mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-04-27 10:55:50 +00:00
Code Activity

fixes to sat.euf ematching #5573

Browse source
This commit is contained in:
Nikolaj Bjorner 2021-10-16 15:52:37 -07:00
parent f78546cd7c
commit 115203e87c
6 changed files with 98 additions and 95 deletions
Download patch file Download diff file Expand all files Collapse all files

6
src/sat/smt/q_clause.cpp
View file

@ -32,8 +32,8 @@ namespace q {
<< mk_bounded_pp(rhs, m, 2);
}
std::ostream& binding::display(euf::solver& ctx, unsigned num_nodes, std::ostream& out) const {
for (unsigned i = 0; i < num_nodes; ++i)
std::ostream& binding::display(euf::solver& ctx, std::ostream& out) const {
for (unsigned i = 0; i < size(); ++i)
out << ctx.bpp((*this)[i]) << " ";
return out;
}
@ -46,7 +46,7 @@ namespace q {
if (!b)
return out;
do {
b->display(ctx, num_decls(), out) << "\n";
b->display(ctx, out) << "\n";
b = b->next();
}
while (b != m_bindings);

2
src/sat/smt/q_clause.h
View file

@ -79,7 +79,7 @@ namespace q {
euf::enode* operator[](unsigned i) const { return m_nodes[i]; }
std::ostream& display(euf::solver& ctx, unsigned num_nodes, std::ostream& out) const;
std::ostream& display(euf::solver& ctx, std::ostream& out) const;
unsigned size() const { return c->num_decls(); }

9
src/sat/smt/q_ematch.cpp
View file

@ -270,13 +270,14 @@ namespace q {
}
void ematch::on_binding(quantifier* q, app* pat, euf::enode* const* _binding, unsigned max_generation, unsigned min_gen, unsigned max_gen) {
TRACE("q", tout << "on-binding " << mk_pp(q, m) << "\n";);
unsigned idx = m_q2clauses[q];
clause& c = *m_clauses[idx];
bool new_propagation = false;
binding* b = alloc_binding(c, pat, _binding, max_generation, min_gen, max_gen);
if (!b)
return;
TRACE("q", b->display(ctx, tout << "on-binding " << mk_pp(q, m) << "\n") << "\n";);
if (false && propagate(false, _binding, max_generation, c, new_propagation))
return;
@ -558,7 +559,7 @@ namespace q {
m_mam->propagate();
bool propagated = flush_prop_queue();
if (m_qhead >= m_clause_queue.size())
return m_inst_queue.propagate();
return m_inst_queue.propagate() || propagated;
ctx.push(value_trail<unsigned>(m_qhead));
ptr_buffer<binding> to_remove;
for (; m_qhead < m_clause_queue.size(); ++m_qhead) {
@ -613,8 +614,8 @@ namespace q {
return true;
for (unsigned i = 0; i < m_clauses.size(); ++i)
if (m_clauses[i]->m_bindings)
std::cout << "missed propagation " << i << "\n";
IF_VERBOSE(0, verbose_stream() << "missed propagation " << i << "\n");
TRACE("q", tout << "no more propagation\n";);
return false;
}

