3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2025-04-07 09:55:19 +00:00
z3/src/tactic/sls/sls_engine.cpp
Christoph M. Wintersteiger e9482a1447 disabled old code
2015-01-12 17:14:13 +00:00

1196 lines
40 KiB
C++

/*++
Copyright (c) 2012 Microsoft Corporation
Module Name:
sls_engine.cpp
Abstract:
A Stochastic Local Search (SLS) engine
Author:
Christoph (cwinter) 2014-03-19
Notes:
--*/
#include<iomanip>
#include"map.h"
#include"ast_smt2_pp.h"
#include"ast_pp.h"
#include"var_subst.h"
#include"model_pp.h"
#include"tactic.h"
#include"cooperate.h"
#include"luby.h"
#include"sls_compilation_settings.h"
#include"sls_params.hpp"
#include"sls_engine.h"
sls_engine::sls_engine(ast_manager & m, params_ref const & p) :
m_manager(m),
m_powers(m_mpz_manager),
m_zero(m_mpz_manager.mk_z(0)),
m_one(m_mpz_manager.mk_z(1)),
m_two(m_mpz_manager.mk_z(2)),
m_cancel(false),
m_bv_util(m),
m_tracker(m, m_bv_util, m_mpz_manager, m_powers),
m_evaluator(m, m_bv_util, m_tracker, m_mpz_manager, m_powers)
{
updt_params(p);
}
sls_engine::~sls_engine() {
m_mpz_manager.del(m_zero);
m_mpz_manager.del(m_one);
m_mpz_manager.del(m_two);
}
double sls_engine::get_restart_armin(unsigned cnt_restarts)
{
unsigned outer_id = (unsigned)(0.5 + sqrt(0.25 + 2 * cnt_restarts));
unsigned inner_id = cnt_restarts - (outer_id - 1) * outer_id / 2;
//printf("armin: %f\n", pow(1.1, inner_id + 1));
return pow((double) _RESTART_CONST_ARMIN_, (int) inner_id + 1);
}
void sls_engine::updt_params(params_ref const & _p) {
sls_params p(_p);
m_produce_models = _p.get_bool("model", false);
m_max_restarts = p.restarts();
m_tracker.set_random_seed(p.random_seed());
m_plateau_limit = p.plateau_limit();
}
void sls_engine::checkpoint() {
if (m_cancel)
throw tactic_exception(TACTIC_CANCELED_MSG);
cooperate("sls");
}
bool sls_engine::full_eval(model & mdl) {
bool res = true;
unsigned sz = m_assertions.size();
for (unsigned i = 0; i < sz && res; i++) {
checkpoint();
expr_ref o(m_manager);
if (!mdl.eval(m_assertions[i], o, true))
exit(ERR_INTERNAL_FATAL);
res = m_manager.is_true(o.get());
}
TRACE("sls", tout << "Evaluation: " << res << std::endl;);
return res;
}
double sls_engine::top_score() {
#if 0
double min = m_tracker.get_score(g->form(0));
unsigned sz = g->size();
for (unsigned i = 1; i < sz; i++) {
double q = m_tracker.get_score(g->form(i));
if (q < min) min = q;
}
TRACE("sls_top", tout << "Score distribution:";
for (unsigned i = 0; i < sz; i++)
tout << " " << m_tracker.get_score(g->form(i));
tout << " MIN: " << min << std::endl;);
return min;
#else
double top_sum = 0.0;
unsigned sz = m_assertions.size();
for (unsigned i = 0; i < sz; i++) {
expr * e = m_assertions[i];
top_sum += m_tracker.get_score(e);
}
TRACE("sls_top", tout << "Score distribution:";
for (unsigned i = 0; i < sz; i++)
tout << " " << m_tracker.get_score(m_assertions[i]);
tout << " AVG: " << top_sum / (double)sz << std::endl;);
#if _CACHE_TOP_SCORE_
m_tracker.set_top_sum(top_sum);
#endif
return top_sum / (double)sz;
#endif
}
double sls_engine::rescore() {
m_evaluator.update_all();
m_stats.m_full_evals++;
return top_score();
}
double sls_engine::serious_score(func_decl * fd, const mpz & new_value) {
m_evaluator.serious_update(fd, new_value);
m_stats.m_incr_evals++;
#if _CACHE_TOP_SCORE_
return (m_tracker.get_top_sum() / m_assertions.size());
#else
return top_score();
#endif
}
double sls_engine::incremental_score(func_decl * fd, const mpz & new_value) {
m_evaluator.update(fd, new_value);
m_stats.m_incr_evals++;
#if _CACHE_TOP_SCORE_
return (m_tracker.get_top_sum() / m_assertions.size());
#else
return top_score();
#endif
}
double sls_engine::incremental_score_prune(func_decl * fd, const mpz & new_value) {
#if _EARLY_PRUNE_
m_stats.m_incr_evals++;
if (m_evaluator.update_prune(fd, new_value))
#if _CACHE_TOP_SCORE_
return (m_tracker.get_top_sum() / m_assertions.size());
#else
return top_score();
#endif
else
return 0.0;
#else
NOT_IMPLEMENTED_YET();
#endif
}
