[trivial] Update pdb synopsis. #232
Conversation
src/search/pdbs/pdb_heuristic.cc
Outdated
| "Planning with Pattern Databases", | ||
| "https://ecp01.icaps-conference.org/papers/A-longpapers/2/edelkamp.pdf", | ||
| "Proceedings of the Sixth European Conference on Planning (ECP 2001)", | ||
| "84--90", |
There was a problem hiding this comment.
The tools we use don't use LaTeX syntax, so we use a single dash for page ranges, not two dashes.
The linked version of the paper is not the one described by the reference. (The linked version has 12 pages, not 7; its page numbers would be 13-24.) I think the linked version is the pre-proceedings version (in Springer style) from 2001, not the actual "published" version produced in AAAI style much later. Perhaps that's a good compromise if we can't link the actual published paper, but it is confusing.
There was a problem hiding this comment.
I looked around a bit, and the officially published version is actually available online in the AAAI Library. I'm not sure how stable the links are, though. Here is a link to the page for the paper from which the PDF is references, which should be reasonably stable: https://aaai.org/papers/7280-ecp-01-2001/
| "Computes dictionaries for state-to-goal distances in" | ||
| "state space abstractions based on projections." | ||
| "First used in domain idependent planning by:" | ||
| + paper_references()); |
There was a problem hiding this comment.
There are spaces missing, so I think this will generate "instate" and "projections.First".
Typo "idependent", and I think "domain-independent" should be written with a hyphen, not as two words.
I don't think people describe PDBs as "dictionaries" (in particular because they don't store keys, only values). I think it's not necessary to get into such details in the first place, this could just be changed to "Computes goal distance in state space abstractions based on projections".
The particular PDB implementation in Fast Downward has its own paper (by Ortlieb et al.), which I think is worth referencing. It is already referenced from another PDB-related plug-in, see https://www.fast-downward.org/Doc/PatternCollectionGenerator#Hill_climbing
There was a problem hiding this comment.
I mean the paragraph "For implementation notes, see:" followed by the Ortlieb et al. reference. Most of that paper is actually about the basic PDB implementation, not the hill-climbing pattern collection paper. (But the paper discusses that too, so it's not wrong that the paper is also referenced from that plugin.)
No description provided.