Optimal Planning with Axioms
Authors: Franc Ivankovic, Patrik Haslum
IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Figure 2: Node expansions required to prove optimality on equivalent problem formulations with and without axioms: (a) Sokoban problems; (b) controller verification problems due to Ghosh et al. [2015]. The door controller problems are marked with a dot ( ) in (b).", "Figure 3: Total nodes expanded (above) and total planning time (below) with different heuristics. |
| Researcher Affiliation | Collaboration | Franc Ivankovic and P@trik Haslum Australian National University & NICTA Optimisation Research Group firstname.lastname@anu.edu.au", "NICTA is funded by the Australian Government through the Department of Communications and the Australian Research Council through the ICT Centre of Excellence Program. |
| Pseudocode | No | The paper describes procedures in narrative text, such as the stratified fixpoint procedure, but does not present them in a formally labeled pseudocode or algorithm block. |
| Open Source Code | No | The paper states: 'Our implementation is built on the Fast Downward planner (fast-downward.org). The ASP solver is Clasp v2.1.3 (potassco.sourceforge.net).' This refers to third-party tools used, not open-source code released by the authors for their specific methodology. |
| Open Datasets | Yes | We use five sets of problems: the verification problems from Ghosh et al. [2015], the PSR domain (middle-size set from IPC 2004), compiled multi-agent planning problems due to Kominis and Geffner [2015], instances of the trapping game (cf. Section 3) played on graphs with 4 47 nodes, and random instances of the Min-Cut domain, with graphs of size 12 20 and 3 4 roadblocks. |
| Dataset Splits | No | The paper mentions various problem sets (Sokoban, IPC domains, verification problems from other works) used for evaluation, but it does not specify explicit training, validation, or test data splits (e.g., percentages, sample counts, or methodology for splitting). |
| Hardware Specification | No | The paper mentions experimental limits like 'CPU time is limited to 1h and memory to 3Gb per problem,' but it does not specify any particular hardware details such as CPU/GPU models, processor types, or memory amounts used for running the experiments. |
| Software Dependencies | Yes | Our implementation is built on the Fast Downward planner (fast-downward.org). The ASP solver is Clasp v2.1.3 (potassco.sourceforge.net). |
| Experiment Setup | No | The paper states 'Each planner was run with up to 1h CPU time and 3 Gb memory per problem,' and describes the problem sets used, but it does not provide specific experimental setup details such as hyperparameters, optimizer settings, or model initialization procedures for the algorithms or models being evaluated. |