| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
manually brought these changes over. Changed the tests used by test/regress/regress0/arith/integers/Makefile.am to be 15 of the more interesting tests. Did a bit of cleanup on TheoryArith to eliminate a warning and remove dead code.
|
|
r2650 to r2779.
- This excludes revision 2777. This revision had some strange performance implications and was delaying the merge.
- This includes the new DioSolver. The DioSolver can discover conflicts, produce substitutions, and produce cuts.
- The DioSolver can be disabled at command line using --disable-dio-solver.
- This includes a number of changes to the arithmetic normal form.
- The Integer class features a number of new number theoretic function.
- This commit includes a few rather loud warning. I will do my best to take care of them today.
|
|
|
|
|
|
resolves bug 289. Adds failing tests to regress1.
|
|
merged-in
* since two theories can propagate the same literal minisat now explicitly checks that a propagated literal has not been asserted yet
|
|
now supports propagating equalities when a slack variable corresponding to a difference of shared terms must be 0. Similarly disequalities are propagated when these variables cannot be zero.
|
|
arithmetic normal form when assertions are enabled.
|
|
* OutputChannel::lemma() now returns an unsigned int. This facility isn't functional yet, but the signature is there. For now, it always returns the current user level (which is "correct" from the interface point of view, but not what we want).
* Pseudobooleans disabled. This should fix some quantifier benchmarks Andy's been working with on the quantifiers2 branch.
* --limit / --time-limit options renamed --rlimit and --tlimit.
There may be slowdown from disabling pseudobooleans.
|
|
into trunk. Arithmetic should now be closer to being able to support push and pop.
|
|
|
|
and model gen also.
I also expect this commit to fix bug #273.
No performance change is expected on regressions with this commit, see
http://goedel.cims.nyu.edu/regress-results/compare_jobs.php?job_id=2871&reference_id=2863
|
|
MacOS-production-dynamic builds)
|
|
|
|
|
|
|
|
Dumping infrastructure. Can dump preprocessed queries and clauses. Can
also dump queries (for testing with another solver) to see if any conflicts
are missed, T-propagations are missed, all lemmas are T-valid, etc. For a
full list of options see --dump=help.
CUDD building much cleaner.
Documentation and assertion fixes.
Printer improvements, printing of commands in language-defined way, etc.
Typechecker stuff in expr package now autogenerated, no need to manually
edit the expr package when adding a new theory.
CVC3 compatibility layer (builds as libcompat).
SWIG detection and language binding support (infrastructure).
Support for some Z3 extended commands (like datatypes) in SMT-LIBv2 mode
(when not in compliance mode).
Copyright and file headers regenerated.
|
|
infrastructure, and takes care not to affect CVC4's performance on LRA
benchmarks.
|
|
SAT solver on every backtrack
* Updated UF to handle the context dependent pre-registration
* Additionally some small changes in order to satisfy warnings of the eclipse code analysis tool
|
|
The theoryOf is not all in one place, theory::theoryOf. The uninterpreted sorts belong to the builtin theory and are dispatched to the apropriate theory (QF_UF, QF_AX) through theoryOf based on the setting in the Theory class.
|
|
1) arithmetic should check for subterms when solving equations, for instance x = if b then x + 1 else x -1 is not a valid substitution
2) a memory problem in minisat - explanations are constructed during conflict analysis, so the clause database might resize and relocate, which invalidates any references to clauses
|
|
|
|
|
|
more than one "real" theory (not BUILTIN or BOOL) active
|
|
theory of arithmetic.
* This code has been partially tested. (My testing situation is currently not so great.) The code for testing not preregistering equalities can be compile time enabled by setting the boolean turnOffEqualityPreRegister. Don't be shocked by slowdowns or failures. This does pass make regress as well as a fresh checkout does. (The Mac version has issues.)
* I need to disable the permanent row removal heuristic by default. We need to discuss why this needs to happen. We should probably detect pure QF_LRA/QF_RDL problems and enable this when this can safely be done.
* I have disabled the arithmetic rewrite equality flag. This code needs to be added to the parser.