165
src/sat/smt/q_mam.cpp
View file

@ -409,16 +409,17 @@ namespace q {
unsigned m_num_args; //!< we need this information to avoid the nary *,+ crash bug
bool m_filter_candidates;
unsigned m_num_regs;
unsigned m_num_choices;
instruction * m_root;
unsigned m_num_choices = 0;
instruction * m_root = nullptr;
enode_vector m_candidates;
unsigned m_qhead = 0;
#ifdef Z3DEBUG
egraph * m_egraph;
egraph * m_egraph = nullptr;
svector<std::pair<quantifier*, app*>> m_patterns;
#endif
#ifdef _PROFILE_MAM
stopwatch m_watch;
unsigned m_counter;
unsigned m_counter = 0;
#endif
friend class compiler;
friend class code_tree_manager;
@ -492,13 +493,7 @@ namespace q {
m_root_lbl(lbl),
m_num_args(num_args),
m_filter_candidates(filter_candidates),
m_num_regs(num_args + 1),
m_num_choices(0),
m_root(nullptr) {
DEBUG_CODE(m_egraph = nullptr;);
#ifdef _PROFILE_MAM
m_counter = 0;
#endif
m_num_regs(num_args + 1) {
(void)m_lbl_hasher;
}
@ -546,16 +541,40 @@ namespace q {
return m_root;
}
void add_candidate(enode * n) {
void add_candidate(euf::solver& ctx, enode * n) {
m_candidates.push_back(n);
ctx.push(push_back_trail<enode*, false>(m_candidates));
}
void unmark(unsigned head) {
for (unsigned i = m_candidates.size(); i-- > head; ) {
enode* app = m_candidates[i];
if (app->is_marked1())
app->unmark1();
}
}
struct scoped_unmark {
unsigned m_qhead;
code_tree* t;
scoped_unmark(code_tree* t) : m_qhead(t->m_qhead), t(t) {}
~scoped_unmark() { t->unmark(m_qhead); }
};
bool has_candidates() const {
return !m_candidates.empty();
return m_qhead < m_candidates.size();
}
void reset_candidates() {
m_candidates.reset();
void save_qhead(euf::solver& ctx) {
ctx.push(value_trail<unsigned>(m_qhead));
}
enode* next_candidate() {
if (m_qhead < m_candidates.size())
return m_candidates[m_qhead++];
else
return nullptr;
}
enode_vector const & get_candidates() const {
@ -1987,24 +2006,18 @@ namespace q {
m_backtrack_stack.resize(t->get_num_choices());
}
struct scoped_unmark {
code_tree* t;
scoped_unmark(code_tree* t) : t(t) {}
~scoped_unmark() {
for (enode* app : t->get_candidates())
if (app->is_marked1())
app->unmark1();
}
};
void execute(code_tree * t) {
if (!t->has_candidates())
return;
TRACE("trigger_bug", tout << "execute for code tree:\n"; t->display(tout););
init(t);
t->save_qhead(ctx);
enode* app;
if (t->filter_candidates()) {
scoped_unmark _unmark(t);
for (unsigned i = 0; i < t->get_candidates().size() && !ctx.resource_limits_exceeded(); ++i) {
enode* app = t->get_candidates()[i];
TRACE("trigger_bug", tout << "candidate\n" << mk_ismt2_pp(app->get_expr(), m) << "\n";);
code_tree::scoped_unmark _unmark(t);
while ((app = t->next_candidate()) && !ctx.resource_limits_exceeded()) {
TRACE("trigger_bug", tout << "candidate\n" << ctx.bpp(app) << "\n";);
if (!app->is_marked1() && app->is_cgr()) {
execute_core(t, app);
app->mark1();
@ -2012,9 +2025,8 @@ namespace q {
}
}
else {
for (unsigned i = 0; i < t->get_candidates().size() && !ctx.resource_limits_exceeded(); ++i) {
enode* app = t->get_candidates()[i];
TRACE("trigger_bug", tout << "candidate\n" << mk_ismt2_pp(app->get_expr(), m) << "\n";);
while ((app = t->next_candidate()) && !ctx.resource_limits_exceeded()) {
TRACE("trigger_bug", tout << "candidate\n" << ctx.bpp(app) << "\n";);
if (app->is_cgr())
execute_core(t, app);
}
@ -2477,17 +2489,18 @@ namespace q {
case YIELD1:
m_bindings[0] = m_registers[static_cast<const yield *>(m_pc)->m_bindings[0]];
#define ON_MATCH(NUM) \
#define ON_MATCH(NUM) \
m_max_generation = std::max(m_max_generation, get_max_generation(NUM, m_bindings.begin())); \
if (!m.inc()) { \
return false; \
} \
if (!m.inc()) \
return false; \
\
m_mam.on_match(static_cast<const yield *>(m_pc)->m_qa, \
static_cast<const yield *>(m_pc)->m_pat, \
NUM, \
m_bindings.begin(), \
m_max_generation)
SASSERT(static_cast<const yield *>(m_pc)->m_qa->get_decl_sort(0) == m_bindings[0]->get_expr()->get_sort());
ON_MATCH(1);
goto backtrack;
@ -3059,29 +3072,9 @@ namespace q {
// temporary field used to collect candidates
ptr_vector<path_tree> m_todo;
enode * m_root { nullptr }; // temp field
enode * m_other { nullptr }; // temp field
bool m_check_missing_instances { false };
class pop_to_match : public trail {
mam_impl& i;
public:
pop_to_match(mam_impl& i):i(i) {}
void undo() override {
code_tree* t = i.m_to_match.back();
t->reset_candidates();
i.m_to_match.pop_back();
}
};
class reset_new_patterns : public trail {
mam_impl& i;
public:
reset_new_patterns(mam_impl& i):i(i) {}
void undo() override {
i.m_new_patterns.reset();
}
};
enode * m_root = nullptr; // temp field
enode * m_other = nullptr; // temp field
bool m_check_missing_instances = false;
enode_vector * mk_tmp_vector() {
enode_vector * r = m_pool.mk();
@ -3094,14 +3087,14 @@ namespace q {
}
void add_candidate(code_tree * t, enode * app) {
if (t) {
TRACE("mam_candidate", tout << "adding candidate:\n" << mk_ll_pp(app->get_expr(), m););
if (!t->has_candidates()) {
ctx.push(pop_to_match(*this));
m_to_match.push_back(t);
}
t->add_candidate(app);
if (!t)
return;
TRACE("q", tout << "candidate " << t << " " << ctx.bpp(app) << "\n";);
if (!t->has_candidates()) {
ctx.push(push_back_trail<code_tree*, false>(m_to_match));
m_to_match.push_back(t);
}
t->add_candidate(ctx, app);
}
void add_candidate(enode * app) {
@ -3678,14 +3671,20 @@ namespace q {
}
}
void match_new_patterns() {
unsigned m_new_patterns_qhead = 0;
void propagate_new_patterns() {
if (m_new_patterns_qhead >= m_new_patterns.size())
return;
ctx.push(value_trail<unsigned>(m_new_patterns_qhead));
TRACE("mam_new_pat", tout << "matching new patterns:\n";);
m_tmp_trees_to_delete.reset();
for (auto const& kv : m_new_patterns) {
for (; m_new_patterns_qhead < m_new_patterns.size(); ++m_new_patterns_qhead) {
if (!m.inc())
break;
quantifier * qa = kv.first;
app * mp = kv.second;
auto [qa, mp] = m_new_patterns[m_new_patterns_qhead];
SASSERT(m.is_pattern(mp));
app * p = to_app(mp->get_arg(0));
func_decl * lbl = p->get_decl();
@ -3717,7 +3716,6 @@ namespace q {
m_tmp_trees[lbl_id] = nullptr;
dealloc(tmp_tree);
}
m_new_patterns.reset();
}
public:
@ -3753,7 +3751,7 @@ namespace q {
return; // ignore multi-pattern containing ground pattern.
update_filters(qa, mp);
m_new_patterns.push_back(qp_pair(qa, mp));
ctx.push(reset_new_patterns(*this));
ctx.push(push_back_trail<qp_pair, false>(m_new_patterns));
// The matching abstract machine implements incremental
// e-matching. So, for a multi-pattern [ p_1, ..., p_n ],
// we have to make n insertions. In the i-th insertion,
@ -3764,7 +3762,6 @@ namespace q {
void reset() override {
m_trees.reset();
m_new_patterns.reset();
m_is_plbl.reset();
m_is_clbl.reset();
reset_pp_pc();
@ -3779,22 +3776,18 @@ namespace q {
return out;
}
void propagate() override {
TRACE("trigger_bug", tout << "match\n"; display(tout););
if (m_to_match_head >= m_to_match.size())
void propagate_to_match() {
if (m_to_match_head >= m_to_match.size())
return;
ctx.push(value_trail<unsigned>(m_to_match_head));
for (; m_to_match_head < m_to_match.size(); ++m_to_match_head) {
code_tree* t = m_to_match[m_to_match_head];
if (t->has_candidates()) {
m_interpreter.execute(t);
t->reset_candidates();
}
}
if (!m_new_patterns.empty()) {
match_new_patterns();
m_new_patterns.reset();
}
for (; m_to_match_head < m_to_match.size(); ++m_to_match_head)
m_interpreter.execute(m_to_match[m_to_match_head]);
}
void propagate() override {
TRACE("trigger_bug", tout << "match\n"; display(tout););
propagate_to_match();
propagate_new_patterns();
}
void rematch(bool use_irrelevant) override {

