Many resources are needed to download a project. Please understand that we have to compensate our server costs. Thank you in advance. Project price only 1 $
You can buy this project and download/modify it how often you want.
/*++
Copyright (c) 2006 Microsoft Corporation
Module Name:
smt_theory.h
Abstract:
Author:
Leonardo de Moura (leonardo) 2008-02-20.
Revision History:
--*/
#pragma once
#include "ast/ast_pp.h"
#include "smt/smt_enode.h"
#include "smt/smt_quantifier.h"
#include "util/obj_hashtable.h"
#include "util/statistics.h"
#include
namespace smt {
class model_generator;
class model_value_proc;
class theory {
protected:
theory_id m_id;
context & ctx;
ast_manager & m;
enode_vector m_var2enode;
unsigned_vector m_var2enode_lim;
unsigned m_lazy_scopes;
bool m_lazy;
friend class context;
friend class arith_value;
protected:
/* ---------------------------------------------------
In the logical context, expressions are 'internalized'. That
is, the logical context creates auxiliary data-structures
(e.g., enodes) and attach them to the expressions. The logical
context does not know the internals of each theory. So, during
the internalization process, it notifies the theory (plugin)
whenever it finds an application with a theory function
symbol.
A theory variable created at scope level n must be deleted
when scope level n is backtracked.
The logical context uses the method is_attached_to_var
to decide whether an enode is already associated with a theory
variable or not.
------------------------------------------------------ */
virtual theory_var mk_var(enode * n) {
SASSERT(!is_attached_to_var(n));
theory_var v = m_var2enode.size();
m_var2enode.push_back(n);
return v;
}
theory_var get_th_var(expr* e) const;
theory_var get_th_var(enode* n) const {
return n->get_th_var(get_id());
}
bool lazy_push();
bool lazy_pop(unsigned& num_scopes);
void force_push();
public:
/**
\brief Return true if the given enode is attached to a
variable of the theory.
\remark The result is not equivalent to
n->get_th_var(get_id()) != null_theory_var
A theory variable v may be in the list of variables of n,
but it may be inherited from another enode n' during an
equivalence class merge. That is, get_enode(v) != n.
*/
bool is_attached_to_var(enode const * n) const {
theory_var v = n->get_th_var(get_id());
return v != null_theory_var && get_enode(v) == n;
}
struct scoped_trace_stream {
ast_manager& m;
scoped_trace_stream(ast_manager& m, std::function& fn): m(m) {
if (m.has_trace_stream()) {
fn();
}
}
scoped_trace_stream(theory& th, std::function& fn): m(th.get_manager()) {
if (m.has_trace_stream()) {
expr_ref body(fn(), m);
th.log_axiom_instantiation(body);
}
}
scoped_trace_stream(theory& th, std::function& fn): m(th.get_manager()) {
if (m.has_trace_stream()) {
th.log_axiom_instantiation(fn());
}
}
scoped_trace_stream(theory& th, literal_vector const& lits): m(th.get_manager()) {
if (m.has_trace_stream()) {
th.log_axiom_instantiation(lits);
}
}
scoped_trace_stream(theory& th, literal_buffer const& lits) : m(th.get_manager()) {
if (m.has_trace_stream()) {
th.log_axiom_instantiation(lits);
}
}
scoped_trace_stream(theory& th, literal lit): m(th.get_manager()) {
if (m.has_trace_stream()) {
literal_vector lits;
lits.push_back(lit);
th.log_axiom_instantiation(lits);
}
}
scoped_trace_stream(theory& th, literal lit1, literal lit2): m(th.get_manager()) {
if (m.has_trace_stream()) {
literal_vector lits;
lits.push_back(lit1);
lits.push_back(lit2);
th.log_axiom_instantiation(lits);
}
}
scoped_trace_stream(theory& th, std::function& fn): m(th.get_manager()) {
if (m.has_trace_stream()) {
literal_vector ls;
ls.push_back(fn());
th.log_axiom_instantiation(ls);
}
}
~scoped_trace_stream() {
if (m.has_trace_stream()) {
m.trace_stream() << "[end-of-instance]\n";
}
}
};
struct if_trace_stream {
ast_manager& m;
if_trace_stream(ast_manager& m, std::function& fn): m(m) {
if (m.has_trace_stream()) {
fn();
}
}
};
protected:
/**
\brief Return true if the theory uses default internalization:
"the internalization of an application internalizes all arguments".
