Probabilistic Generalization of Backdoor Trees with Application to SAT

Authors: Alexander Semenov, Daniil Chivilikhin, Stepan Kochemazov, Ibragim Dzhiblavi

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In the experimental part of the paper, we show that moving from the metaheuristic search for ρ-backdoors to that of ρ-backdoor trees allows drastically reducing the time required to construct the required decompositions without compromising their quality.
Researcher Affiliation Academia Alexander Semenov, Daniil Chivilikhin, Stepan Kochemazov, Ibragim Dzhiblavi ITMO University, St. Petersburg, Russia
Pseudocode No The paper describes algorithms and procedures in prose but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code Yes The search algorithm for ρ-backdoor trees was implemented in C++1. 1https://github.com/ctlab/itmo parsat
Open Datasets No The paper mentions using instances from a GitHub repository (Pavlenko 2022) but does not explicitly state that these datasets are publicly available or open, nor does it provide concrete access information in terms of citations with author/year or specific links to the datasets themselves, only to the repository for the instances.
Dataset Splits No The paper does not specify any training, validation, or test dataset splits.
Hardware Specification Yes All experiments were run on a computer with a 32-Core Intel(R) Xeon(R) 106 CPU @ 1.99 GHz and 128 GB of RAM.
Software Dependencies No The paper mentions using the Minisat solver (E en and S orensson 2004) and solvers Kissat (Biere 2022) and Ca Di Ca L (Biere 2021) but does not provide specific version numbers for these software dependencies. It only lists the commit hashes for Kissat and Ca Di Ca L in their references/footnotes, which are not explicitly presented as version numbers in the main text.
Experiment Setup Yes In the search algorithm, for all ρ-backdoor trees with |B| < 16, the value ρ was calculated exactly, and for larger backdoors the corresponding random sampling was used during fitness function evaluation. We used a sample of size 103 in these cases. The number is computed using Chernoff bounds in a way similar to (Semenov et al. 2022).