// checks whether the score outcome of a given move is better than the previous score
bool sls_engine::what_if(
func_decl * fd,
const unsigned & fd_inx,
const mpz & temp,
double & best_score,
unsigned & best_const,
mpz & best_value) {
#ifdef Z3DEBUG
mpz old_value;
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
#endif
#if _EARLY_PRUNE_
double r = incremental_score_prune(fd, temp);
#else
double r = incremental_score(fd, temp);
#endif
#ifdef Z3DEBUG
TRACE("sls_whatif", tout << "WHAT IF " << fd->get_name() << " WERE " << m_mpz_manager.to_string(temp) <<
" --> " << r << std::endl;);
m_mpz_manager.del(old_value);
#endif
// if (r >= best_score) {
if (r > best_score) {
best_score = r;
best_const = fd_inx;
m_mpz_manager.set(best_value, temp);
return true;
}
return false;
}
// same as what_if, but only applied to the score of a specific atom, not the total score
bool sls_engine::what_if_local(
expr * e,
func_decl * fd,
const unsigned & fd_inx,
const mpz & temp,
double & best_score,
unsigned & best_const,
mpz & best_value)
{
m_evaluator.update(fd, temp);
double r = m_tracker.get_score(e);
if (r >= best_score) {
best_score = r;
best_const = fd_inx;
m_mpz_manager.set(best_value, temp);
return true;
}
return false;
}
void sls_engine::mk_add(unsigned bv_sz, const mpz & old_value, mpz & add_value, mpz & result) {
mpz temp, mask, mask2;
m_mpz_manager.add(old_value, add_value, temp);
m_mpz_manager.set(mask, m_powers(bv_sz));
m_mpz_manager.bitwise_not(bv_sz, mask, mask2);
m_mpz_manager.bitwise_and(temp, mask2, result);
m_mpz_manager.del(temp);
m_mpz_manager.del(mask);
m_mpz_manager.del(mask2);
}
// Andreas: do we really need all those temporary mpzs?
void sls_engine::mk_mul2(unsigned bv_sz, const mpz & old_value, mpz & result) {
mpz temp, mask, mask2;
m_mpz_manager.mul(old_value, m_two, temp);
m_mpz_manager.set(mask, m_powers(bv_sz));
m_mpz_manager.bitwise_not(bv_sz, mask, mask2);
m_mpz_manager.bitwise_and(temp, mask2, result);
m_mpz_manager.del(temp);
m_mpz_manager.del(mask);
m_mpz_manager.del(mask2);
}
void sls_engine::mk_div2(unsigned bv_sz, const mpz & old_value, mpz & result) {
m_mpz_manager.div(old_value, m_two, result);
}
void sls_engine::mk_inc(unsigned bv_sz, const mpz & old_value, mpz & incremented) {
unsigned shift;
m_mpz_manager.add(old_value, m_one, incremented);
if (m_mpz_manager.is_power_of_two(incremented, shift) && shift == bv_sz)
m_mpz_manager.set(incremented, m_zero);
}
void sls_engine::mk_dec(unsigned bv_sz, const mpz & old_value, mpz & decremented) {
if (m_mpz_manager.is_zero(old_value)) {
m_mpz_manager.set(decremented, m_powers(bv_sz));
m_mpz_manager.dec(decremented);
}
else
m_mpz_manager.sub(old_value, m_one, decremented);
}
void sls_engine::mk_inv(unsigned bv_sz, const mpz & old_value, mpz & inverted) {
m_mpz_manager.bitwise_not(bv_sz, old_value, inverted);
}
void sls_engine::mk_flip(sort * s, const mpz & old_value, unsigned bit, mpz & flipped) {
m_mpz_manager.set(flipped, m_zero);
if (m_bv_util.is_bv_sort(s)) {
mpz mask;
m_mpz_manager.set(mask, m_powers(bit));
m_mpz_manager.bitwise_xor(old_value, mask, flipped);
m_mpz_manager.del(mask);
}
else if (m_manager.is_bool(s))
m_mpz_manager.set(flipped, (m_mpz_manager.is_zero(old_value)) ? m_one : m_zero);
else
NOT_IMPLEMENTED_YET();
}
void sls_engine::mk_random_move(ptr_vector<func_decl> & unsat_constants)
{
unsigned rnd_mv = 0;
unsigned ucc = unsat_constants.size();
unsigned rc = (m_tracker.get_random_uint((ucc < 16) ? 4 : (ucc < 256) ? 8 : (ucc < 4096) ? 12 : (ucc < 65536) ? 16 : 32)) % ucc;
func_decl * fd = unsat_constants[rc];
mpz new_value;
sort * srt = fd->get_range();
if (m_manager.is_bool(srt))
m_mpz_manager.set(new_value, (m_mpz_manager.is_zero(m_tracker.get_value(fd))) ? m_one : m_zero);
else
{
#if _USE_ADDSUB_
if (m_mpz_manager.is_one(m_tracker.get_random_bool())) rnd_mv = 2;
if (m_mpz_manager.is_one(m_tracker.get_random_bool())) rnd_mv++;
move_type mt = (move_type)rnd_mv;
// inversion doesn't make sense, let's do a flip instead.