9
src/sat/smt/q_queue.cpp
View file

@ -149,6 +149,12 @@ namespace q {
if (false && em.propagate(true, f.nodes(), gen, *f.c, new_propagation))
return;
#if 0
std::cout << mk_pp(q, m) << "\n";
std::cout << num_bindings << "\n";
for (unsigned i = 0; i < num_bindings; ++i)
std::cout << mk_pp(f[i]->get_expr(), m) << " " << mk_pp(f[i]->get_sort(), m) << "\n";
#endif
auto* ebindings = m_subst(q, num_bindings);
for (unsigned i = 0; i < num_bindings; ++i)
ebindings[i] = f[i]->get_expr();
@ -161,6 +167,9 @@ namespace q {
stat->inc_num_instances();
m_stats.m_num_instances++;
// f.display(ctx, std::cout << mk_pp(f.q(), m) << "\n" << instance << "\n") << "\n";
euf::solver::scoped_generation _sg(ctx, gen);
sat::literal result_l = ctx.mk_literal(instance);

2
src/smt/smt_context_pp.cpp
View file

@ -676,7 +676,7 @@ namespace smt {
std::ostream& operator<<(std::ostream& out, enode_pp const& p) {
ast_manager& m = p.ctx.get_manager();
enode* n = p.n;
return out << "[#" << n->get_owner_id() << " " << mk_bounded_pp(n->get_expr(), m) << "]";
return out << n->get_owner_id() << ": " << mk_bounded_pp(n->get_expr(), m);
}
std::ostream& operator<<(std::ostream& out, enode_eq_pp const& p) {