Theories like arithmetic do not use default internalization.
For example, in the application (+ a (+ b c)), no enode is created
for (+ b c).
*/
virtual bool default_internalizer() const {
return true;
}
/**
\brief This method is invoked by the logical context when
atom is being internalized. The theory may return false if it
does not want to implement the given predicate symbol.
After the execution of this method the given atom must be
associated with a new boolean variable.
*/
virtual bool internalize_atom(app * atom, bool gate_ctx) = 0;
/**
\brief This method is invoked by the logical context
after the given equality atom is internalized.
*/
virtual void internalize_eq_eh(app * atom, bool_var v) {
}
/**
\brief This method is invoked by the logical context when
the term is being internalized. The theory may return false
if it does not want to implement the given function symbol.
After the execution of this method the given term must be
associated with a new enode.
*/
virtual bool internalize_term(app * term) = 0;
/**
\brief Apply (interpreted) sort constraints on the given enode.
*/
virtual void apply_sort_cnstr(enode * n, sort * s) {
}
/**
\brief This method is invoked when a truth value is
assigned to the given boolean variable.
*/
virtual void assign_eh(bool_var v, bool is_true) {
}
/**
\brief use the theory to determine phase of the variable.
*/
virtual lbool get_phase(bool_var v) {
return l_undef;
}
/**
\brief Equality propagation (v1 = v2): Core -> Theory
*/
virtual void new_eq_eh(theory_var v1, theory_var v2) = 0;
/**
\brief Return true if the theory does something with the
disequalities implied by the core.
*/
virtual bool use_diseqs() const {
return true;
}
/**
\brief Disequality propagation (v1 /= v2): Core -> Theory
*/
virtual void new_diseq_eh(theory_var v1, theory_var v2) = 0;
/**
\brief This method is invoked when the theory application n
is marked as relevant.
*/
virtual void relevant_eh(app * n) {
}
/**
\brief This method is invoked when a new backtracking point
is created.
*/
virtual void push_scope_eh();
/**
\brief This method is invoked during backtracking.
*/
virtual void pop_scope_eh(unsigned num_scopes);
/**
\brief This method is invoked when the logical context is being
restarted.
*/
virtual void restart_eh() {
}
/**
\brief This method is called by smt_context before the search starts
to get any extra assumptions the theory wants to use.
(See theory_str for an example)
*/
virtual void add_theory_assumptions(expr_ref_vector & assumptions) {
}
/**
\brief This method is called from the smt_context when an unsat core is generated.
The theory may change the answer to UNKNOWN by returning l_undef from this method.
*/
virtual lbool validate_unsat_core(expr_ref_vector & unsat_core) {
return l_false;
}
/**
\brief This method is called from the smt_context when an unsat core is generated.
The theory may tell the solver to perform iterative deepening by invalidating
this unsat core and increasing some resource constraints.
*/
virtual bool should_research(expr_ref_vector & unsat_core) {
return false;
}
/**
\brief This method is invoked before the search starts.
*/
virtual void init_search_eh() {
}
/**
\brief This method is invoked when the logical context assigned
a truth value to all boolean variables and no inconsistency was
detected.
*/
virtual final_check_status final_check_eh() {
return FC_DONE;
}
/**
\brief Parametric theories (e.g. Arrays) should implement this method.
See example in context::is_shared
*/
virtual bool is_shared(theory_var v) const {
return false;
}
/**
\brief Determine if node \c n under parent \c p is in a beta redex position.
