mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-04-08 18:31:49 +00:00
Code Activity

fixes to maxres/mss

Browse source 
Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com>
This commit is contained in:
Nikolaj Bjorner 2014-09-03 10:03:56 -07:00
parent b5bbf83847
commit 18b491eee0
10 changed files with 76 additions and 114 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

50
src/opt/inc_sat_solver.cpp
View file

@ -37,6 +37,7 @@ class inc_sat_solver : public solver {
sat::solver m_solver;
goal2sat m_goal2sat;
params_ref m_params;
bool m_optimize_model; // parameter
expr_ref_vector m_fmls;
expr_ref_vector m_current_fmls;
unsigned_vector m_fmls_lim;
@ -54,10 +55,12 @@ class inc_sat_solver : public solver {
expr_ref_vector m_soft;
vector<rational> m_weights;
typedef obj_map<expr, sat::literal> dep2asm_t;
public:
inc_sat_solver(ast_manager& m, params_ref const& p):
m(m), m_solver(p,0), m_params(p),
m(m), m_solver(p,0),
m_params(p), m_optimize_model(false),
m_fmls(m), m_current_fmls(m), m_core(m), m_map(m),
m_num_scopes(0),
m_dep_core(m),
@ -165,6 +168,7 @@ public:
m_params = p;
m_params.set_bool("elim_vars", false);
m_solver.updt_params(m_params);
m_optimize_model = m_params.get_bool("optimize_model", false);
}
virtual void collect_statistics(statistics & st) const {
m_preprocess->collect_statistics(st);
@ -226,32 +230,26 @@ private:
lbool r = internalize_formulas();
if (r != l_true) return r;
r = internalize_assumptions(soft.size(), soft.c_ptr(), dep2asm);
if (r != l_true) return r;
sat::literal_vector lits;
svector<double> weights;
sat::literal lit;
if (r == l_true) {
for (unsigned i = 0; i < soft.size(); ++i) {
weights.push_back(m_weights[i].get_double());
expr* s = soft[i].get();
bool is_neg = m.is_not(s, s);
if (!dep2asm.find(s, lit)) {
std::cout << "not found: " << mk_pp(s, m) << "\n";
dep2asm_t::iterator it = dep2asm.begin(), end = dep2asm.end();
for (; it != end; ++it) {
std::cout << mk_pp(it->m_key, m) << " " << it->m_value << "\n";
}
UNREACHABLE();
}
if (is_neg) {
lit.neg();
}
lits.push_back(lit);
for (unsigned i = 0; i < soft.size(); ++i) {
weights.push_back(m_weights[i].get_double());
expr* s = soft[i].get();
if (!dep2asm.find(s, lit)) {
IF_VERBOSE(0,
verbose_stream() << "not found: " << mk_pp(s, m) << "\n";
dep2asm_t::iterator it = dep2asm.begin();
dep2asm_t::iterator end = dep2asm.end();
for (; it != end; ++it) {
verbose_stream() << mk_pp(it->m_key, m) << " " << it->m_value << "\n";
}
UNREACHABLE(););
}
m_solver.initialize_soft(lits.size(), lits.c_ptr(), weights.c_ptr());
m_params.set_bool("optimize_model", true);
m_solver.updt_params(m_params);
lits.push_back(lit);
}
m_solver.initialize_soft(lits.size(), lits.c_ptr(), weights.c_ptr());
return r;
}
@ -324,12 +322,8 @@ private:
m_core.reset();
for (unsigned i = 0; i < core.size(); ++i) {
expr* e;
if (asm2dep.find(core[i].index(), e)) {
if (core[i].sign()) {
e = m.mk_not(e);
}
m_core.push_back(e);
}
VERIFY (asm2dep.find(core[i].index(), e));
m_core.push_back(e);
}
TRACE("opt",
dep2asm_t::iterator it = dep2asm.begin();

