mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-07-17 09:56:39 +00:00
Code Activity

Cleanup of hypothesis_reducer

Browse source
This commit is contained in:
Arie Gurfinkel 2018-05-16 16:25:49 -07:00
parent 2db38fedd6
commit 8d312f9d1f
2 changed files with 160 additions and 189 deletions
Download patch file Download diff file Expand all files Collapse all files

345
src/muz/spacer/spacer_proof_utils.cpp
View file

@ -133,7 +133,7 @@ proof_ref theory_axiom_reducer::reduce(proof* pr) {
SASSERT(m.get_fact(res) == m.get_fact(p)); SASSERT(m.get_fact(res) == m.get_fact(p));
} }
else { else {
// proof is dirty, if a subproof of one of its premises // proof is dirty, if a sub-proof of one of its premises
// has been transformed // has been transformed
bool dirty = false; bool dirty = false;
@ -173,85 +173,67 @@ proof_ref theory_axiom_reducer::reduce(proof* pr) {
return proof_ref(res, m); return proof_ref(res, m);
} }
void hypothesis_reducer::reset() void hypothesis_reducer::reset() {
{
m_cache.reset(); m_cache.reset();
m_units.reset(); m_units.reset();
m_active_hyps.reset(); m_active_hyps.reset();
m_parent_hyps.reset(); m_parent_hyps.reset();
for (auto t : m_pinned_active_hyps) dealloc(t);
m_pinned_active_hyps.reset(); m_pinned_active_hyps.reset();
for (auto t : m_pinned_parent_hyps) dealloc(t);
m_pinned_parent_hyps.reset(); m_pinned_parent_hyps.reset();
m_pinned.reset();
} }
void hypothesis_reducer::compute_hypsets(proof* pr) void hypothesis_reducer::compute_hypsets(proof *pr) {
{
ptr_vector<proof> todo; ptr_vector<proof> todo;
todo.push_back(pr); todo.push_back(pr);
while (!todo.empty()) while (!todo.empty()) {
{
proof* p = todo.back(); proof* p = todo.back();
if (m_active_hyps.contains(p)) if (m_active_hyps.contains(p)) {
{
SASSERT(m_parent_hyps.contains(p)); SASSERT(m_parent_hyps.contains(p));
todo.pop_back(); todo.pop_back();
continue;
} }
// if we haven't already visited the current unit
else
{
bool existsUnvisitedParent = false;
// add unprocessed premises to stack for DFS. If there is at least one unprocessed premise, don't compute the result bool dirty = false;
// for p now, but wait until those unprocessed premises are processed. for (unsigned i = 0, sz = m.get_num_parents(p); i < sz; ++i) {
for (unsigned i = 0; i < m.get_num_parents(p); ++i) { SASSERT(m.is_proof(p->get_arg(i)));
SASSERT(m.is_proof(p->get_arg(i))); proof *parent = to_app(p->get_arg(i));
proof* premise = to_app(p->get_arg(i));
// if we haven't visited the premise yet if (!m_active_hyps.contains(parent)) {
if (!m_active_hyps.contains(premise)) SASSERT(!m_parent_hyps.contains(parent));
{ todo.push_back(parent);
SASSERT(!m_parent_hyps.contains(premise)); dirty = true;
// add it to the stack
todo.push_back(premise);
existsUnvisitedParent = true;
}
} }
}
if (dirty) continue;
// if we already visited all premises, we can visit p too todo.pop_back();
if (!existsUnvisitedParent)
{
todo.pop_back();
// create active_hyps-set and parent_hyps-set for step p // create active_hyps-set and parent_hyps-set for step p
proof_set* active_hyps = alloc(proof_set); proof_set* active_hyps = alloc(proof_set);
m_pinned_active_hyps.insert(active_hyps); m_pinned_active_hyps.insert(active_hyps);
m_active_hyps.insert(p, active_hyps); m_active_hyps.insert(p, active_hyps);
expr_set* parent_hyps = alloc(expr_set); expr_set* parent_hyps = alloc(expr_set);
m_pinned_parent_hyps.insert(parent_hyps); m_pinned_parent_hyps.insert(parent_hyps);
m_parent_hyps.insert(p, parent_hyps); m_parent_hyps.insert(p, parent_hyps);
