alarsyo/spot
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

ltlcross: Complement deterministic automata.

Browse source 
* src/bin/ltlcross.cc: Complement deterministic automata,
and use them for additional intersection checks.
* NEWS, doc/org/ltlcross.org, src/bin/man/ltlcross.x: Document it.
This commit is contained in:
Alexandre Duret-Lutz 2013-08-18 22:37:25 +02:00
parent 2dda2c9122
commit 1029d08a77
4 changed files with 182 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

22
src/bin/man/ltlcross.x
View file

@ -81,10 +81,7 @@ Technology. The main motivation for the reimplementation was to
support PSL, and output more statistics about the translations.
The sanity checks performed on the result of each translator (by
either LBTT or ltlcross) are described in the following paper. Our
implementation will detect and reports problems (like inconsistencies
between two translations) but unlike LBTT it does not offer an
interactive mode to investigate such problems.
either LBTT or ltlcross) are described in the following paper.
.TP
th02
@ -92,3 +89,20 @@ H. Tauriainen and K. Heljanko: Testing LTL formula translation into
Büchi automata. Int. J. on Software Tools for Technology Transfer.
Volume 4, number 1, October 2002.
LBTT did not implement Test 2 described in this paper. ltlcross
implements a slight variation: when an automaton produced by some
translator is deterministic, its complement is built and used for
additional cross-comparisons with other tools. If the translation P1
of the positive formula and the translation N1 of the negative formula
both yield deterministic automata (this may only happen for obligation
properties) then the emptiness check of Comp(P1)*Comp(N1) is
equivalent to Test 2 of Tauriainen and Heljanko. If only one
automaton is deterministic, say P1, it can still be used to check we
can be used to check the result of another translators, for instance
checking the emptiness of Comp(P1)*P2.
Our implementation will detect and reports problems (like
inconsistencies between two translations) but unlike LBTT it does not
offer an interactive mode to investigate such problems.