*/
virtual bool is_beta_redex(enode* p, enode* n) const {
return false;
}
/**
\brief Return true if the theory has something to propagate
*/
virtual bool can_propagate() {
return false;
}
/**
\brief This method is invoked to give a theory a chance to perform
theory propagation.
*/
virtual void propagate() {
}
/**
\brief This method allows a theory to contribute to
disequality propagation.
*/
virtual justification * why_is_diseq(theory_var v1, theory_var v2) {
return nullptr;
}
/**
\brief Just releases memory.
*/
virtual void flush_eh() {
}
/**
\brief This method is invoked when the logical context is being reset.
*/
virtual void reset_eh();
// ----------------------------------------------------
//
// Model validation
//
// ----------------------------------------------------
virtual void validate_model(proto_model& mdl) {}
// ----------------------------------------------------
//
// Conflict resolution event handler
//
// ----------------------------------------------------
public:
/**
\brief This method is invoked when a theory atom is used
during conflict resolution. This allows the theory to bump
the activity of the enodes contained in the given atom.
*/
virtual void conflict_resolution_eh(app * atom, bool_var v) {
}
public:
theory(context& ctx, family_id fid);
virtual ~theory();
virtual void setup() {}
virtual void init() {}
theory_id get_id() const {
return m_id;
}
family_id get_family_id() const {
return m_id;
}
context & get_context() const {
return ctx;
}
ast_manager & get_manager() const {
return m;
}
smt_params const& get_fparams() const;
enode * get_enode(theory_var v) const {
SASSERT(v < static_cast(m_var2enode.size()));
return m_var2enode[v];
}
app * get_expr(theory_var v) const {
return get_enode(v)->get_expr();
}
/**
\brief Return the equivalence class representative
of the given theory variable.
*/
theory_var get_representative(theory_var v) const {
SASSERT(v != null_theory_var);
theory_var r = get_enode(v)->get_root()->get_th_var(get_id());
SASSERT(r != null_theory_var);
return r;
}
/**
\brief Return true if the theory variable is the representative
of its equivalence class.
*/
bool is_representative(theory_var v) const {
return get_representative(v) == v;
}
virtual bool is_safe_to_copy(bool_var v) const { return true; }
unsigned get_num_vars() const {
return m_var2enode.size();
}
unsigned get_old_num_vars(unsigned num_scopes) const {
return m_var2enode_lim[m_var2enode_lim.size() - num_scopes];
}
virtual void display(std::ostream & out) const = 0;
virtual void display_var2enode(std::ostream & out) const;
virtual void collect_statistics(::statistics & st) const {
}
std::ostream& display_app(std::ostream & out, app * n) const;
std::ostream& display_flat_app(std::ostream & out, app * n) const;
protected:
void log_axiom_instantiation(app * r, unsigned axiom_id = UINT_MAX, unsigned num_bindings = 0,
app * const * bindings = nullptr, unsigned pattern_id = UINT_MAX,
const vector> & used_enodes = vector>());
void log_axiom_instantiation(expr * r, unsigned axiom_id = UINT_MAX, unsigned num_bindings = 0,
app * const * bindings = nullptr, unsigned pattern_id = UINT_MAX,
const vector> & used_enodes = vector>()) {
log_axiom_instantiation(to_app(r), axiom_id, num_bindings, bindings, pattern_id, used_enodes);
}
void log_axiom_instantiation(literal_vector const& ls);
void log_axiom_instantiation(literal_buffer const& ls);
void log_axiom_instantiation(app * r, unsigned num_blamed_enodes, enode ** blamed_enodes) {
vector> used_enodes;
for (unsigned i = 0; i < num_blamed_enodes; ++i) {
used_enodes.push_back(std::make_tuple(nullptr, blamed_enodes[i]));
}
log_axiom_instantiation(r, UINT_MAX, 0, nullptr, UINT_MAX, used_enodes);
}
void log_axiom_unit(app* r) {
expr_ref _r(r, m);
log_axiom_instantiation(r);
m.trace_stream() << "[end-of-instance]\n";
}
public:
/**
\brief Assume eqs between variable that are equal with respect to the given table.
