1160 lines
37 KiB
Python
1160 lines
37 KiB
Python
"""
|
|
SMT-based Model Checking Module for RS with Concentrations and Context Automaton
|
|
"""
|
|
|
|
from z3 import *
|
|
from time import time
|
|
from sys import stdout
|
|
from itertools import chain
|
|
import resource
|
|
from colour import *
|
|
|
|
from logics import rsLTL_Encoder
|
|
from logics import ParamConstr_Encoder
|
|
|
|
from rs.reaction_system_with_concentrations_param import ParameterObj, is_param
|
|
|
|
|
|
def z3_max(a, b):
|
|
return If(a > b, a, b)
|
|
|
|
|
|
class SmtCheckerRSCParam(object):
|
|
def __init__(self, rsca, optimise=False):
|
|
rsca.sanity_check()
|
|
|
|
if not rsca.is_concentr_and_param_compatible():
|
|
raise RuntimeError(
|
|
"RS and CA with concentrations (and parameters) expected"
|
|
)
|
|
|
|
self.rs = rsca.rs
|
|
self.ca = rsca.ca
|
|
|
|
self.optimise = optimise
|
|
|
|
self.initialise()
|
|
|
|
def initialise(self):
|
|
"""Initialises all the variables used by the checker"""
|
|
|
|
# "Currently" used variables (loaded from self.path_v...)
|
|
self.v = None
|
|
self.v_ctx = None
|
|
self.ca_state = None
|
|
|
|
# intermediate products:
|
|
self.v_improd = None
|
|
self.v_improd_for_entities = None
|
|
|
|
# Per-path variables
|
|
self.path_v = dict()
|
|
self.path_v_ctx = dict()
|
|
self.path_ca_state = dict()
|
|
|
|
# intermediate products:
|
|
self.path_v_improd = dict()
|
|
self.path_v_improd_for_entities = dict()
|
|
|
|
# parameters:
|
|
self.v_param = dict()
|
|
|
|
self.next_level_to_encode = 0
|
|
|
|
#
|
|
# WARNING: improd vs. improducible
|
|
# there is some confusion related to the variable naming:
|
|
#
|
|
# * improd - intermediate products
|
|
# * improducible - entities that are never produces (there is no reaction that produces that entity)
|
|
#
|
|
|
|
# TODO: number of loops == number of paths
|
|
self.loop_position = None
|
|
self.path_loop_position = dict()
|
|
|
|
if self.optimise:
|
|
self.solver = Optimize()
|
|
else:
|
|
self.solver = Solver() # For("QF_FD")
|
|
|
|
self.verification_time = None
|
|
|
|
self.prepare_param_variables()
|
|
|
|
def reset(self):
|
|
"""Reinitialises the state of the checker"""
|
|
|
|
self.initialise()
|
|
|
|
def prepare_all_variables(self, num_of_paths):
|
|
for path_idx in range(num_of_paths):
|
|
self.prepare_all_path_variables(path_idx)
|
|
self.next_level_to_encode += 1
|
|
|
|
def prepare_all_path_variables(self, path_idx):
|
|
"""Prepares the variables for a given path index"""
|
|
|
|
print_info(
|
|
"Preparing variables for path={:d} (level={:d})".format(
|
|
path_idx, self.next_level_to_encode
|
|
)
|
|
)
|
|
|
|
self.prepare_state_variables(path_idx)
|
|
self.prepare_context_variables(path_idx)
|
|
self.prepare_intermediate_product_variables(path_idx)
|
|
self.prepare_loop_position_variables(path_idx)
|
|
|
|
def prepare_loop_position_variables(self, path_idx):
|
|
"""Prepares the variables for loop positions"""
|
|
|
|
self.path_loop_position[path_idx] = Int("p{:d}_loop_pos".format(path_idx))
|
|
|
|
def prepare_context_variables(self, path_idx):
|
|
"""Prepares all the context variables"""
|
|
|
|
level = self.next_level_to_encode
|
|
|
|
self.path_v_ctx.setdefault(path_idx, [])
|
|
assert len(self.path_v_ctx[path_idx]) == level
|
|
|
|
variables = []
|
|
for entity in self.rs.background_set:
|
|
new_var = Int("p{:d}C{:d}_{:s}".format(path_idx, level, entity))
|
|
variables.append(new_var)
|
|
|
|
self.path_v_ctx[path_idx].append(variables)
|
|
|
|
def prepare_state_variables(self, path_idx):
|
|
"""Prepares all the state variables"""
|
|
|
|
level = self.next_level_to_encode
|
|
|
|
# RS vars
|
|
self.path_v.setdefault(path_idx, [])
|
|
assert len(self.path_v[path_idx]) == level
|
|
|
|
variables = []
|
|
for entity in self.rs.background_set:
|
|
new_var = Int("p{:d}L{:d}_{:s}".format(path_idx, level, entity))
|
|
variables.append(new_var)
|
|
self.path_v[path_idx].append(variables)
|
|
|
|
# Context automaton states:
|
|
self.path_ca_state.setdefault(path_idx, [])
|
|
assert len(self.path_ca_state[path_idx]) == level
|
|
|
|
ca_state_var = Int("p{:d}CA{:d}_state".format(path_idx, level))
|
|
self.path_ca_state[path_idx].append(ca_state_var)
|
|
|
|
def prepare_intermediate_product_variables(self, path_idx):
|
|
"""
|
|
Prepares the intermediate product variables
|
|
carrying the individual concentration levels produced
|
|
by the reactions.