// fill both sets // fill both sets
if (m.is_hypothesis(p)) if (m.is_hypothesis(p)) {
{ active_hyps->insert(p);
active_hyps->insert(p); parent_hyps->insert(m.get_fact(p));
parent_hyps->insert(m.get_fact(p)); }
} else {
else for (unsigned i = 0, sz = m.get_num_parents(p); i < sz; ++i) {
{ proof* parent = m.get_parent(p, i);
for (unsigned i = 0, sz = m.get_num_parents(p); i < sz; ++i) set_union(*parent_hyps, *m_parent_hyps.find(parent));
{
proof* pp = m.get_parent(p, i);
set_union(*parent_hyps, *m_parent_hyps.find(pp));
if (!m.is_lemma(p)) // lemmas clear all hypotheses if (!m.is_lemma(p))
{ // lemmas clear all hypotheses
set_union(*active_hyps, *m_active_hyps.find(pp)); set_union(*active_hyps, *m_active_hyps.find(parent));
}
}
}
} }
} }
} }
@ -259,48 +241,41 @@ void hypothesis_reducer::compute_hypsets(proof* pr)
// collect all units that are hyp-free and are used as hypotheses somewhere // collect all units that are hyp-free and are used as hypotheses somewhere
// requires that m_active_hyps and m_parent_hyps have been computed // requires that m_active_hyps and m_parent_hyps have been computed
void hypothesis_reducer::collect_units(proof* pr) void hypothesis_reducer::collect_units(proof* pr) {
{
expr_set* all_hyps = m_parent_hyps.find(pr); expr_set* all_hyps = m_parent_hyps.find(pr);
SASSERT(all_hyps != nullptr); SASSERT(all_hyps);
proof_post_order pit(pr, m); proof_post_order pit(pr, m);
while (pit.hasNext()) { while (pit.hasNext()) {
proof* p = pit.next(); proof* p = pit.next();
if (!m.is_hypothesis(p)) if (!m.is_hypothesis(p)) {
{
proof_set* active_hyps = m_active_hyps.find(p); proof_set* active_hyps = m_active_hyps.find(p);
SASSERT(active_hyps != nullptr); SASSERT(active_hyps);
// collect units that are hyp-free and are used as hypotheses in the proof pr // collect units that are hyp-free and are used as
if (active_hyps->empty() && m.has_fact(p) && all_hyps->contains(m.get_fact(p))) // hypotheses in the proof pr
{ if (active_hyps->empty() && m.has_fact(p) &&
all_hyps->contains(m.get_fact(p)))
m_units.insert(m.get_fact(p), p); m_units.insert(m.get_fact(p), p);
}
} }
} }
} }
proof_ref hypothesis_reducer::reduce(proof* pr) proof_ref hypothesis_reducer::reduce(proof* pr) {
{
compute_hypsets(pr); compute_hypsets(pr);
collect_units(pr); collect_units(pr);
proof* res = compute_transformed_proof(pr); proof_ref res(compute_transformed_proof(pr), m);
SASSERT(res != nullptr); SASSERT(res);
proof_ref res_ref(res,m);
reset(); reset();
DEBUG_CODE(proof_checker pc(m); DEBUG_CODE(proof_checker pc(m);
expr_ref_vector side(m); expr_ref_vector side(m);
SASSERT(pc.check(res, side)); SASSERT(pc.check(res, side)););
); return res;
return res_ref;
} }
proof* hypothesis_reducer::compute_transformed_proof(proof* pf) proof* hypothesis_reducer::compute_transformed_proof(proof* pf) {
{
proof *res = NULL; proof *res = NULL;
ptr_vector<proof> todo; ptr_vector<proof> todo;
@ -325,204 +300,198 @@ proof* hypothesis_reducer::compute_transformed_proof(proof* pf)
pp = m.get_parent(p, i); pp = m.get_parent(p, i);
if (m_cache.find(pp, tmp)) { if (m_cache.find(pp, tmp)) {
args.push_back(tmp); args.push_back(tmp);
dirty = dirty || pp != tmp; dirty |= pp != tmp;
} else { } else {
todo.push_back(pp); todo.push_back(pp);
} }
} }
if (todo_sz < todo.size()) { continue; } if (todo_sz < todo.size()) continue;
else { todo.pop_back(); }
todo.pop_back();
// here the proof transformation begins // transform the proof
// INV: whenever we visit p, active_hyps and parent_hyps have been computed for the args.
