Age | Commit message (Collapse) | Author |
|
Eager incremental solving is achieved via solving under assumptions. As soon as incremental and eager bit-blasting is enabled, assertions are passed to the SAT solver as assumptions rather than assertions. This requires the eager SAT solver to support incremental solving, which is currently only supported by CryptoMiniSat.
|
|
|
|
Fixes issue #2137. CnfProof has a stack of assertions that are being
converted to clauses. Previously, it could happen that while an
assertion was being added, TheoryProxy::explainPropagation() would be
called from Solver::reason() and push an assertion to the stack that was
then not removed. This lead to a clause id of the assertion being
associated with the explanation instead, which in turn could lead to a
wrong unsat core. This commit identifies two cases where
TheoryProxy::explainPropagation() is called without cleaning up the
assertion stack afterwards. It also adds an assertion that the assertion
stack must be empty when we are getting the unsat core.
|
|
In MiniSat's analyze(), we were taking a reference of a clause that
could be invalidated by a call to resolveOutUnit(). resolveOutUnit() can
lead to allocation of clauses which in turn can lead to clauses being
reallocated, making the reference stale. The commit encloses the
reference in a code block that makes the lifetime of the reference more
obvious and removes uses of the potentially stale reference.
|
|
|
|
This commit implements a workaround for computing unsat cores while
doing incremental solving to address #1349. Currently, our resolution
proofs do not handle clauses with a lower-than-assertion-level correctly
because the resolution proofs for them get removed when popping the
context but the SAT solver keeps using them. The workaround changes the
behavior of the SAT solver to add clauses always at the current level if
incremental solving and unsat cores are enabled. This makes sure that
all learned clauses are removed when we pop below the level that they
were learned at. This may degrade performance because the SAT solver has
to relearn clauses.
|
|
Some quick background: CVC4 has two primary contexts: the assertion
context (which corresponds to the push/pops issued by the user) and the
decision/SAT context (which corresponds to the push/pops done by the
user and each decision made by MiniSat.
Before 2cc0e2c6a691fb6d30ed8879832b49bc1f277d77, we had an additonal
push/pop when doing the actual solving. With these removed, it could
happen that we get a wrong result when doing incremental solving:
```
...
; check-sat call returns sat, the decision level in the SAT solver is
; non-zero at this point
(check-sat)
; Solver::addClause_() determines that we cannot add the clause to the
; SAT solver directly because the decision level is non-zero (i.e. the
; clause is treated like a theory lemma), thus it is added to the lemmas
; list.
(assert false)
; The solver stores some information about the current state, including
; the value of the "ok" flag in Solver, which indicates whether the
; current constraints are unsatisfiable. Note that "ok" is true at this
; point.
(push)
; Now, in Solver::updateLemmas(), we add clauses from the lemmas list.
; The problem is that empty clauses (which "false" is after removing "false"
; literals) and unit clauses are not added like normal clauses. Instead,
; the empty clause, essentially just results in the "ok" flag being set to
; false (unit clauses are added to the decision trail).
(check-sat)
; Here, the solver restores the information stored during
; (push), including the "ok" flag, which is now true again.
(pop)
; At this point, the solver has "forgotten" about (assert false) since
; the "ok" flag is back to true and it answers sat.
(check-sat)
```
There are multiple ways to look at the problem and to fix it:
- One could argue that an input assertion should always be added
directly to the clauses in the SAT solver, i.e. the solver should
always be at decision level 0 when we are adding input clauses. The
advantage of this is that it is relatively easy to implement,
corresponds to what the original MiniSat code does (it calls
Solver::cancelUntil(0) after solving), and is no worse than what we had
before commit 2cc0e2c6a691fb6d30ed8879832b49bc1f277d77 (push/pop do a
strict superset of what resetting the decision at the current assertion
level does). The disadvantage is that we might throw away some useful work.
- One could argue that is fine that (assert false) is treated like a
theory lemma. The advantage being that it might result in more efficient
solving and the disadvantage being that it is much trickier to
implement.
Unfortunately, it is not quite clear from the code what the original
intention was but we decided to implement the first solution. This
commit exposes a method in MiniSat to reset the decisions at the current
assertion level. We call this method from SmtEngine after solving.
Resetting the decisions directly after solving while we are still in
MiniSat does not work because it causes issues with datastructures in
the SMT solver that use the decision context (e.g.