Table is a hashtable indexed by the variable value.
table.contains(v) should be true if there is v' in table such that assignment of
v is equal to v'.
This method assumes that class VarValueTable contains the methods:
- void reset()
- theory_var insert_if_not_there(theory_var v)
*/
template
bool assume_eqs(VarValueTable & table) {
TRACE("assume_eqs", tout << "starting...\n";);
table.reset();
bool result = false;
int num = get_num_vars();
for (theory_var v = 0; v < num; v++) {
enode * n = get_enode(v);
theory_var other = null_theory_var;
TRACE("assume_eqs",
tout << "#" << n->get_owner_id() << " is_relevant_and_shared: " << is_relevant_and_shared(n) << "\n";);
if (n != nullptr && is_relevant_and_shared(n)) {
other = table.insert_if_not_there(v);
if (other != v) {
enode * n2 = get_enode(other);
TRACE("assume_eqs", tout << "value(#" << n->get_owner_id() << ") = value(#" << n2->get_owner_id() << ")\n";);
if (assume_eq(n, n2)) {
TRACE("assume_eqs", tout << "new assumed eq\n";);
result = true;
}
}
}
}
return result;
}
/**
\brief When an eq atom n is created during the search, the default behavior is
to make sure that the n->get_arg(0)->get_id() < n->get_arg(1)->get_id().
This may create some redundant atoms, since some theories/families use different
conventions in their simplifiers. For example, arithmetic always force a numeral
to be in the right hand side. So, this method should be redefined if the default
behavior conflicts with a convention used by the theory/family.
*/
virtual app * mk_eq_atom(expr * lhs, expr * rhs) {
ast_manager& m = get_manager();
if (lhs->get_id() > rhs->get_id())
std::swap(lhs, rhs);
if (m.are_distinct(lhs, rhs))
return m.mk_false();
if (m.are_equal(lhs, rhs))
return m.mk_true();
return get_manager().mk_eq(lhs, rhs);
}
literal mk_eq(expr * a, expr * b, bool gate_ctx);
literal mk_preferred_eq(expr* a, expr* b);
literal mk_literal(expr* e);
enode* ensure_enode(expr* e);
enode* get_root(expr* e) { return ensure_enode(e)->get_root(); }
// -----------------------------------
//
// Model generation
//
// -----------------------------------
/**
\brief Return true if theory support model construction
*/
virtual bool build_models() const {
return true;
}
virtual void init_model(model_generator & m) {
}
virtual void finalize_model(model_generator & m) {
}
/**
\brief Return a functor that can build the value (interpretation) for n.
*/
virtual model_value_proc * mk_value(enode * n, model_generator & mg) {
return nullptr;
}
virtual bool include_func_interp(func_decl* f) {
return false;
}
// -----------------------------------
//
// Model checker
//
// -----------------------------------
virtual bool get_value(enode * n, expr_ref & r) {
return false;
}
virtual char const * get_name() const { return "unknown"; }
// -----------------------------------
//
// Return a fresh new instance of the given theory.
// This function is used to create a fresh context (new_ctx) containing the same theories of the context that owns this theory.
//
// We need the parameter new_ctx because of user_smt_theory :-(
//
// -----------------------------------
virtual theory * mk_fresh(context * new_ctx) = 0;
protected:
// ----------------------------------------------------
//
// Auxiliary methods for assume_eqs
//
// smt::context is not defined at this point.
//
// ----------------------------------------------------
bool is_relevant_and_shared(enode * n) const;
bool assume_eq(enode * n1, enode * n2);
/**
* \brief theory plugin for fixed values.
*/
virtual bool is_fixed_propagated(theory_var v, expr_ref& val, literal_vector & explain) { return false; }
};
};