104
src/opt/maxres.cpp
View file

@ -72,14 +72,12 @@ public:
enum strategy_t {
s_mus,
s_mus_mss,
s_mus_mss2,
s_mss
};
private:
expr_ref_vector m_B;
expr_ref_vector m_asms;
obj_map<expr, rational> m_asm2weight;
obj_map<expr, bool> m_asm2value;
ptr_vector<expr> m_new_core;
mus m_mus;
mss m_mss;
@ -186,44 +184,6 @@ public:
return l_true;
}
lbool mus_mss_solver() {
init();
init_local();
sls();
exprs mcs;
vector<exprs> cores;
while (m_lower < m_upper) {
TRACE("opt",
display_vec(tout, m_asms.size(), m_asms.c_ptr());
s().display(tout);
tout << "\n";
display(tout);
);
lbool is_sat = try_improve_bound(cores, mcs);
if (m_cancel) {
return l_undef;
}
switch (is_sat) {
case l_undef:
return l_undef;
case l_false:
SASSERT(cores.empty() && mcs.empty());
m_lower = m_upper;
return l_true;
case l_true:
SASSERT(cores.empty() || mcs.empty());
for (unsigned i = 0; i < cores.size(); ++i) {
process_unsat(cores[i]);
}
if (cores.empty()) {
process_sat(mcs);
}
break;
}
}
m_lower = m_upper;
return l_true;
}
lbool mss_solver() {
init();
@ -234,6 +194,9 @@ public:
lbool is_sat = l_true;
while (m_lower < m_upper && is_sat == l_true) {
IF_VERBOSE(1, verbose_stream() << "(opt.maxres [" << m_lower << ":" << m_upper << "])\n";);
if (m_cancel) {
return l_undef;
}
vector<exprs> cores;
exprs mss;
model_ref mdl;
@ -241,6 +204,12 @@ public:
mcs.reset();
s().get_model(mdl);
update_assignment(mdl.get());
exprs cs;
get_current_correction_set(mdl.get(), cs);
process_sat(cs);
#if 0
is_sat = get_mss(mdl.get(), cores, mss, mcs);
switch (is_sat) {
@ -249,15 +218,12 @@ public:
case l_false:
m_lower = m_upper;
return l_true;
case l_true: {
case l_true:
process_sat(mcs);
get_mss_model();
break;
}
}
if (m_cancel) {
return l_undef;
break;
}
#endif
if (m_lower < m_upper) {
is_sat = s().check_sat(0, 0);
}
@ -294,7 +260,7 @@ public:
Suppose correction set is huge. Do we really need it?
*/
lbool mus_mss2_solver() {
lbool mus_mss_solver() {
init();
init_local();
sls();
@ -356,14 +322,8 @@ public:
// obtained from the current best model.
//
//
// TBD: throttle blocking on correction sets if they are too big.
// likewise, if the cores are too big, don't block the cores.
//
exprs cs;
get_current_correction_set(cs);
get_current_correction_set(mdl.get(), cs);
unsigned max_core = max_core_size(cores);
if (!cs.empty() && cs.size() < max_core) {
process_sat(cs);
@ -379,7 +339,6 @@ public:
void found_optimum() {
s().get_model(m_model);
m_asm2value.reset();
DEBUG_CODE(
for (unsigned i = 0; i < m_asms.size(); ++i) {
SASSERT(is_true(m_asms[i].get()));
@ -397,8 +356,6 @@ public:
return mus_solver();
case s_mus_mss:
return mus_mss_solver();
case s_mus_mss2:
return mus_mss2_solver();
case s_mss:
return mss_solver();
}
@ -462,14 +419,14 @@ public:
return is_sat;
}
void get_current_correction_set(exprs& cs) {
void get_current_correction_set(model* mdl, exprs& cs) {
cs.reset();
if (!mdl) return;
for (unsigned i = 0; i < m_asms.size(); ++i) {
if (!is_true(m_asms[i].get())) {
if (!is_true(mdl, m_asms[i].get())) {
cs.push_back(m_asms[i].get());
}
}
IF_VERBOSE(2, verbose_stream() << "(opt.maxres correction set size: " << cs.size() << ")\n";);
TRACE("opt", display_vec(tout << "new correction set: ", cs.size(), cs.c_ptr()););
}
@ -554,7 +511,7 @@ public:
IF_VERBOSE(1, verbose_stream() << "(opt.maxres [" << m_lower << ":" << m_upper << "])\n";);
}
void get_mus_model(model_ref& mdl) {
bool get_mus_model(model_ref& mdl) {
rational w(0);
if (m_c.sat_enabled()) {
// SAT solver core extracts some model
@ -567,6 +524,7 @@ public:
if (mdl.get() && w < m_upper) {
update_assignment(mdl.get());
}
return 0 != mdl.get();
}
void get_mss_model() {
@ -682,17 +640,13 @@ public:
s().assert_expr(fml);
fml = m.mk_implies(dd, b_i);
s().assert_expr(fml);
m_asm2value.insert(dd, is_true(d) && is_true(b_i));
d = dd;
}
else {
dd = m.mk_and(b_i, d);
m_asm2value.insert(dd, is_true(d) && is_true(b_i));
m_trail.push_back(dd);
d = dd;
d = m.mk_and(b_i, d);
m_trail.push_back(d);
}
asum = mk_fresh_bool("a");
m_asm2value.insert(asum, is_true(b_i1) || is_true(d));
cls = m.mk_or(b_i1, d);
fml = m.mk_implies(asum, cls);
new_assumption(asum, w);
@ -809,7 +763,6 @@ public:
return l_undef;
}
void update_assignment(model* mdl) {
rational upper(0);
expr_ref tmp(m);
@ -826,7 +779,6 @@ public:
return;
}
m_model = mdl;
m_asm2value.reset();
for (unsigned i = 0; i < m_soft.size(); ++i) {
m_assignment[i] = is_true(m_soft[i].get());
@ -851,16 +803,16 @@ public:
s().assert_expr(fml);
}
bool is_true(expr* e) {
bool truth_value;
if (m_asm2value.find(e, truth_value)) {
return truth_value;
}
bool is_true(model* mdl, expr* e) {
expr_ref tmp(m);
VERIFY(m_model->eval(e, tmp));
VERIFY(mdl->eval(e, tmp));
return m.is_true(tmp);
}
bool is_true(expr* e) {
return is_true(m_model.get(), e);
}
void remove_soft(exprs const& core, expr_ref_vector& asms) {
for (unsigned i = 0; i < asms.size(); ++i) {
if (core.contains(asms[i].get())) {
@ -933,7 +885,7 @@ opt::maxsmt_solver_base* opt::mk_maxres(
opt::maxsmt_solver_base* opt::mk_mus_mss_maxres(
context& c, weights_t& ws, expr_ref_vector const& soft) {
return alloc(maxres, c, ws, soft, maxres::s_mus_mss2);
return alloc(maxres, c, ws, soft, maxres::s_mus_mss);
}
opt::maxsmt_solver_base* opt::mk_mss_maxres(