TheoryEngine::d_incomplete seems to be reset too early, resulting in us
answering sat instead of unknown).
|
|
In `addClause_()`, we were checking the condition `ca[confl].size() ==
1` regardless if proofs were enabled or not, even though both branches
of the if statement only do something when proofs are enabled. This lead
to issue #1978 occurring even when not generating proofs. Note: This
commit is *not* a fix for #1978 but it makes sure that the issue does
not occur when not generating proofs/unsat cores.
|
|
|
|
If analyzeFinal() in BVMiniSat was called with a literal that also
occurred in the trail, it could happen that the set of assumptions that
led to the assignment of `p` contained `p` twice.
BitVectorProof::endBVConflict would then register that conflict with the
duplicate literal but at the same time compute a conflict expression
with the duplicate removed. LFSCSatProof::printAssumptionsResolution
would then output two resolutions for the same literal, which made the
simplify_clause side condition fail because it couldn't find the literal
in the clause anymore after the first removal.
The fix simply avoid adding `p` to the set of assumptions if it is found
in the trail. In this situation, `p` is guaranteed to be in the set of
assumptions already because it has been added at the beginning of
analyzeFinal(). This issue originally came up in PR #1384. With this fix
the regression tests pass in #1384.
|
|
While reviewing #1695, I realized that bvminisat is leaking memory for
each call to setNotify(). This commit uses std::unique_ptr to fix the
issue. It also adds std::unique_ptr to manage d_minisat.
|
|
|
|
|
|
Cryptominisat has name conflicts with the other Minisat implementations since
the Minisat implementations export var_Undef, l_True, ... as macro whereas
Cryptominisat uses static const. In order to avoid these conflicts we
forward declare CMSat::SATSolver and include the cryptominisat header only
in cryptominisat.cpp.
Further, the helper functions are moved into an anonymous namespace in the .cpp file and functions that were not used are removed.
|
|
* Fixes --hide-zero-stats (and really skips the 0 values)
* Removes the additional newline after each statistic
* Introduces theory::getStatsPrefix(TheoryId) to generate consistent
prefixes for statistics based on the theory id
(e.g., THEORY_BV -> "theory::bv").
|
|
Adds missing override keywords.
|
|
Clang formatted sat_solver_factor.cpp
|
|
Removing more miscellaneous throw specifiers. Also fixing the spelling of amount in several places.
|
|
Also, moved implementation of BVMinisatSatSolver::MinisatNotify::notify to .cpp file.
|
|
|
|
* Replacing __gnu_cxx::hash_map with std::unordered_map.
* Replacing __gnu_cxx::hash_set with std::unordered_set.
* Replacing __gnu_cxx::hash with std::hash.
* Adding missing includes.
|
|
|
|
|
|
This commit fixes bug 821. As written in the description of the bug, the issue
is that `id` is not being set on one of the paths in addClause(), specifically
in the case where all but one literal are assigned false and the remaining
literal is assigned true. In that case, we are not actually adding anything and
set the `id` to `ClauseIdUndef`.
|
|
|
|
Boolean terms, treats distinguished BOOLEAN_TERM_VARIABLE kind as theory literal. Fixes bugs 597, 604, 651, 652, 691, 694. Add regressions.
|
|
|
|
This reverts commit 3395c5c13cd61d98aec0d9806e3b9bc3d707968a, reversing
changes made to 5f415d4585134612bc24e9a823289fee35541a01.
|
|
Conflicts:
src/options/quantifiers_options
|
|
context/ classes, and their subclasses. Fixes compilation issues with clang 3.5 and -std=c++11 'exception specification of overriding function is more lax than base version' for a couple of different classes.
|
|
|
|
Record a separate recipe for each conjunct, but have as the "original lemma" in this recipe the complete conjunction, so that we can report this to the theory solver later, if asked.
Refactoring: instead of propagating the proof recipes from the theory engine to the prop engine and cnf stream to be registered there, just register them at the theory engine - as the prop engine and cnf stream don't change them.