* For all of the above changes, I have annotated the code with the key word BREADCRUMB.
* I have renamed ArithUnatePropagator to ArithAtomDatabase.
|
|
easier replication of experiment)
|
|
|
|
* Preprocessing-time, non-clausal, Boolean simplification round to
support "quasi-non-linear rewrites" as discussed at last few meetings.
* --simplification=none is the default for now, but we'll probably
change that to --simplification=incremental. --simplification=batch
is also a possibility. See --simplification=help for details.
* RecursionBreaker<T> now uses a hash set for the seen trail.
* Fixes to TLS stuff to support that.
* Fixes to theory and SmtEngine documentation.
* Fixes to stream indentation.
* Other miscellaneous stuff.
|
|
* new "well-foundedness" type property (like cardinality) specified in
Theory kinds files; specifies well-foundedness and a ground term
* well-foundedness / finite checks in Datatypes now superseded by type
system isFinite(), isWellFounded(), mkGroundTerm().
* new "RecursionBreaker" template class, a convenient class that keeps
a "seen" trail without you having to pass it around (which is
difficult in cases of mutual recursion) of the idea of passing
around a "seen" trail
|
|
* Type::getCardinality() returns the cardinality for for all types.
Theories give a cardinality in the their kinds file. For
cardinalities that depend on a type argument, a "cardinality
computer" function is named in the kinds file, which takes a
TypeNode and returns its cardinality.
* There's a bitmap for the set of "active theories" in the
TheoryEngine. Theories become "active" when a term that is owned by
them, or whose type is owned by them, is pre-registered (run CVC4
with --verbose to see theory activation). Non-active theories don't
get any calls for check() or propagate() or anything, and if we're
running in single-theory mode, the shared term manager doesn't have
to get involved. This is really important for get() performance
(which can only skimp on walking the entire sub-DAG only if the
theory doesn't require it AND the shared term manager doesn't
require it).
* TheoryEngine now does not call presolve(), registerTerm(),
notifyRestart(), etc., on a Theory if that theory doesn't declare
that property in its kinds file. To avoid coding errors,
mktheorytraits greps the theory header and gives warnings if:
+ the theory appears to declare one of the functions (check,
propagate, etc.) that isn't listed among its kinds file properties
(but probably should be)
+ the theory appears NOT to declare one of the functions listed in
its kinds file properties
* some bounded token stream work
|
|
|
|
- This adds code for bounds refinement, and conflict weakening.
- This adds util/boolean_simplification.h.
- This adds a propagation manager to theory of arithmetic.
- Propagation is disabled by default.
- Propagation can be enabled by the command line flag "--enable-arithmetic-propagation"
- Propagation interacts *heavily* with rewriting equalities, and will work best if the command line flag "--rewrite-arithmetic-equalities" is enabled.
|
|
|
|
called. The OutputChannel is now untouched by TheoryArith during preregistration.
|
|
* Makes Options an "omnipresent thread-local global" (like the notion
of the "current NodeManager" was already). Options::current() accesses
this structure.
* Removes Options from constructors and data structures everywhere
(this cleans up a lot of things).
* No longer uses StatisticsRegistry statically. An instance of the
registry is created and linked to a NodeManager.
* StatisticsRegistry::current() is similar to Options::current(), but
the pointer is stowed in the NodeManager (rather than stored)
* The static functions of StatisticsRegistry have been left, for backward
compatibility; they now use the "current" statistics registry.
* SmtEngine::getStatisticsRegistry() is a public accessor for the
registry; this is needed by main() to reach in and get the registry,
for flushing statistics at the end.
|
|
static flag in Options that the ArithRewriter uses to determine the equality rewriting policy.
|
|
|
|
are somewhat disparate but belonged on the same branch because they were
held back from trunk all for the same reason (to keep the trunk stable
for furious bitvector development). Dejan has now given me the go-ahead
for a merge.
=========================================
THIS COMMIT CHANGES THE THEORY INTERFACE!