if (mt == MV_INV) mt = MV_FLIP;
#else
mt = MV_FLIP;
#endif
unsigned bit = 0;
switch (mt)
{
case MV_FLIP: {
unsigned bv_sz = m_bv_util.get_bv_size(srt);
bit = (m_tracker.get_random_uint((bv_sz < 16) ? 4 : (bv_sz < 256) ? 8 : (bv_sz < 4096) ? 12 : (bv_sz < 65536) ? 16 : 32)) % bv_sz;
mk_flip(fd->get_range(), m_tracker.get_value(fd), bit, new_value);
break;
}
case MV_INC:
mk_inc(m_bv_util.get_bv_size(fd->get_range()), m_tracker.get_value(fd), new_value);
break;
case MV_DEC:
mk_dec(m_bv_util.get_bv_size(fd->get_range()), m_tracker.get_value(fd), new_value);
break;
case MV_INV:
mk_inv(m_bv_util.get_bv_size(fd->get_range()), m_tracker.get_value(fd), new_value);
break;
default:
NOT_IMPLEMENTED_YET();
}
TRACE("sls", tout << "Randomization candidates: ";
for (unsigned i = 0; i < unsat_constants.size(); i++)
tout << unsat_constants[i]->get_name() << ", ";
tout << std::endl;
tout << "Random move: ";
switch (mt) {
case MV_FLIP: tout << "Flip #" << bit << " in " << fd->get_name() << std::endl; break;
case MV_INC: tout << "+1 for " << fd->get_name() << std::endl; break;
case MV_DEC: tout << "-1 for " << fd->get_name() << std::endl; break;
case MV_INV: tout << "NEG for " << fd->get_name() << std::endl; break;
}
tout << "Locally randomized model: " << std::endl; m_tracker.show_model(tout););
}
m_evaluator.update(fd, new_value);
m_mpz_manager.del(new_value);
}
void sls_engine::mk_random_move() {
mk_random_move(m_tracker.get_unsat_constants(m_assertions, m_stats.m_moves));
}
// will use VNS to ignore some possible moves and increase the flips per second
double sls_engine::find_best_move_vns(
ptr_vector<func_decl> & to_evaluate,
double score,
unsigned & best_const,
mpz & best_value,
unsigned & new_bit,
move_type & move)
{
mpz old_value, temp;
unsigned bv_sz, max_bv_sz = 0;
double new_score = score;
for (unsigned i = 0; i < to_evaluate.size() && new_score < 1.0; i++) {
func_decl * fd = to_evaluate[i];
sort * srt = fd->get_range();
bv_sz = (m_manager.is_bool(srt)) ? 1 : m_bv_util.get_bv_size(srt);
if (max_bv_sz < bv_sz) max_bv_sz = bv_sz;
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
if (m_bv_util.is_bv_sort(srt) && bv_sz > 1) {
if (!m_mpz_manager.is_even(old_value)) {
// for odd values, try +1
mk_inc(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_INC;
}
else {
// for even values, try -1
mk_dec(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_DEC;
}
// try inverting
mk_inv(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_INV;
// try to flip lsb
mk_flip(srt, old_value, 0, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value)) {
new_bit = 0;
move = MV_FLIP;
}
}
// reset to what it was before
double check = incremental_score(fd, old_value);
SASSERT(check == score);
}
// we can either check the condition once in the beginning or check it repeatedly after every bit
#if _VNS_ == 1
for (unsigned j = 1; j < max_bv_sz && new_score <= score; j++)
#else
if (new_score <= score)
for (unsigned j = 1; j < max_bv_sz && new_score < 1.0; j++)
#endif
for (unsigned i = 0; i < to_evaluate.size() && new_score < 1.0; i++) {
func_decl * fd = to_evaluate[i];
sort * srt = fd->get_range();
bv_sz = (m_manager.is_bool(srt)) ? 1 : m_bv_util.get_bv_size(srt);
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
// What would happen if we flipped bit #j ?
if (j < bv_sz)
{
mk_flip(srt, old_value, j, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value)) {
new_bit = j;
move = MV_FLIP;
}
}
// reset to what it was before
double check = incremental_score(fd, old_value);
SASSERT(check == score);
}
m_mpz_manager.del(old_value);
m_mpz_manager.del(temp);
return new_score;
}
// finds the move that increased score the most. returns best_const = -1, if no increasing move exists.
double sls_engine::find_best_move(
ptr_vector<func_decl> & to_evaluate,
double score,
unsigned & best_const,
mpz & best_value,
unsigned & new_bit,
move_type & move)
{
mpz old_value, temp;
#if _USE_MUL3_ || _USE_UNARY_MINUS_
mpz temp2;
#endif
unsigned bv_sz;
double new_score = score;
for (unsigned i = 0; i < to_evaluate.size() && new_score < 1.0; i++) {
func_decl * fd = to_evaluate[i];
sort * srt = fd->get_range();
bv_sz = (m_manager.is_bool(srt)) ? 1 : m_bv_util.get_bv_size(srt);
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
// first try to flip every bit
#if _SKIP_BITS_
for (unsigned j = (i + m_stats.m_moves) % (_SKIP_BITS_ + 1); j < bv_sz && new_score < 1.0; j += (_SKIP_BITS_ + 1)) {
#else
for (unsigned j = 0; j < bv_sz && new_score < 1.0; j++) {
#endif
// What would happen if we flipped bit #i ?
