Parity games have been solved semi-locally so far.
We deduced a strategy for the reachable part of the arena
This lead to some inconsistencies when not all state were
rachable.
Now you can chose to solve parity games truely globally.
* spot/twaalgos/game.cc, spot/twaalgos/game.hh: Here
* tests/python/games.ipynb: Test
Fixes issue #525.
* spot/twaalgos/dbranch.hh, NEWS: Document.
* spot/twaalgos/dbranch.cc: Detect cases where the acceptance should
be changed from state-based to transition-based.
* tests/python/dbranch.py: Add a test case.
* spot/twaalgos/game.hh, spot/twaalgos/game.cc (get_state_players,
get_strategy, get_state_winners): Take argument by reference, not
copy.
* spot/twaalgos/synthesis.cc, spot/twaalgos/mealy_machine.cc: Replace
auto by actual type for readability.
Fixes#524, reported by Rüdiger Ehlers.
* spot/twaalgos/dbranch.cc: When merging an edge going to state
without successors simply delete it.
* bin/spot-x.cc: Typo in documentation.
* tests/core/ltlcross.test: Add a test case.
* NEWS: Mention the bug.
Put the new function to use in order to speed up
mealy machine minimization
* spot/twaalgos/mealy_machine.cc: Here
* spot/twaalgos/synthesis.cc
, spot/twaalgos/synthesis.hh: Helper function to relabel games
* tests/python/_mealy.ipynb
, tests/python/except.py
, tests/python/_partitioned_relabel.ipynb: Adapt/expand tests
Function taking an automaton and trying to relabel
it by partitioning the old conditions and encode the
different subsets of the partition with new variables
* spot/priv/Makefile.am: Add
* spot/priv/partitioned_relabel.hh
, spot/priv/partitioned_relabel.cc: try_partition_me,
computes the partition of a given vector of bdds
* spot/twaalgos/relabel.hh
, spot/twaalgos/relabel.cc: Here. Adapt also relabel()
to cope with the different type of relabeling_maps
* tests/python/_partitioned_relabel.ipynb
, tests/python/except.py: Test and Usage
* tests/Makefile.am: Add test
When calling solve_parity_game() multiple times on the same
automaton the strategies are appended one after the other.
Reported by Dávid Smolka.
* NEWS: Mention the bug.
* spot/twaalgos/game.cc: Fix it.
* tests/python/game.py: Test it.
* THANKS: Add Dávid.
* spot/twaalgos/mealy_machine.cc (minimize_mealy): Do not compare
result to the original unsplit machine without splitting it first.
* tests/python/mealy.py: Add a test case.
This issue surfaced in twacube.test after the previous patches.
* spot/twaalgos/ltl2tgba_fm.cc: Release the formula namer on abort.
* NEWS: Mention the bug.
Fixes#521.
* spot/tl/simplify.cc, spot/tl/simplify.hh,
spot/twaalgos/translate.cc, spot/twaalgos/translate.hh: Add an option
to limit automata-based implication checks of n-ary operators when too
many operands are used. Defaults to 16.
* bin/spot-x.cc, NEWS, doc/tl/tl.tex: Document it.
* tests/core/bdd.test: Disable the limit for this test.
* spot/priv/satcommon.cc, spot/twaalgos/dtbasat.cc,
spot/twaalgos/dtwasat.cc: When setting exception on std::ofstream, use
ofstream::failbit and ofstream::badbit instead of ifstream::failbit
and ifstream::badbit.
* spot/twaalgos/relabel.cc (relabel_here): This function has multiple
exit paths, and none of them were calling bdd_freepair. Use a
unique_ptr to ensure that.
Doing so reduced the number of GC passes tested in bdd.test, which is
good.
* spot/twaalgos/ltl2tgba_fm.cc: Simplify minato loops with
bdd_restrict.
* spot/twaalgos/synthesis.cc (split_2step): Use bdd_restrict instead
of bdd_appex.
* tests/core/bdd.test, tests/core/ltlf.test: Adjust test cases.
Based on a request from Pierre Ganty.
* spot/twaalgos/stats.cc, spot/twaalgos/stats.hh,
bin/common_aoutput.cc, bin/common_aoutput.hh: Implement those
options.
* tests/core/format.test: Add test case.
* doc/org/autfilt.org: Update doc.
* NEWS: Mention them.