|
|
Conflicts:
proofs/signatures/Makefile.am
src/Makefile.am
src/expr/datatype.cpp
src/options/datatypes_options
src/options/options_template.cpp
src/options/quantifiers_options
src/proof/arith_proof.cpp
src/proof/arith_proof.h
src/proof/array_proof.cpp
src/proof/array_proof.h
src/proof/bitvector_proof.cpp
src/proof/bitvector_proof.h
src/proof/cnf_proof.cpp
src/proof/cnf_proof.h
src/proof/proof_manager.cpp
src/proof/proof_manager.h
src/proof/sat_proof.h
src/proof/sat_proof_implementation.h
src/proof/skolemization_manager.h
src/proof/theory_proof.cpp
src/proof/theory_proof.h
src/proof/uf_proof.cpp
src/proof/uf_proof.h
src/prop/cnf_stream.cpp
src/prop/cnf_stream.h
src/prop/minisat/core/Solver.cc
src/prop/prop_engine.cpp
src/prop/prop_engine.h
src/prop/theory_proxy.cpp
src/smt/smt_engine.cpp
src/smt/smt_engine_check_proof.cpp
src/theory/arrays/array_proof_reconstruction.cpp
src/theory/arrays/theory_arrays.cpp
src/theory/bv/eager_bitblaster.cpp
src/theory/bv/lazy_bitblaster.cpp
src/theory/datatypes/theory_datatypes.cpp
src/theory/quantifiers/alpha_equivalence.cpp
src/theory/quantifiers/candidate_generator.cpp
src/theory/quantifiers/candidate_generator.h
src/theory/quantifiers/ce_guided_single_inv.cpp
src/theory/quantifiers/ceg_instantiator.cpp
src/theory/quantifiers/conjecture_generator.cpp
src/theory/quantifiers/equality_infer.cpp
src/theory/quantifiers/equality_infer.h
src/theory/quantifiers/inst_match_generator.cpp
src/theory/quantifiers/inst_propagator.cpp
src/theory/quantifiers/inst_propagator.h
src/theory/quantifiers/inst_strategy_e_matching.cpp
src/theory/quantifiers/inst_strategy_e_matching.h
src/theory/quantifiers/instantiation_engine.cpp
src/theory/quantifiers/model_builder.cpp
src/theory/quantifiers/model_engine.cpp
src/theory/quantifiers/quant_conflict_find.cpp
src/theory/quantifiers/quant_conflict_find.h
src/theory/quantifiers/quant_split.cpp
src/theory/quantifiers/quant_util.cpp
src/theory/quantifiers/quantifiers_rewriter.cpp
src/theory/quantifiers/quantifiers_rewriter.h
src/theory/quantifiers/term_database.cpp
src/theory/quantifiers/term_database.h
src/theory/quantifiers/trigger.cpp
src/theory/quantifiers/trigger.h
src/theory/quantifiers_engine.cpp
src/theory/quantifiers_engine.h
src/theory/sets/kinds
src/theory/sets/theory_sets_private.cpp
src/theory/sets/theory_sets_private.h
src/theory/sets/theory_sets_rewriter.cpp
src/theory/sets/theory_sets_type_rules.h
src/theory/strings/theory_strings.cpp
src/theory/strings/theory_strings.h
src/theory/theory_engine.cpp
src/theory/theory_engine.h
src/theory/uf/equality_engine.cpp
test/regress/regress0/fmf/Makefile.am
test/regress/regress0/quantifiers/Makefile.am
test/regress/regress0/strings/Makefile.am
test/regress/regress0/sygus/Makefile.am
test/regress/regress0/sygus/max2-univ.sy
|
|
|
|
This reverts commit 89ba584531115b7f6d47088d7614368ea05ab9d8.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Breaking an edge between the sat solver and command.h.
|
|
- Options
-- Added the new option attribute :notify. One can get a notify() call on the Listener after a the option's value is updated. This is the new preferred way to achieve dynamic dispatch for options.
-- Removed SmtOptionsHandler and pushed its functionality into OptionsHandler and Listeners.
-- Added functions to Options for registering listeners of the notify calls.
-- Changed a number of options to use the new listener infrastructure.
-- Fixed a number of warnings in options.
-- Added the ArgumentExtender class to better capture how arguments are inserted while parsing options and ease memory management. Previously this was the "preemptGetopt" procedure.
-- Moved options/options_handler_interface.{cpp,h} to options/options_handler.{cpp,h}.
- Theories
-- Reimplemented alternative theories to use a datastructure stored on TheoryEngine instead of on Options.
- Ostream Handling:
-- Added new functionality that generalized how ostreams are opened, options/open_stream.h.
-- Simplified the memory management for different ostreams, smt/managed_ostreams.h.
-- Had the SmtEnginePrivate manage the memory for the ostreams set by options.
-- Simplified how the setting of ostreams are updated, smt/update_ostream.h.
- Configuration and Tags:
-- Configuration can now be used during predicates and handlers for options.