mk_flip(srt, old_value, j, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value)) {
new_bit = j;
move = MV_FLIP;
}
}
if (m_bv_util.is_bv_sort(srt) && bv_sz > 1) {
#if _USE_ADDSUB_
if (!m_mpz_manager.is_even(old_value)) {
// for odd values, try +1
mk_inc(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_INC;
}
else {
// for even values, try -1
mk_dec(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_DEC;
}
#endif
// try inverting
mk_inv(bv_sz, old_value, temp);
if (what_if(fd, i, temp, new_score, best_const, best_value))
move = MV_INV;
#if _USE_UNARY_MINUS_
mk_inc(bv_sz, temp, temp2);
if (what_if(g, fd, i, temp2, new_score, best_const, best_value))
move = MV_UMIN;
#endif
#if _USE_MUL2DIV2_
// try multiplication by 2
mk_mul2(bv_sz, old_value, temp);
if (what_if(g, fd, i, temp, new_score, best_const, best_value))
move = MV_MUL2;
#if _USE_MUL3_
// try multiplication by 3
mk_add(bv_sz, old_value, temp, temp2);
if (what_if(g, fd, i, temp2, new_score, best_const, best_value))
move = MV_MUL3;
#endif
// try division by 2
mk_div2(bv_sz, old_value, temp);
if (what_if(g, fd, i, temp, new_score, best_const, best_value))
move = MV_DIV2;
#endif
}
// reset to what it was before
double check = incremental_score(fd, old_value);
// Andreas: does not hold anymore now that we use top level score caching
//SASSERT(check == score);
}
m_mpz_manager.del(old_value);
m_mpz_manager.del(temp);
#if _USE_MUL3_
m_mpz_manager.del(temp2);
#endif
return new_score;
}
// same as find_best_move but only considers the score of the current expression instead of the overall score
double sls_engine::find_best_move_local(expr * e, ptr_vector<func_decl> & to_evaluate,
unsigned & best_const, mpz & best_value, unsigned & new_bit, move_type & move) {
mpz old_value, temp;
unsigned bv_sz;
double new_score = m_tracker.get_score(e);
// Andreas: tie breaking not implemented yet
// double tie_score = top_score(g);
for (unsigned i = 0; i < to_evaluate.size(); i++) {
func_decl * fd = to_evaluate[i];
sort * srt = fd->get_range();
bv_sz = (m_manager.is_bool(srt)) ? 1 : m_bv_util.get_bv_size(srt);
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
// first try to flip every bit
for (unsigned j = 0; j < bv_sz; j++) {
// What would happen if we flipped bit #i ?
mk_flip(srt, old_value, j, temp);
if (what_if_local(e, fd, i, temp, new_score, best_const, best_value)) {
new_bit = j;
move = MV_FLIP;
}
}
if (m_bv_util.is_bv_sort(srt) && bv_sz > 1) {
if (!m_mpz_manager.is_even(old_value)) {
// for odd values, try +1
mk_inc(bv_sz, old_value, temp);
if (what_if_local(e, fd, i, temp, new_score, best_const, best_value))
move = MV_INC;
}
else {
// for even values, try -1
mk_dec(bv_sz, old_value, temp);
if (what_if_local(e, fd, i, temp, new_score, best_const, best_value))
move = MV_DEC;
}
// try inverting
mk_inv(bv_sz, old_value, temp);
if (what_if_local(e, fd, i, temp, new_score, best_const, best_value))
move = MV_INV;
}
// reset to what it was before
m_evaluator.update(fd, old_value);
}
m_mpz_manager.del(old_value);
m_mpz_manager.del(temp);
return new_score;
}
// first try of intensification ... does not seem to be efficient
bool sls_engine::handle_plateau()
{
unsigned sz = m_assertions.size();
#if _BFS_
unsigned pos = m_stats.m_moves % sz;
#else
unsigned pos = m_tracker.get_random_uint(16) % sz;
#endif
expr * e = m_tracker.get_unsat_assertion(sz, pos);
if (!e)
return 0;
expr * q = m_tracker.get_unsat_expression(e);
ptr_vector<func_decl> & to_evaluate = m_tracker.get_constants(q);
for (unsigned i = 0; i < to_evaluate.size(); i++)
{
m_tracker.get_value(to_evaluate[i]);
m_old_values.push_back(&m_tracker.get_value(to_evaluate[i]));
}
unsigned new_const = (unsigned)-1, new_bit = 0;
mpz new_value;
move_type move;
for (unsigned i = 0; i < _INTENSIFICATION_TRIES_; i++)
{
// Andreas: Could be extended to use (best) score but this is computationally more expensive.
find_best_move_local(q, to_evaluate, new_const, new_value, new_bit, move);
if (new_const == static_cast<unsigned>(-1)) {
// Andreas: Actually this should never happen.
