The aim of the project is to improve constraint programming and Boolean satisfaction techniques to provide more efficiency in solving problems motivated by artificial intelligence. The project assumes to use tasks from automated planning (a sub-area of artificial intelligence) as a source of difficult problems to drive the improvements of constraint programming and Boolean satisfiability. The project is targeted on studying several particular questions, namely: (i) How to apply network flows for relaxing hard problems arising in constraint programming and Boolean satisfiability? (ii) How to exploit structural properties of problems arising in constraint programming and Boolean satisfiability for developing new types of consistencies? (iii) How does the structure of highly parallel planning problems look like? (iv) What is the characterization of easy and difficult problems in relation to questions (i), (ii), and (iii)? (en)
(i) Aplikovat toky v sítích pro relaxaci těžkých problémů z umělé inteligence. (ii) Navrhnout konzistenční techniky založené na strukturálních vlastnostech. (iii) Popsat strukturu vysoce paralelního plánování. (iv) Charakterizovat obtížnost problémů s využitím (i), (ii) a (iii).