mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-08-24 03:57:51 +00:00
Code Activity

update todos, working on assignment minimization

Browse source
This commit is contained in:
Nikolaj Bjorner 2021-09-22 13:29:36 -07:00
parent 3447d80310
commit c82bbaad7d
8 changed files with 51 additions and 55 deletions
Download patch file Download diff file Expand all files Collapse all files

29
src/math/polysat/conflict.cpp
View file

@ -12,21 +12,12 @@ Author:
Notes:
TODO: constraints containing v could be tracked incrementally when constraints are added/removed using an index.
TODO: try a final core reduction step or other core minimization
TODO: If we have e.g. 4x+y=2 and y=0, then we have a conflict no matter the value of x, so we should drop x=? from the core.
(works currently if x is unassigned; for other cases we would need extra info from constraint::is_currently_false)
TODO: build_lemma:
note that we may have added too many variables: e.g., y disappears in x*y if x=0
TODO: keep is buggy. The assert
SASSERT(premise.is_currently_true(s()) || premise.bvalue(s()) == l_true);
does not necessarily hold. A saturation premise could be inserted that is a resolvent that evaluates to false
and therefore not a current Boolean literal on the search stack.
TODO: revert(pvar v) is too weak.
It should apply saturation rules currently only available for propagated values.
@ -46,9 +37,7 @@ Notes:
namespace polysat {
conflict::conflict(solver& s):s(s) {
ex_engines.push_back(alloc(ex_polynomial_superposition));
for (auto* engine : ex_engines)
engine->set_solver(s);
ex_engines.push_back(alloc(ex_polynomial_superposition, s));
ve_engines.push_back(alloc(ve_reduction));
inf_engines.push_back(alloc(inf_saturate));
for (auto* engine : inf_engines)
@ -77,7 +66,6 @@ namespace polysat {
m_literals.reset();
m_vars.reset();
m_conflict_var = null_var;
m_saturation_premises.reset();
m_bailout = false;
SASSERT(empty());
}
@ -146,16 +134,17 @@ namespace polysat {
m_constraints.push_back(c);
}
// NOTE: maybe we should skip intermediate steps and just collect the leaf premises for c?
// Ensure that c is assigned and justified
void conflict::insert(signed_constraint c, vector<signed_constraint> const& premises) {
insert(c);
// NOTE: maybe we should skip intermediate steps and just collect the leaf premises for c?
clause_builder c_lemma(s);
for (auto premise : premises) {
LOG_H3("premise: " << premise);
keep(premise);
SASSERT(premise->has_bvar());
SASSERT(premise.bvalue(s) == l_true);
// otherwise the propagation doesn't make sense
c_lemma.push(~premise.blit());
}
c_lemma.push(c.blit());
@ -210,12 +199,12 @@ namespace polysat {
* insert it (and recursively, its premises) into \Gamma
*/
void conflict::keep(signed_constraint c) {
if (!c->has_bvar()) {
remove(c);
cm().ensure_bvar(c.get());
insert(c);
}
if (c->has_bvar())
return;
LOG_H3("keeping: " << c);
remove(c);
cm().ensure_bvar(c.get());
insert(c);
}
clause_builder conflict::build_lemma() {