NOT_IMPLEMENTED_YET();
}
else {
m_stats.m_moves++;
func_decl * fd = to_evaluate[new_const];
switch (move) {
case MV_FLIP: m_stats.m_flips++; break;
case MV_INC: m_stats.m_incs++; break;
case MV_DEC: m_stats.m_decs++; break;
case MV_INV: m_stats.m_invs++; break;
case MV_UMIN: m_stats.m_umins++; break;
case MV_MUL2: m_stats.m_mul2s++; break;
case MV_MUL3: m_stats.m_mul3s++; break;
case MV_DIV2: m_stats.m_div2s++; break;
}
m_evaluator.update(fd, new_value);
}
if (m_mpz_manager.is_one(m_tracker.get_value(q)))
return 1;
}
for (unsigned i = 0; i < to_evaluate.size(); i++)
m_tracker.set_value(to_evaluate[i], *m_old_values[i]);
m_old_values.reset();
return 0;
}
// what_if version needed in the context of 2nd intensification try, combining local and global score
bool sls_engine::what_if(
expr * e,
func_decl * fd,
const mpz & temp,
double & best_score,
mpz & best_value,
unsigned i)
{
double global_score = incremental_score(fd, temp);
double local_score = m_tracker.get_score(e);
double new_score = i * local_score / _INTENSIFICATION_TRIES_ + (_INTENSIFICATION_TRIES_ - i) * global_score / _INTENSIFICATION_TRIES_;
if (new_score >= best_score) {
best_score = new_score;
m_mpz_manager.set(best_value, temp);
return true;
}
return false;
}
// find_best_move version needed in the context of 2nd intensification try
double sls_engine::find_best_move_local(expr * e, func_decl * fd, mpz & best_value, unsigned i)
{
mpz old_value, temp;
double best_score = 0;
sort * srt = fd->get_range();
unsigned bv_sz = (m_manager.is_bool(srt)) ? 1 : m_bv_util.get_bv_size(srt);
m_mpz_manager.set(old_value, m_tracker.get_value(fd));
for (unsigned j = 0; j < bv_sz && best_score < 1.0; j++) {
mk_flip(srt, old_value, j, temp);
what_if(e, fd, temp, best_score, best_value, i);
}
m_mpz_manager.del(old_value);
m_mpz_manager.del(temp);
return best_score;
}
// second try to use intensification ... also not very effective
bool sls_engine::handle_plateau(double old_score)
{
unsigned sz = m_assertions.size();
#if _BFS_
unsigned new_const = m_stats.m_moves % sz;
#else
unsigned new_const = m_tracker.get_random_uint(16) % sz;
#endif
expr * e = m_tracker.get_unsat_assertion(m_assertions, sz, new_const);
if (!e)
return 0;
expr * q = m_tracker.get_unsat_expression(e);
ptr_vector<func_decl> & to_evaluate = m_tracker.get_constants(q);
new_const = m_tracker.get_random_uint(16) % to_evaluate.size();
func_decl * fd = to_evaluate[new_const];
mpz new_value;
//m_mpz_manager.set(new_value, m_tracker.get_value(fd));
unsigned new_bit = 0;
double global_score = old_score, local_score = m_tracker.get_score(q), new_score = old_score;
for (unsigned i = 1; i <= _INTENSIFICATION_TRIES_; i++)
{
new_score = find_best_move_local(q, fd, new_value, i);
m_stats.m_moves++;
m_stats.m_flips++;
global_score = incremental_score(fd, new_value);
local_score = m_tracker.get_score(q);
SASSERT(new_score == i * local_score / _INTENSIFICATION_TRIES_ + (_INTENSIFICATION_TRIES_ - i) * global_score / _INTENSIFICATION_TRIES_);
if (m_mpz_manager.is_one(m_tracker.get_value(q)))
return 1;
}
return 0;
}
// main search loop
lbool sls_engine::search() {
lbool res = l_undef;
double score = 0.0, old_score = 0.0;
unsigned new_const = (unsigned)-1, new_bit = 0;
mpz new_value;
move_type move;
unsigned plateau_cnt = 0;
score = rescore();
unsigned sz = m_assertions.size();
#if _PERC_STICKY_
expr * e = m_tracker.get_unsat_assertion(g, m_stats.m_moves);
#endif
#if _RESTARTS_ == 1
while (check_restart(m_stats.m_moves) && m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_) {
#elif _RESTARTS_ == 2
while (check_restart(plateau_cnt) && m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_) {
#elif _RESTARTS_ == 3
while (check_restart((unsigned)m_stats.m_stopwatch.get_current_seconds()) && m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_) {
#else
while (m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_) {
#endif
checkpoint();
m_stats.m_moves++;
#if _REAL_RS_ || _REAL_PBFS_
//m_tracker.debug_real(g, m_stats.m_moves);
#endif
#if _FOCUS_
#if _PERC_STICKY_
if (m_tracker.get_random_uint(16) % 100 >= _PERC_STICKY_ || m_mpz_manager.eq(m_tracker.get_value(e), m_one))
e = m_tracker.get_unsat_assertion(g, m_stats.m_moves);
#else
expr * e = m_tracker.get_unsat_assertion(m_assertions, m_stats.m_moves);
#endif
if (!e)
{
res = l_true;
goto bailout;
}
ptr_vector<func_decl> & to_evaluate = m_tracker.get_unsat_constants_walksat(e);
#else
ptr_vector<func_decl> & to_evaluate = m_tracker.get_unsat_constants_gsat(g, sz);
if (!to_evaluate.size())
{
res = l_true;
goto bailout;
}
#endif
#if _TYPE_RSTEP_
if (m_tracker.get_random_uint(16) % 1000 < _PERM_RSTEP_)
{
#if _TYPE_RSTEP_ == 1
m_evaluator.randomize_local(to_evaluate);
#elif _TYPE_RSTEP_ == 2
mk_random_move(to_evaluate);
#endif
#if _CACHE_TOP_SCORE_
score = m_tracker.get_top_sum() / g->size();
#else
score = top_score(g);
#endif
}
continue;
#endif
#if _WEIGHT_DIST_ == 4
m_tracker.set_weight_dist_factor(m_stats.m_stopwatch.get_current_seconds() / _TIMELIMIT_);
#endif
old_score = score;
new_const = (unsigned)-1;
#if _VNS_
score = find_best_move_vns(g, to_evaluate, score, new_const, new_value, new_bit, move);
#else
score = find_best_move(to_evaluate, score, new_const, new_value, new_bit, move);
#endif
if (new_const == static_cast<unsigned>(-1)) {
score = old_score;
plateau_cnt++;
#if _INTENSIFICATION_
handle_plateau(g, score);
//handle_plateau(g);
//e = m_tracker.get_unsat_assertion(g, m_stats.m_moves);
//to_evaluate = m_tracker.get_unsat_constants_walksat(e);
#else
#if _PERC_PLATEAU_MOVES_
if (m_tracker.get_random_uint(8) % 100 < _PERC_PLATEAU_MOVES_)
mk_random_move(to_evaluate);
else
#endif
#if _REPICK_
m_evaluator.randomize_local(m_assertions, m_stats.m_moves);
#else
m_evaluator.randomize_local(to_evaluate);
#endif
#endif
#if _CACHE_TOP_SCORE_
score = m_tracker.get_top_sum() / m_assertions.size();
#else
score = top_score(g);
#endif
}
else {
func_decl * fd = to_evaluate[new_const];
#if _REAL_RS_ || _REAL_PBFS_
score = serious_score(g, fd, new_value);
#else
score = incremental_score(fd, new_value);
#endif
}
}
bailout:
m_mpz_manager.del(new_value);
return res;
}
#if 0 // Old code.