* spot/twaalgos/parity.cc, spot/twaalgos/parity.hh: Add a
reduce_parity_data class for access to the vectors of colors
computed by reduce_parity.
* python/spot/impl.i: Add bindings for std::vector<int>.
* spot/twaalgos/parity.cc: Implement it.
* spot/twaalgos/parity.hh, NEWS: Document it.
* tests/python/parity.ipynb: Demonstrate it. This is the only test so
far, but more uses are coming.
* spot/twaalgos/translate.cc: When producing Parity output, split LTL
as we do in the Generic case.
* spot/twaalgos/postproc.hh, spot/twaalgos/postproc.cc: Use
acd_transform() and add an "acd" option to disable this.
* bin/spot-x.cc, NEWS: Document this.
* tests/core/genltl.test, tests/core/minusx.test,
tests/core/parity2.test: Adjust test cases for improved outputs.
This is motivated by an example sent by Edmond Irani Liu,
that will be tested in next patch.
* spot/twaalgos/dbranch.cc, spot/twaalgos/dbranch.hh: New files.
* python/spot/impl.i, spot/twaalgos/Makefile.am: Add them.
* spot/twaalgos/translate.cc: Call delay_branching_here
unconditionally.
* spot/twa/twagraph.cc (defrag_states): Do not assume
that games are alternating.
* tests/core/genltl.test: Adjust expected numbers.
* tests/python/dbranch.py: New file.
* tests/Makefile.am: Add it.
Typically intead of doing
minato_isop isop(rel & letter);
while (bdd cube = isop.next()) {
bdd res = bdd_exists(cube, ap)
...
}
do
minato_isop isop(bdd_relprod(rel, letter, ap);
while (bdd res = isop.next()) {
...
}
this way the existential quantification is done once at the same time
of the conjunction, and isop has fewer variable to work with.
* spot/twaalgos/alternation.cc, spot/twaalgos/dualize.cc,
spot/twaalgos/simulation.cc, spot/twaalgos/toweak.cc: Here.
* spot/twaalgos/postproc.cc, spot/twaalgos/postproc.hh: Introduce a
merge-states-min option.
* bin/spot-x.cc: Document it.
* spot/gen/automata.cc, spot/gen/automata.hh, bin/genaut.cc: Add
option to generate cyclist test cases.
* NEWS: Document the above.
* tests/core/included.test: Add test cases that used to be too slow.
* spot/twaalgos/degen.hh, spot/twaalgos/degen.cc: Adjust
degeneralize() and degeneralize_tba() to work on generalized-co-Büchi.
* NEWS: Mention this.
* spot/twaalgos/cobuchi.hh, spot/twaalgos/cobuchi.cc (to_nca): Use
degeneralization on generalized-co-Büchi.
* spot/twaalgos/postproc.cc: Use degeneralization for generalized
co-Büchi as well.
* bin/autfilt.cc: Improve chain products of co-Büchi automata by using
generalization if too many colors are needed.
* tests/core/prodchain.test, tests/python/pdegen.py: Add test cases.
Improve the construction of the above constructions, saving colors.
* spot/twaalgos/product.cc: Here.
* spot/twaalgos/product.hh, NEWS: Mention it.
* tests/core/prodchain.test, tests/core/prodor.test,
tests/python/_product_weak.ipynb: Adjust.
* NEWS: Mention those change.
* spot/twaalgos/game.hh (print_pg): New function.
(pg_print): Mark as deprecated.
* spot/twaalgos/game.cc (pg_print): Redirect to print_pg().
(print_pg): Update to output state names.
* python/spot/__init__.py: Teach to_str() about print_pg().
* bin/ltlsynt.cc: Adjust to call print_pg().
* tests/python/games.ipynb: Add an example.
* tests/core/ltlsynt.test: Adjust to remove the "INIT" note.
* spot/twaalgos/genem.cc, spot/twaalgos/genem.hh: add detection
of edges that are in at least one accepting cycle.
* spot/twaalgos/toparity.cc,
spot/twaalgos/toparity.hh: add parity_type_to_parity and
buchi_type_to_buchi.
* spot/twaalgos/game.cc: solve_parity_game now works for any of
the four parity types and partially colored graphs.
Also removing unnescessary steps from Zielonka.
h: Update
* tests/python/game.py: Update and additional tests
* tests/python/except.py: Remove outdated exception