1
src/opt/mus.cpp
View file

@ -70,6 +70,7 @@ struct mus::imp {
lbool get_mus(unsigned_vector& mus) {
// SASSERT: mus does not have duplicates.
m_model.reset();
unsigned_vector core;
for (unsigned i = 0; i < m_cls2expr.size(); ++i) {
core.push_back(i);

6
src/opt/opt_context.cpp
View file

@ -465,9 +465,13 @@ namespace opt {
void context::enable_sls(expr_ref_vector const& soft, vector<rational> const& weights) {
SASSERT(soft.size() == weights.size());
if (m_enable_sls && m_sat_solver.get()) {
if (m_sat_solver.get()) {
set_soft_inc_sat(m_sat_solver.get(), soft.size(), soft.c_ptr(), weights.c_ptr());
}
if (m_enable_sls && m_sat_solver.get()) {
m_params.set_bool("optimize_model", true);
m_sat_solver->updt_params(m_params);
}
}
struct context::is_bv {

8
src/sat/sat_mus.cpp
View file

@ -32,6 +32,7 @@ namespace sat {
m_core.reset();
m_mus.reset();
m_model.reset();
m_best_value = 0;
}
void mus::set_core() {
@ -96,8 +97,15 @@ namespace sat {
if (!core.empty()) {
// mr(); // TBD: measure
}
double new_value = s.m_wsls.evaluate_model(s.m_model);
if (m_model.empty()) {
m_model.append(s.m_model);
m_best_value = new_value;
}
else if (m_best_value > new_value) {
m_model.reset();
m_model.append(s.m_model);
m_best_value = new_value;
}
break;
}

1
src/sat/sat_mus.h
View file

@ -25,6 +25,7 @@ namespace sat {
literal_vector m_mus;
bool m_is_active;
model m_model; // model obtained during minimal unsat core
double m_best_value;
solver& s;

8
src/sat/sat_sls.cpp
View file

@ -354,7 +354,7 @@ namespace sat {
//
// Initialize m_clause_weights, m_hscore, m_sscore.
//
m_best_value = m_false.empty()?evaluate_model():-1.0;
m_best_value = m_false.empty()?evaluate_model(m_model):-1.0;
m_best_model.reset();
m_clause_weights.reset();
m_hscore.reset();
@ -382,7 +382,7 @@ namespace sat {
for (; !m_cancel && m_best_value > 0 && i < m_max_tries; ++i) {
wflip();
if (m_false.empty()) {
double val = evaluate_model();
double val = evaluate_model(m_model);
if (val < m_best_value || m_best_value < 0.0) {
m_best_value = val;
m_best_model.reset();
@ -511,12 +511,12 @@ namespace sat {
DEBUG_CODE(check_invariant(););
}
double wsls::evaluate_model() {
double wsls::evaluate_model(model& mdl) {
SASSERT(m_false.empty());
double result = 0.0;
for (unsigned i = 0; i < m_soft.size(); ++i) {
literal lit = m_soft[i];
if (value_at(lit, m_model) != l_true) {
if (value_at(lit, mdl) != l_true) {
result += m_weights[i];
}
}

2
src/sat/sat_sls.h
View file

@ -99,12 +99,12 @@ namespace sat {
void opt(unsigned sz, literal const* tabu, bool reuse_model);
model const& get_model() { return m_best_model; }
virtual void display(std::ostream& out) const;
double evaluate_model(model& mdl);
private:
void wflip();
void wflip(literal lit);
void update_hard_weights();
bool pick_wflip(literal & lit);
double evaluate_model();
virtual void check_invariant();
void refresh_scores(bool_var v);
int compute_hscore(bool_var v);

2
src/sat/sat_solver.cpp
View file

@ -1737,7 +1737,7 @@ namespace sat {
m_mus(); // ignore return value on cancelation.
m_model.reset();
m_model.append(m_mus.get_model());
m_model.append(m_mus.get_model());
}
}

8
src/sat/tactic/goal2sat.cpp
View file

@ -372,7 +372,9 @@ struct goal2sat::imp {
SASSERT(m_result_stack.empty());
}
void insert_dep(expr* dep, bool sign) {
void insert_dep(expr* dep0, expr* dep, bool sign) {
SASSERT(sign || dep0 == dep); // !sign || (not dep0) == dep.
SASSERT(!sign || m.is_not(dep0));
expr_ref new_dep(m), fml(m);
if (is_uninterp_const(dep)) {
new_dep = dep;
@ -386,7 +388,7 @@ struct goal2sat::imp {
}
convert_atom(new_dep, false, false);
sat::literal lit = m_result_stack.back();
m_dep2asm.insert(dep, sign?(~lit):lit);
m_dep2asm.insert(dep0, sign?~lit:lit);
m_result_stack.pop_back();
}
@ -411,7 +413,7 @@ struct goal2sat::imp {
SASSERT(m.is_bool(d));
bool sign = m.is_not(d, d1);
insert_dep(d1, sign);
insert_dep(d, d1, sign);
if (d == f) {
goto skip_dep;
}