// main search loop
lbool sls_engine::search_old() {
lbool res = l_undef;
double score = 0.0, old_score = 0.0;
unsigned new_const = (unsigned)-1, new_bit = 0;
mpz new_value;
move_type move;
score = rescore();
TRACE("sls", tout << "Starting search, initial score = " << std::setprecision(32) << score << std::endl;
tout << "Score distribution:";
for (unsigned i = 0; i < m_assertions.size(); i++)
tout << " " << std::setprecision(3) << m_tracker.get_score(m_assertions[i]);
tout << " TOP: " << score << std::endl;);
unsigned plateau_cnt = 0;
// Andreas: Why do we only allow so few plateaus?
#if _RESTARTS_
while (m_stats.m_stopwatch.get_current_seconds() < 200 * (m_stats.m_restarts + 1) * 0.2) {
//while (plateau_cnt < m_plateau_limit && m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_) {
#else
while (m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_ && (_RESTARTS_ == 0 || m_stats.m_moves < _RESTARTS_)) {
#endif
do {
checkpoint();
#if _WEIGHT_DIST_ == 4
m_tracker.set_weight_dist_factor(m_stats.m_stopwatch.get_current_seconds() / _TIMELIMIT_);
#endif
#if _TYPE_RSTEP_
if (m_tracker.get_random_uint(16) % 1000 < _PERM_RSTEP_)
{
#if _TYPE_RSTEP_ == 1
m_evaluator.randomize_local(g, m_stats.m_moves);
#elif _TYPE_RSTEP_ == 2
mk_random_move(g);
#endif
score = top_score(g);
if (score >= 1.0) {
bool all_true = true;
for (unsigned i = 0; i < g->size() && all_true; i++)
if (!m_mpz_manager.is_one(m_tracker.get_value(g->form(i))))
all_true = false;
if (all_true) {
res = l_true; // sat
goto bailout;
}
else
TRACE("sls", tout << "Imprecise 1.0 score" << std::endl;);
}
}
#endif
old_score = score;
new_const = (unsigned)-1;
ptr_vector<func_decl> & to_evaluate = m_tracker.get_unsat_constants(m_assertions, m_stats.m_moves);
if (!to_evaluate.size())
{
res = l_true;
goto bailout;
}
TRACE("sls_constants", tout << "Evaluating these constants: " << std::endl;
for (unsigned i = 0; i < to_evaluate.size(); i++)
tout << to_evaluate[i]->get_name() << std::endl;);
#if _VNS_
score = find_best_move_vns(to_evaluate, score, new_const, new_value, new_bit, move);
#else
score = find_best_move(to_evaluate, score, new_const, new_value, new_bit, move);
#endif
if (new_const == static_cast<unsigned>(-1)) {
TRACE("sls", tout << "Local maximum reached; unsatisfied constraints: " << std::endl;
for (unsigned i = 0; i < m_assertions.size(); i++) {
if (!m_mpz_manager.is_one(m_tracker.get_value(m_assertions[i])))
tout << mk_ismt2_pp(m_assertions[i], m_manager) << std::endl;
});
TRACE("sls_max", m_tracker.show_model(tout);
tout << "Scores: " << std::endl;
for (unsigned i = 0; i < m_assertions.size(); i++)
tout << mk_ismt2_pp(m_assertions[i], m_manager) << " ---> " <<
m_tracker.get_score(m_assertions[i]) << std::endl;);
// Andreas: If new_const == -1, shouldn't score = old_score anyway?
score = old_score;
}
else {
// Andreas: Why does randomizing not count as a move? (Now it does.)
m_stats.m_moves++;
func_decl * fd = to_evaluate[new_const];
TRACE("sls", tout << "Setting " << fd->get_name() << " to " << m_mpz_manager.to_string(new_value) << " (Move: ";
switch (move) {
case MV_FLIP:
tout << "Flip";
if (!m_manager.is_bool(fd->get_range())) tout << " #" << new_bit;
break;
case MV_INC:
tout << "+1";
break;
case MV_DEC:
tout << "-1";
break;
case MV_INV:
tout << "NEG";
break;
};
tout << ") ; new score = " << std::setprecision(32) << score << std::endl;);
switch (move) {
case MV_FLIP: m_stats.m_flips++; break;
case MV_INC: m_stats.m_incs++; break;
case MV_DEC: m_stats.m_decs++; break;
case MV_INV: m_stats.m_invs++; break;
case MV_UMIN: m_stats.m_umins++; break;
case MV_MUL2: m_stats.m_mul2s++; break;
case MV_MUL3: m_stats.m_mul3s++; break;
case MV_DIV2: m_stats.m_div2s++; break;
}
#if _REAL_RS_ || _REAL_PBFS_
score = serious_score(fd, new_value);
#else
score = incremental_score(fd, new_value);
#endif
TRACE("sls", tout << "Score distribution:";
for (unsigned i = 0; i < m_assertions.size(); i++)
tout << " " << std::setprecision(3) << m_tracker.get_score(m_assertions[i]);
tout << " TOP: " << score << std::endl;);
}
if (score >= 0.99999) {
// if (score >= 1.0) {
// score could theoretically be imprecise.
// Andreas: it seems using top level score caching can make the score unprecise also in the other direction!
bool all_true = true;
for (unsigned i = 0; i < m_assertions.size() && all_true; i++)
if (!m_mpz_manager.is_one(m_tracker.get_value(m_assertions[i])))
all_true = false;
if (all_true) {
res = l_true; // sat
goto bailout;
}
else
TRACE("sls", tout << "Imprecise 1.0 score" << std::endl;);
}
/*
if (m_stats.m_moves % 100 == 0)
{
verbose_stream() << "(" << std::fixed << std::setprecision(10) << score << ")" << std::endl;
verbose_stream() << "(" << std::fixed << std::setprecision(2) << (m_stats.m_moves / m_stats.m_stopwatch.get_current_seconds()) << ")" << std::endl;
}*/
} while (score > old_score && res == l_undef);
// Andreas: Why do you check for old_score? This should always be equal due to the loop invariant.
if (score != old_score) {
report_tactic_progress("This should not happen I guess.", plateau_cnt);
plateau_cnt = 0;
}
else {
m_stats.m_moves++;
plateau_cnt++;
//report_tactic_progress("Plateau.", plateau_cnt);
// Andreas: Right now, a useless assignment is created in case of a restart. But we don't want to use restarts anyway.
//if (plateau_cnt < m_plateau_limit) {
TRACE("sls", tout << "In a plateau (" << plateau_cnt << "/" << m_plateau_limit << "); randomizing locally." << std::endl;);
#if _INTENSIFICATION_
handle_plateau(g, score);
//handle_plateau();
#else
m_evaluator.randomize_local(m_assertions, m_stats.m_moves);
#endif
//mk_random_move();
score = top_score();
if (score >= 1.0) {
bool all_true = true;
for (unsigned i = 0; i < m_assertions.size() && all_true; i++)
if (!m_mpz_manager.is_one(m_tracker.get_value(m_assertions[i])))
all_true = false;
if (all_true) {
res = l_true; // sat
goto bailout;
}
else
TRACE("sls", tout << "Imprecise 1.0 score" << std::endl;);
}
}
}
#endif
bailout:
m_mpz_manager.del(new_value);
return res;
}
void sls_engine::operator()(goal_ref const & g, model_converter_ref & mc) {
if (g->inconsistent()) {
mc = 0;
return;
}
m_produce_models = g->models_enabled();
for (unsigned i = 0; i < g->size(); i++)
assert_expr(g->form(i));
verbose_stream() << "_BFS_ " << _BFS_ << std::endl;
verbose_stream() << "_FOCUS_ " << _FOCUS_ << std::endl;
verbose_stream() << "_PERC_STICKY_ " << _PERC_STICKY_ << std::endl;
verbose_stream() << "_RESTARTS_ " << _RESTARTS_ << std::endl;
verbose_stream() << "_RESTART_LIMIT_ " << _RESTART_LIMIT_ << std::endl;
verbose_stream() << "_RESTART_INIT_ " << _RESTART_INIT_ << std::endl;
verbose_stream() << "_RESTART_SCHEME_ " << _RESTART_SCHEME_ << std::endl;
verbose_stream() << "_TIMELIMIT_ " << _TIMELIMIT_ << std::endl;
verbose_stream() << "_SCORE_AND_AVG_ " << _SCORE_AND_AVG_ << std::endl;
verbose_stream() << "_SCORE_OR_MUL_ " << _SCORE_OR_MUL_ << std::endl;
verbose_stream() << "_VNS_ " << _VNS_ << std::endl;
verbose_stream() << "_WEIGHT_DIST_ " << _WEIGHT_DIST_ << std::endl;
