ForewordThe procedural reconstruction is a feature of the proof assistant Matita for translating the system's logical representation of formalized mathematical knowledge (mainly of formalized proofs) into the system's highlevel specification language.This feature has its justification in the actual fact that proof assistants allow to edit some knowledge only starting from its highlevel representation, so the adaptation and maintenance of this knowledge can not be performed at the logical level. The knowledge to which this translation applies, can come from any source that contributes to HELM (that is the digital library of Matita). For instance it can come from another digital library or from another proof assistant. Therefore this transformation may help the crossplatform maintenance of the certified specifications (as in the case of the formal system λδ). This transformation is focused on proofs. In general terms a proof has two main aspects:
In fact adapting or maintaining a proof usually means changing some aspects of its contents while trying to preserve its structure. Since a proof may need some transformation before being translated, for the user convenience we preprocess the proof with an optimizing procedure. An optimized proof does not contain detours or unused information and is normalized according to the direct proof paradigm to possibly enhance its readability and ease its maintenance. News
Documentation
ImplementationThe procedural reconstruction is activated by invoking the Matita highlevel command inline.The parameters of the inline command affecting the procedural reconstruction procedure, are the following:
