mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-04-28 11:25:51 +00:00
Code Activity

don't have bv-ackerman influence simplification

Browse source 
previous scheme has Ackmerman module instrument main solver to backjump and simplify when reaching a threshold.
This destroys overall performance: simplification does many more things than invoking Ackerman axioms.
Having a dependency between simplification (in-processing) and depleting a priority queue of auxiliary axioms therefore hurts overall performance. It has to be decoupled. The current approach is now to empty the axiom queue on occasion.
It is still not ideal - it should be coupled with the search level - axioms don't survive higher levels where redundant clauses get garbage collected as they don't have a chance of being used.
This commit is contained in:
Nikolaj Bjorner 2022-08-21 15:24:30 -07:00
parent be0cd74c71
commit 17fc438476
3 changed files with 13 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

1
src/sat/sat_solver.h
View file

@ -714,7 +714,6 @@ namespace sat {
//
// -----------------------
public:
void set_should_simplify() { m_next_simplify = m_conflicts_since_init; }
bool_var_vector const& get_vars_to_reinit() const { return m_vars_to_reinit; }
bool is_probing() const { return m_is_probing; }
bool is_free(bool_var v) const { return m_free_vars.contains(v); }

12
src/sat/smt/bv_ackerman.cpp
View file

@ -57,8 +57,8 @@ namespace bv {
remove(other);
add_cc(v1, v2);
}
else if (other->m_count > m_propagate_high_watermark)
s.s().set_should_simplify();
else if (other->m_count > 2*m_propagate_high_watermark)
propagate();
}
void ackerman::used_diseq_eh(euf::theory_var v1, euf::theory_var v2) {
@ -76,8 +76,8 @@ namespace bv {
new_tmp();
gc();
}
if (other->m_count > m_propagate_high_watermark)
s.s().set_should_simplify();
if (other->m_count > 2*m_propagate_high_watermark)
propagate();
}
void ackerman::update_glue(vv& v) {
@ -137,6 +137,9 @@ namespace bv {
if (m_num_propagations_since_last_gc <= s.get_config().m_dack_gc)
return;
m_num_propagations_since_last_gc = 0;
if (m_table.size() > m_gc_threshold)
propagate();
while (m_table.size() > m_gc_threshold)
remove(m_queue->prev());
@ -147,7 +150,6 @@ namespace bv {
}
void ackerman::propagate() {
SASSERT(s.s().at_base_lvl());
auto* n = m_queue;
vv* k = nullptr;
unsigned num_prop = static_cast<unsigned>(s.s().get_stats().m_conflict * s.get_config().m_dack_factor);

12
src/sat/smt/bv_ackerman.h
View file

@ -51,13 +51,13 @@ namespace bv {
solver& s;
table_t m_table;
vv* m_queue { nullptr };
vv* m_tmp_vv { nullptr };
vv* m_queue = nullptr;
vv* m_tmp_vv = nullptr;
unsigned m_gc_threshold { 100 };
unsigned m_propagate_high_watermark { 10000 };
unsigned m_propagate_low_watermark { 10 };
unsigned m_num_propagations_since_last_gc { 0 };
unsigned m_gc_threshold = 100;
unsigned m_propagate_high_watermark = 10000;
unsigned m_propagate_low_watermark = 10;
unsigned m_num_propagations_since_last_gc = 0;
bool_vector m_diff_levels;
void update_glue(vv& v);