-- Moved configuration.{cpp,h,i} and configuration_private.h from util/ into base/.
-- Moved {Debug,Trace}_tags.* from being generated in options/ into base/.
- cvc4_private.h
-- Upgraded #warning's in cvc4_private.h and cvc4_private_library.h to #error's.
-- Added public first-order (non-templatized) member functions for options get and set the value of options outside of libcvc4. Fixed all of the use locations.
-- Made lib/lib/clock_gettime.h a cvc4_private_library.h header.
- Antlr
-- Fixed antlr and cvc4 macro definition conflicts that caused warnings.
- SmtGlobals
-- Refactored replayStream and replayLog out of SmtGlobals.
-- Renamed SmtGlobals to LemmaChannels and moved the implementation into smt_util/lemma_channels.{h,cpp}.
|
|
|
|
Listeners for reporting hard and soft resource out() events.
|
|
- The functionality the get the StatisticsRegistry attached to the SmtEngine was previously through StatisticsRegistry::current(). This is the dominant StatisticsRegistry in the code. (There is another StatisticsRegistry attached to the NodeManager.) Having this be a static function on StatisticsRegistry requires the use of an SmtEngine in the wrong compilation unit.
- Usages of StatisticsRegistry::current() that were visible in prop/{bvminisat,minisat} has been removed. A pointer to the relevant StatisticsRegistry should be passed instead into the constructor.
- The function StatisticsRegistry::current() has been replaced by SmtScope::currentStatisticsRegistry(). SmtScope is in the libcvc4 package, where SmtEngine is available in the compilation unit.
- The function smtStatisticsRegistry() is a synonym for SmtScope::currentStatisticsRegistry() in smt/smt_statistics_registry.h. This header has fewer include dependencies than the one for SmtScope.
- Correspondingly, the static functions StatisticsRegistry::{registerStat, unregisterStat} have been removed. One should instead use smtStatisticsRegistry()->{registerStat,unregisterStat} instead.
- The KEEP_STATISTIC macro has been moved into smt/smt_statistics_registry.h.
- Documents the reason StatisticsRegistry is CVC4_PUBLIC. This lets me remove the warning I added.
- Removing most operators for timespec from statistics_registry.h file. These a bit error prone in clang.
- Most of the really confusing ifdef's in util/statistics_registry.h are gone.
|
|
- The options replayStream, lemmaInputChannel, lemmaOutputChannel have been removed due to their datatypes. These datatypes were previously pointers to types that were not usable from the options/ library.
- The option replayLog has been removed due to inconsistent memory management.
- SmtGlobals is a class that wraps a pointer to each of these removed options. These can each be set independently.
- There is a single SmtGlobals per SmtEngine with the lifetime of the SmtEngine.
- A pointer to this is freely given to the user of an SmtEngine to parameterize the solver after construction.
- Selected classes have been given a copy of this pointer in their constructors.
- Removed the dependence on Node from Result. Moving Result back into util/.
|
|
- Adding a script contrib/test_install_headers.h that tests whether one can include all cvc4_public headers. CVC4 can pass this test after this commit.
- Making lib/{clock_gettime.h,ffs.h,strtok_r.h} cvc4_private.
- Making prop/sat_solver_factory.h cvc4_private.
- Moving the expr iostream manipulators into their own files: expr_iomanip.{h,cpp}.
- Setting the generated *_options.h files back to being cvc4_private.
-- Removing the usage of options/expr_options.h from expr.h.
-- Removing the include of base_options.h from options.h.
- Cleaning up CPP macros in cvc4_public headers.
-- Changing the ROLL macro in floatingpoint.h into an inline function.
-- Removing the now unused flag -D__BUILDING_STATISTICS_FOR_EXPORT.
|
|
- Splitting the two instances of CheckArgument. The template version is now always defined in base/exception.h and is available in a cvc4_public header. This version has lost its variadic version (due to swig not supporting va_list's). The CPP macro version has been renamed PrettyCheckArgument. (Taking suggestions for a better name.) This is now only defined in base/cvc4_assert.h. Only use this in cvc4_private headers and in .cpp files that can use cvc4_private headers. To use a variadic version of CheckArguments, outside of this scope, you need to duplicate this macro locally. See cvc3_compat.cpp for an example.
- Making fitsSignedInt() and fitsUnsignedInt() work more robustly for CLN on 32 bit systems.
- Refactoring ArrayStoreAll to avoid potential problems with circular header inclusions.
- Changing some headers to use iosfwd when possible.
|