if (m.is_hypothesis(p)) // INV: whenever p is visited, active_hyps and parent_hyps
{ // have already been computed for everything in args.
if (m.is_hypothesis(p)) {
// hyp: replace by a corresponding unit // hyp: replace by a corresponding unit
if (m_units.find(m.get_fact(p), tmp)) if (m_units.find(m.get_fact(p), tmp)) {
{
// look up the proof of the unit: // look up the proof of the unit:
// if there is a transformed proof use that one // if there is a transformed proof use that one
// otherwise use the original proof // otherwise use the original proof
proof* proof_of_unit; proof* proof_of_unit;
if (!m_cache.find(tmp,proof_of_unit)) if (!m_cache.find(tmp, proof_of_unit)) {
{
proof_of_unit = tmp; proof_of_unit = tmp;
} }
// compute hypsets (have not been computed in general, since the unit can be anywhere in the proof) // compute hypsets (have not been computed in general,
// since the unit can be anywhere in the proof)
compute_hypsets(proof_of_unit); compute_hypsets(proof_of_unit);
// if the transformation doesn't create a cycle, perform it // if the transformation doesn't create a cycle, perform it
SASSERT(m_parent_hyps.contains(proof_of_unit)); SASSERT(m_parent_hyps.contains(proof_of_unit));
expr_set* parent_hyps = m_parent_hyps.find(proof_of_unit); expr_set* parent_hyps = m_parent_hyps.find(proof_of_unit);
if (!parent_hyps->contains(p)) if (!parent_hyps->contains(p))
{
res = proof_of_unit;
// hypsets have already been computed for proof_of_unit // hypsets have already been computed for proof_of_unit
} res = proof_of_unit;
// otherwise don't transform the proof and just use the hypothesis // otherwise don't transform the proof and just use
// the hypothesis
else else
{
res = p;
// hypsets have already been computed for p // hypsets have already been computed for p
} res = p;
} }
else else
{
res = p;
// hypsets have already been computed for p // hypsets have already been computed for p
} res = p;
} }
else if (!dirty) { res = p; } else if (!dirty)
res = p;
else if (m.is_lemma(p)) else if (m.is_lemma(p)) {
{ //lemma: reduce the premise; remove reduced consequences
//lemma: reduce the premise; remove reduced consequences from conclusion //from conclusion
SASSERT(args.size() == 1); SASSERT(args.size() == 1);
res = mk_lemma_core(args[0], m.get_fact(p)); res = mk_lemma_core(args[0], m.get_fact(p));
compute_hypsets(res); compute_hypsets(res);
} }
else if (m.is_unit_resolution(p)) else if (m.is_unit_resolution(p)) {
{ // unit: reduce untis; reduce the first premise; rebuild
// unit: reduce untis; reduce the first premise; rebuild unit resolution // unit resolution
res = mk_unit_resolution_core(args); res = mk_unit_resolution_core(args);
compute_hypsets(res); compute_hypsets(res);
} }
else else {
{ res = mk_proof_core(p, args);
res = mk_step_core(p, args);
compute_hypsets(res); compute_hypsets(res);
} }
SASSERT(res); SASSERT(res);
m_cache.insert(p, res); m_cache.insert(p, res);
// bail out as soon as found a sub-proof of false
SASSERT(m_active_hyps.contains(res)); SASSERT(m_active_hyps.contains(res));
proof_set* active_hyps = m_active_hyps.find(res); proof_set* active_hyps = m_active_hyps.find(res);
if (active_hyps->empty() && m.has_fact(res) && m.is_false(m.get_fact(res))) if (active_hyps->empty() && m.has_fact(res) && m.is_false(m.get_fact(res)))
{
return res; return res;
}
} }
UNREACHABLE(); UNREACHABLE();
return nullptr; return nullptr;
} }
proof* hypothesis_reducer::mk_lemma_core(proof* premise, expr *fact) proof* hypothesis_reducer::mk_lemma_core(proof* premise, expr *fact) {
{
SASSERT(m.is_false(m.get_fact(premise))); SASSERT(m.is_false(m.get_fact(premise)));
SASSERT(m_active_hyps.contains(premise)); SASSERT(m_active_hyps.contains(premise));
proof_set* active_hyps = m_active_hyps.find(premise); proof_set* active_hyps = m_active_hyps.find(premise);
// if there is no active hypothesis return the premise // if there is no active hypothesis return the premise
if (active_hyps->empty()) if (active_hyps->empty()) {
{ // XXX just in case premise might go away
m_pinned.push_back(premise);
return premise; return premise;
} }
// otherwise build disjunction of the negated active hypothesis' and add lemma step.
