Abstract
The architecture of the WALDMEISTER prover for unit equational deduction is based on a strict separation of active and passive facts. After an inspection of the system's proof procedure, the representation of each of the central data structures is outlined, namely indexing for the active facts, compression for the passive facts, successor sets for the hypotheses, and minimal recording of inference steps for the proof object. In order to cope with large search spaces, specialized redundancy criteria are employed, and the empirically gained control knowledge is integrated to ease the use of the system. The paper concludes with a quantitative comparison of the WALDMEISTER versions over the years, and a view of the future prospects.
Get full access to this article
View all access options for this article.