verbose_stream() << "_WEIGHT_DIST_FACTOR_ " << std::fixed << std::setprecision(2) << _WEIGHT_DIST_FACTOR_ << std::endl;
verbose_stream() << "_INTENSIFICATION_ " << _INTENSIFICATION_ << std::endl;
verbose_stream() << "_INTENSIFICATION_TRIES_ " << _INTENSIFICATION_TRIES_ << std::endl;
verbose_stream() << "_PERC_PLATEAU_MOVES_ " << _PERC_PLATEAU_MOVES_ << std::endl;
verbose_stream() << "_REPICK_ " << _REPICK_ << std::endl;
verbose_stream() << "_UCT_ " << _UCT_ << std::endl;
verbose_stream() << "_UCT_CONSTANT_ " << std::fixed << std::setprecision(2) << _UCT_CONSTANT_ << std::endl;
verbose_stream() << "_UCT_RESET_ " << _UCT_RESET_ << std::endl;
verbose_stream() << "_UCT_INIT_ " << _UCT_INIT_ << std::endl;
verbose_stream() << "_PROBABILISTIC_UCT_ " << _PROBABILISTIC_UCT_ << std::endl;
verbose_stream() << "_UCT_EPS_ " << std::fixed << std::setprecision(4) << _UCT_EPS_ << std::endl;
verbose_stream() << "_USE_ADDSUB_ " << _USE_ADDSUB_ << std::endl;
verbose_stream() << "_USE_MUL2DIV2_ " << _USE_MUL2DIV2_ << std::endl;
verbose_stream() << "_USE_MUL3_ " << _USE_MUL3_ << std::endl;
verbose_stream() << "_USE_UNARY_MINUS_ " << _USE_UNARY_MINUS_ << std::endl;
verbose_stream() << "_UNIFORM_RANDOM_ " << _UNIFORM_RANDOM_ << std::endl;
verbose_stream() << "_REAL_RS_ " << _REAL_RS_ << std::endl;
verbose_stream() << "_REAL_PBFS_ " << _REAL_PBFS_ << std::endl;
verbose_stream() << "_SKIP_BITS_ " << _SKIP_BITS_ << std::endl;
verbose_stream() << "_PERC_CHANGE_ " << _PERC_CHANGE_ << std::endl;
verbose_stream() << "_TYPE_RSTEP_ " << _TYPE_RSTEP_ << std::endl;
verbose_stream() << "_PERM_RSTEP_ " << _PERM_RSTEP_ << std::endl;
verbose_stream() << "_EARLY_PRUNE_ " << _EARLY_PRUNE_ << std::endl;
verbose_stream() << "_CACHE_TOP_SCORE_ " << _CACHE_TOP_SCORE_ << std::endl;
#if _WEIGHT_DIST_ == 4
m_tracker.set_weight_dist_factor(m_stats.m_stopwatch.get_current_seconds() / _TIMELIMIT_);
#endif
#if _WEIGHT_TOGGLE_
m_tracker.set_weight_dist_factor(_WEIGHT_DIST_FACTOR_);
#endif
m_tracker.initialize(m_assertions);
lbool res = l_undef;
m_restart_limit = _RESTART_LIMIT_;
do {
checkpoint();
report_tactic_progress("Searching... restarts left:", m_max_restarts - m_stats.m_restarts);
res = search();
if (res == l_undef)
{
#if _RESTART_INIT_
m_tracker.randomize();
#else
m_tracker.reset(m_assertions);
#endif
}
} while (m_stats.m_stopwatch.get_current_seconds() < _TIMELIMIT_ && res != l_true && m_stats.m_restarts++ < m_max_restarts);
verbose_stream() << "(restarts: " << m_stats.m_restarts << " flips: " << m_stats.m_moves << " time: " << std::fixed << std::setprecision(2) << m_stats.m_stopwatch.get_current_seconds() << " fps: " << (m_stats.m_moves / m_stats.m_stopwatch.get_current_seconds()) << ")" << std::endl;
if (res == l_true) {
report_tactic_progress("Number of flips:", m_stats.m_moves);
for (unsigned i = 0; i < g->size(); i++)
if (!m_mpz_manager.is_one(m_tracker.get_value(g->form(i))))
{
verbose_stream() << "Terminated before all assertions were SAT!" << std::endl;
NOT_IMPLEMENTED_YET();
}
if (m_produce_models) {
model_ref mdl = m_tracker.get_model();
mc = model2model_converter(mdl.get());
TRACE("sls_model", mc->display(tout););
}
g->reset();
}
else
mc = 0;
}
unsigned sls_engine::check_restart(unsigned curr_value)
{
if (curr_value > m_restart_limit)
{
#if _RESTART_SCHEME_ == 4
m_restart_limit += (m_stats.m_restarts & (m_stats.m_restarts + 1)) ? _RESTART_LIMIT_ : (_RESTART_LIMIT_ * m_stats.m_restarts + 1);
#elif _RESTART_SCHEME_ == 3
m_restart_limit += (unsigned)get_restart_armin(m_stats.m_restarts + 1) * _RESTART_LIMIT_;
#elif _RESTART_SCHEME_ == 2
m_restart_limit += get_luby(m_stats.m_restarts + 1) * _RESTART_LIMIT_;
#elif _RESTART_SCHEME_ == 1
if (m_stats.m_restarts & 1)
m_restart_limit += _RESTART_LIMIT_;
else
m_restart_limit += (2 << (m_stats.m_restarts >> 1)) * _RESTART_LIMIT_;
#else
m_restart_limit += _RESTART_LIMIT_;
#endif
#if _WEIGHT_TOGGLE_
printf("Setting weight: %f\n", _WEIGHT_DIST_FACTOR_ * (((m_stats.m_restarts & 2) == 0) + 1));
m_tracker.set_weight_dist_factor(_WEIGHT_DIST_FACTOR_ * (((m_stats.m_restarts & 2) == 0) + 1));
#endif
return 0;
}
return 1;
}