else
{
expr_ref_buffer args(m);
for (auto hyp : *active_hyps)
{
expr* hyp_fact = m.get_fact(hyp);
expr_ref negated_hyp_fact(m);
negated_hyp_fact = m.is_not(hyp_fact) ? to_app(hyp_fact)->get_arg(0) : m.mk_not(hyp_fact);
args.push_back(negated_hyp_fact);
}
expr_ref lemma(m); // otherwise, build a disjunction of the negated active hypotheses
if (args.size() == 1) // and add a lemma proof step
{ expr_ref_buffer args(m);
lemma = args[0]; for (auto hyp : *active_hyps) {
} expr *hyp_fact, *t;
hyp_fact = m.get_fact(hyp);
if (m.is_not(hyp_fact, t))
args.push_back(t);
else else
{ args.push_back(m.mk_not(hyp_fact));
lemma = m.mk_or(args.size(), args.c_ptr());
}
proof_ref res(m);
res = m.mk_lemma(premise, lemma);
m_pinned.push_back(res);
return res;
} }
expr_ref lemma(m);
lemma = mk_or(m, args.size(), args.c_ptr());
proof* res;
res = m.mk_lemma(premise, lemma);
m_pinned.push_back(res);
return res;
} }
proof* hypothesis_reducer::mk_unit_resolution_core(ptr_buffer<proof>& args) proof* hypothesis_reducer::mk_unit_resolution_core(ptr_buffer<proof>& args) {
{
ptr_buffer<proof> pf_args; // the arguments of the transformed unit resolution step
pf_args.push_back(args [0]); // the first element of args is the clause to resolve with
// if any literal is false, we don't need a unit resolution step // if any literal is false, we don't need a unit resolution step
// could be the case due to transformations which already have been done // This can be the case due to some previous transformations
for (unsigned i = 1; i < args.size(); ++i) for (unsigned i = 1, sz = args.size(); i < sz; ++i) {
{ if (m.is_false(m.get_fact(args[i]))) {
if (m.is_false(m.get_fact(args[i]))) // XXX just in case
{ m_pinned.push_back(args[i]);
return args[i]; return args[i];
} }
} }
app *cls_fact = to_app(m.get_fact(args[0])); // BUG: I guess this shouldn't work with quantifiers (since they are no apps) proof* arg0 = args[0];
ptr_buffer<expr> cls; ptr_buffer<proof> pf_args;
if (m.is_or(cls_fact)) { pf_args.push_back(arg0);
for (unsigned i = 0, sz = cls_fact->get_num_args(); i < sz; ++i)
{ cls.push_back(cls_fact->get_arg(i)); }
} else { cls.push_back(cls_fact); }
// construct new resolvent // BUG: I guess this shouldn't work with quantifiers (since they
ptr_buffer<expr> new_fact_cls; // are not apps)
// AG: who is "I"? What is the bug?
