented hadd abs in the ENHSP planner3, and compared it to other heuristics in a greedy best-first search (f(n) = h(n) with ties broken in favor of lower g-values)... Timeout is 1800 seconds.