=========================================
Theory constructors are expected to take an additional "Valuation*"
parameter that each Theory should send along to the base class
constructor. The base class Theory keeps the Valuation* in a
d_valuation field for use by it and by its derived classes.
Theory::getValue() no longer takes a Valuation* (it is expected
to use d_valuation instead). This allows other theory functions
to take advantage of getValue() for debugging or heuristic
purposes.
TODO BEFORE MERGE TO TRUNK:
****implement BitIterator find() in CDAttrHash<bool>.
Specifically:
* Added QF_BV support for SMT-LIB v2.
* Two adjustments to the theory interface as requested by Tim King:
1. As described above.
2. Theories now have const access to the fact queue through base
class functions facts_begin() and facts_end(); useful for
debugging.
* Added an "Asserted" attribute so that theories can check if something
has been asserted or not (and therefore not propagate it). However, this
has been disabled for now, pending more data on the overhead of it, and
pending discussion at the 3/25/2011 meeting.
* Do not define NDEBUG in MiniSat in assertion-enabled builds (so
that MiniSat asserts are evaluated).
* As a result of the new MiniSat assertions, some --incremental
regressions had to be disabled; also, some bitvectors ?!!
* Bug 71 is resolved by adding a specialization for CDAttrHash<> in the
attribute package.
* Fixes for some warnings flagged by clang.
* System tests have arrived! So far mainly infrastructure for having
system tests, but there is a system test aimed at improving code
coverage of the printer package.
* Minor other adjustments to documentation and coding to be more
conformant to CVC4 policy.
Tests have been performed to demonstrate that these changes have no or
negligible effect on performance. In particular, changing the
CDAttrHash<> doesn't have any real effect on performance or memory right
now, since there is only one context-dependent boolean flag (as soon
as another is added, the effect is noticeable but probably still slight).
|
|
importantly removes an unintentional line of code that had it pivoting more times than intended before rechecking the queue. Importantly, it does this without losing any examples with rewrite-equality enabled. This adds a parameter NUM_CHECKS which determines how many times the queue chould be checked during difference mode. A value of 10 for NUM_CHECKS has been empirically determined to be good in practice. See jobs 1815, 1824, 1825, 1821, 1814.
|
|
Collection mode, the arithmetic priority queue is maintained as a set. Compare jobs 1781 and 1782 for the expected performance change.
|
|
slowing down the search. (The effect can be seen in the difference between jobs 1765 and 1755). This happened during commit -r1480 when adding the ArithStaticLearner. This has been fixed.
|
|
|
|
|
|
- Adds ArithStaticLearner. Consolidates and cleans up the code for static learning in arithmetic. Static learning is now associated with a small amount of state between calls. This is used to track the data for the miplib trick. The goal is to make this inference work without relying on the fact that all of the miplib problem is asserted under the same AND node.
- This commit contains miscellaneous other arithmetic cleanup.
|
|
known that x \in {c_i}, then x is not in the interval (c_{i}, c_{i+1}) (assuming the c_i's are sorted). (Compare jobs 1742 and 1739 for the expected performance change on trunk. Compare jobs 1740 and 1738 for the expected performance change with the rewrite-equality patch.)
|
|
set of nodes that are asserted to the theory of the form (=> p_i (= x c_i)). If (or p_1 p_2 ...) is a tautology, then x \in {c_1, c_2, ...}. (This tautology check currently requires CUDD to be installed.) Right now all this does is assert x \leq max{c_i} and x \geq min{c_i}. (Compare jobs 1728 to 1626 for how this affects the miplib examples.)
|
|
against cudd libraries, the propositional_query class (in util/),
which uses cudd if it's available (and otherwise answers UNKNOWN for
all queries), and the arith theory support for it (currently disabled
per Tim's request, so he can clean it up).
Other changes include:
* contrib/debug-keys - script to print all used keys under Debug(), Trace()
* test/regress/run_regression - minor fix (don't export a variable)
* configure.ac - replace a comment removed by dejan's google perf commit
* some minor copyright/documentation updates, and minor changes to source
text to make 'clang --analyze' happy.
|