app *fact = to_app(m.get_fact(arg0));
ptr_buffer<expr> cls;
if (m.is_or(fact)) {
for (unsigned i = 0, sz = fact->get_num_args(); i < sz; ++i)
cls.push_back(fact->get_arg(i));
}
else
cls.push_back(fact);
// construct the new resolvent
ptr_buffer<expr> new_fact;
bool found; bool found;
// XXX quadratic
// -- find all literals that are resolved on
// XXX quadratic implementation
for (unsigned i = 0, sz = cls.size(); i < sz; ++i) { for (unsigned i = 0, sz = cls.size(); i < sz; ++i) {
found = false; found = false;
for (unsigned j = 1; j < args.size(); ++j) { for (unsigned j = 1; j < args.size(); ++j) {
if (m.is_complement(cls.get(i), m.get_fact(args [j]))) { if (m.is_complement(cls.get(i), m.get_fact(args[j]))) {
found = true; found = true;
pf_args.push_back(args [j]); pf_args.push_back(args[j]);
break; break;
} }
} }
if (!found) { if (!found) new_fact.push_back(cls.get(i));
new_fact_cls.push_back(cls.get(i));
}
} }
SASSERT(new_fact_cls.size() + pf_args.size() - 1 == cls.size()); SASSERT(new_fact.size() + pf_args.size() - 1 == cls.size());
expr_ref new_fact(m);
new_fact = mk_or(m, new_fact_cls.size(), new_fact_cls.c_ptr());
// create new proof step // unit resolution got reduced to noop
if (pf_args.size() == 1) // the only premise is the clause itself if (pf_args.size() == 1) {
{ // XXX just in case
return args[0]; m_pinned.push_back(arg0);
} return arg0;
else
{
proof* res = m.mk_unit_resolution(pf_args.size(), pf_args.c_ptr(), new_fact);
m_pinned.push_back(res);
return res;
} }
// make unit resolution proof step
expr_ref tmp(m);
tmp = mk_or(m, new_fact.size(), new_fact.c_ptr());
proof* res = m.mk_unit_resolution(pf_args.size(), pf_args.c_ptr(), tmp);
m_pinned.push_back(res);
return res;
} }
proof* hypothesis_reducer::mk_step_core(proof* old_step, ptr_buffer<proof>& args) proof* hypothesis_reducer::mk_proof_core(proof* old, ptr_buffer<proof>& args) {
{ // if any of the literals are false, we don't need a step
// if any of the literals is false, we don't need a step for (unsigned i = 0; i < args.size(); ++i) {
for (unsigned i = 0; i < args.size(); ++i) if (m.is_false(m.get_fact(args[i]))) {
{ // XXX just in case
if (m.is_false(m.get_fact(args[i]))) m_pinned.push_back(args[i]);
{
return args[i]; return args[i];
} }
} }
// otherwise build step // otherwise build step
args.push_back(to_app(m.get_fact(old_step))); // BUG: I guess this doesn't work with quantifiers (since they are no apps) // BUG: I guess this doesn't work with quantifiers (since they are no apps)
args.push_back(to_app(m.get_fact(old)));
SASSERT(old_step->get_decl()->get_arity() == args.size()); SASSERT(old->get_decl()->get_arity() == args.size());
proof* res = m.mk_app(old_step->get_decl(), args.size(), (expr * const*)args.c_ptr());
proof* res = m.mk_app(old->get_decl(), args.size(),
(expr * const*)args.c_ptr());
m_pinned.push_back(res); m_pinned.push_back(res);
return res; return res;
} }

4
src/muz/spacer/spacer_proof_utils.h
View file

@ -46,10 +46,12 @@ private:
void reset(); void reset();
}; };
/// reduces the number of hypotheses in a proof
class hypothesis_reducer class hypothesis_reducer
{ {
public: public:
hypothesis_reducer(ast_manager &m) : m(m), m_pinned(m) {} hypothesis_reducer(ast_manager &m) : m(m), m_pinned(m) {}
~hypothesis_reducer() {reset();}
// reduce hypothesis and return transformed proof // reduce hypothesis and return transformed proof
proof_ref reduce(proof* pf); proof_ref reduce(proof* pf);
@ -90,7 +92,7 @@ private:
proof* mk_lemma_core(proof *pf, expr *fact); proof* mk_lemma_core(proof *pf, expr *fact);
proof* mk_unit_resolution_core(ptr_buffer<proof>& args); proof* mk_unit_resolution_core(ptr_buffer<proof>& args);
proof* mk_step_core(proof* old_step, ptr_buffer<proof>& args); proof* mk_proof_core(proof* old, ptr_buffer<proof>& args);
}; };
} }
#endif #endif