|
|
|
|
These variables are used later on to encode the final
|
|
concentration levels for all the entities
|
|
"""
|
|
|
|
level = self.next_level_to_encode
|
|
|
|
self.path_v_improd.setdefault(path_idx, [])
|
|
self.path_v_improd_for_entities.setdefault(path_idx, [])
|
|
|
|
if level < 1:
|
|
#
|
|
# If we are at level==0, we add a dummy "level"
|
|
# to match the indices of of the successors
|
|
# which are always at level+1.
|
|
#
|
|
self.path_v_improd[path_idx].append(None)
|
|
self.path_v_improd_for_entities[path_idx].append(None)
|
|
|
|
reactions_dict = dict()
|
|
number_of_reactions = len(self.rs.reactions)
|
|
|
|
all_entities_dict = dict()
|
|
|
|
for reaction in self.rs.reactions:
|
|
*_, products = reaction
|
|
|
|
reaction_id = self.rs.reactions.index(reaction)
|
|
|
|
entities_dict = dict()
|
|
|
|
if is_param(products):
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
entity_name = self.rs.get_entity_name(entity)
|
|
new_var = Int(
|
|
"p{:d}L{:d}_ImProd_r{:d}_{:s}".format(
|
|
path_idx, level, reaction_id, entity_name
|
|
)
|
|
)
|
|
entities_dict[entity] = new_var
|
|
|
|
all_entities_dict.setdefault(entity, [])
|
|
all_entities_dict[entity].append(new_var)
|
|
|
|
else:
|
|
for entity, conc in products:
|
|
entity_name = self.rs.get_entity_name(entity)
|
|
new_var = Int(
|
|
"p{:d}L{:d}_ImProd_r{:d}_{:s}".format(
|
|
path_idx, level, reaction_id, entity_name
|
|
)
|
|
)
|
|
entities_dict[entity] = new_var
|
|
|
|
all_entities_dict.setdefault(entity, [])
|
|
all_entities_dict[entity].append(new_var)
|
|
|
|
reactions_dict[reaction_id] = entities_dict
|
|
|
|
self.path_v_improd[path_idx].append(reactions_dict)
|
|
self.path_v_improd_for_entities[path_idx].append(all_entities_dict)
|
|
|
|
def prepare_param_variables(self):
|
|
"""
|
|
Prepares variables for parameters
|
|
|
|
A parameter (it's valuation) is a subset of the background set,
|
|
therefore we need separate variables for each element of the
|
|
background set.
