Efficiently Implementing GOLOG with Answer Set Programming

Authors: Malcolm Ryan

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We calculate the size of the grounded ASP for basic program structures and then we empirically evaluate each approach on a complex example program. We ran instances of this problem with size k varying from 5 to 100, measuring compilation, grounding and solving times. In each case the maximum plan length was 100 time steps. Each instance was run 100 times and average times are reported. Runs that took more than 100s were aborted. The total run time, including compilation, grounding and solving, for each approach is plotted in Figure 5(a). The difference in performance is striking. The Con GOLOG encoding is almost an order of magnitude faster than the GOLOG encoding and the finite state machine encoding is another order of magnitude faster still. The size of the ground programs, as shown in Figure 5(b), also reflect this.
Researcher Affiliation Academia Malcolm Ryan School of Computer Science and Engineering University of New South Wales, Australia malcolmr@cse.unsw.edu.au
Pseudocode Yes Figure 1: An encoding of GOLOG s do predicate in ASP. Figure 2: Generating all subprograms from a given program. Figure 3: An encoding of Con GOLOG s trans predicate in ASP. Figure 4: (a) The augmented state machine for the GOLOG program (φ? ; a ; (b | c)) ; φ? ; d ; ψ? and (b) its ASP encoding.
Open Source Code No The paper mentions "Available at http://potassco.sourceforge.net/" which refers to the tools used (gringo and clasp), not the authors' own implemented compiler or encodings described in the paper.
Open Datasets No The paper describes generating instances of programs for testing, such as "a ((a|b); a) ((a|b)2; a) . . . ((a|b)k 1; a) ; (a|b)k". It also states: "The action model contained only the actions a and b necessary for this program. Their preconditions were always true and they had no effects. This test is designed to focus on the variation in run-time caused by the GOLOG implementation, so the planning domain itself is deliberately simple." There is no mention of using or providing access to a publicly available dataset.
Dataset Splits No The paper describes running instances of a program and measuring times, but it does not specify any training, validation, or test dataset splits in terms of percentages or counts, or refer to standard predefined splits.
Hardware Specification Yes Measurements were taken on a 3.20GHz Intel Xeon TM.
Software Dependencies Yes We use the Potassco tools gringo and clasp for this purpose5. 5Experiments were run with gringo 3.0.5 and clasp 2.1.5.
Experiment Setup Yes We ran instances of this problem with size k varying from 5 to 100, measuring compilation, grounding and solving times6. In each case the maximum plan length was 100 time steps. Each instance was run 100 times and average times are reported. Runs that took more than 100s were aborted.