Computing Plan-Length Bounds Using Lengths of Longest Paths

Authors: Mohammad Abdulaziz,Dominik Berger11709-11717

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We implement two methods to compute recurrence diameters based on the work of Biere et al. 1999. Unlike Biere et al., who devised their method and tested it on model-checking problems, we test these methods on planning benchmarks and show that it is impractical for most planning problems. We show, however, that computing the recurrence diameter within the compositional bounding method by Abdulaziz, Gretton, and Norrish 2017 leads to bounds that are, as predicted theoretically, much tighter when rd is used instead of td. Lastly, we experimentally show that the improved bounds lead to an improved problem coverage for state-of-the-art SAT-based planner MP (Rintanen 2012), when the bounds are used as horizons for it.
Researcher Affiliation Academia Mohammad Abdulaziz and Dominik Berger Techniche Universit at M unchen, Germany
Pseudocode No The paper describes algorithms and encodings for computing recurrence diameters, but it does so using mathematical notation and descriptive text. It does not include any clearly labeled "Pseudocode" or "Algorithm" blocks with structured steps.
Open Source Code No The paper mentions using specific software like "Yices 2.6.1" and "MP (Rintanen 2012)" which are third-party tools. However, there is no explicit statement, link, or reference to a repository indicating that the authors have released the source code for their own methodology or implementation described in the paper.
Open Datasets No The paper states, "We run the bounding algorithm by Abdulaziz, Gretton, and Norrish 2017 on standard planning benchmarks (from previous planning competitions and ones we modified)." While this indicates the use of existing datasets, it does not provide concrete access information (e.g., specific links, DOIs, or formal citations with author/year for public availability) for these benchmarks.
Dataset Splits No The paper discusses 'train', 'validation', and 'test' in the context of action sequences and states (e.g., 'states traversed by executing π at x'). It does not provide any specific information regarding dataset splits (e.g., percentages or sample counts for training, validation, and test sets) to enable reproduction of data partitioning for machine learning experiments.
Hardware Specification Yes We perform our experiments on a cluster of 2.3GHz Intel Xeon machines with a timeout of 20 minutes and a memory limit of 4GB.
Software Dependencies Yes We use Yices 2.6.1 (Dutertre 2014) as the SMT solver to prove the satisfiability or unsatisfiability of the resulting SMT formulae.
Experiment Setup Yes We perform our experiments on a cluster of 2.3GHz Intel Xeon machines with a timeout of 20 minutes and a memory limit of 4GB.