|
|
"""
|
|
|
|
for param_name in self.rs.parameters.keys():
|
|
# we start collecting bg-related vars for the given param
|
|
vars_for_param = []
|
|
|
|
for entity in self.rs.ordered_list_of_bgset_ids:
|
|
new_var = Int(
|
|
"Pm{:s}_{:s}".format(param_name, self.rs.get_entity_name(entity))
|
|
)
|
|
vars_for_param.append(new_var)
|
|
|
|
self.v_param[param_name] = vars_for_param
|
|
|
|
def get_enc_param(self, param_name, entity_name):
|
|
"""
|
|
Returns encoded param[entity_name]
|
|
"""
|
|
entity_id = self.rs.get_entity_id(entity_name)
|
|
return self.v_param[param_name][entity_id]
|
|
|
|
def load_varset_for_path(self, path_idx):
|
|
"""
|
|
Loads the the variables for the path with path_idx
|
|
"""
|
|
self.v = self.path_v[path_idx]
|
|
self.v_ctx = self.path_v_ctx[path_idx]
|
|
self.ca_state = self.path_ca_state[path_idx]
|
|
self.v_improd = self.path_v_improd[path_idx]
|
|
self.v_improd_for_entities = self.path_v_improd_for_entities[path_idx]
|
|
self.loop_position = self.path_loop_position[path_idx]
|
|
|
|
def enc_param_concentration_levels_assertion(self):
|
|
"""
|
|
Assertions for the parameter variables
|
|
"""
|
|
|
|
if len(self.v_param) == 0:
|
|
return True
|
|
|
|
enc_param_gz = True
|
|
enc_non_empty = True
|
|
|
|
for param_vars in self.v_param.values():
|
|
enc_param_at_least_one = False
|
|
for pvar in param_vars:
|
|
# TODO: fixed upper limit: 100 (have a per-param setting for that)
|
|
enc_param_gz = simplify(And(enc_param_gz, pvar >= 0, pvar < 100))
|
|
enc_param_at_least_one = simplify(Or(enc_param_at_least_one, pvar > 0))
|
|
|
|
enc_non_empty = simplify(And(enc_non_empty, enc_param_at_least_one))
|
|
|
|
return simplify(And(enc_param_gz, enc_non_empty))
|
|
|
|
def assert_param_optimisation(self):
|
|
for param_vars in self.v_param.values():
|
|
for pvar in param_vars:
|
|
# self.solver.add_soft(pvar == 0)
|
|
self.solver.minimize(pvar)
|
|
|
|
def enc_concentration_levels_assertion(self, level, path_idx):
|
|
"""
|
|
Encodes assertions that (some) variables need to be >=0
|
|
|
|
We do not need to actually control all the variables,
|
|
only those that can possibly go below 0.
|
|
"""
|
|
|
|
print_info(
|
|
"Concentration level assertions for path={:d} (level={:d})".format(
|
|
path_idx, level
|
|
)
|
|
)
|
|
|
|
enc_gz = True
|
|
|
|
for e_i in self.rs.set_of_bgset_ids:
|
|
var = self.path_v[path_idx][level][e_i]
|
|
var_ctx = self.path_v_ctx[path_idx][level][e_i]
|
|
e_max = self.rs.get_max_concentration_level(e_i)
|
|
enc_gz = simplify(
|
|
And(enc_gz, var >= 0, var_ctx >= 0, var <= e_max, var_ctx <= e_max)
|
|
)
|
|
|
|
vars_per_reaction = self.path_v_improd_for_entities[path_idx][level + 1]
|
|
if e_i in vars_per_reaction:
|
|
for var_improd in vars_per_reaction[e_i]:
|
|
enc_gz = simplify(And(enc_gz, var_improd >= 0, var_improd <= e_max))
|
|
|
|
return enc_gz
|
|
|
|
def enc_init_state(self, level, path_idx):
|
|
"""Encodes the initial state at the given level"""
|
|
|
|
rs_init_state_enc = True
|
|
|
|
for v in self.path_v[path_idx][level]:
|
|
# the initial concentration levels are zeroed
|
|
rs_init_state_enc = simplify(And(rs_init_state_enc, v == 0))
|
|
|
|
ca_init_state_enc = (
|
|
self.path_ca_state[path_idx][level] == self.ca.get_init_state_id()
|
|
)
|
|
|
|
init_state_enc = simplify(And(rs_init_state_enc, ca_init_state_enc))
|
|
|
|
return init_state_enc
|
|
|
|
def enc_transition_relation(self, level, path_idx):
|
|
return simplify(
|
|
And(
|
|
self.enc_rs_trans(level, path_idx),
|
|
self.enc_automaton_trans(level, path_idx),
|
|
)
|
|
)
|
|
|
|
def enc_param_sanity_for_reactions(self):
|
|
"""R < I constraint (R n I = 0)"""
|
|
|
|
rct_inh_constr = True
|
|
|
|
for reactants, inhibitors, products in self.rs.reactions:
|
|
if is_param(reactants) or is_param(inhibitors):
|
|
# 1. R and I
|
|
if is_param(reactants) and is_param(inhibitors):
|
|
rct_param_name = reactants.name
|
|
inh_param_name = inhibitors.name
|
|
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
rct_inh_constr = And(
|
|
rct_inh_constr,
|
|
Implies(
|
|
self.v_param[inh_param_name][entity] > 0,
|
|
self.v_param[rct_param_name][entity]
|
|
< self.v_param[inh_param_name][entity],
|
|
),
|
|
)
|
|
|
|
elif (not is_param(reactants)) and is_param(inhibitors):
|
|
inh_param_name = inhibitors.name
|
|
|
|
for entity, conc in reactants:
|
|
assert conc > 0, "Unexpected concentration level!"
|
|
rct_inh_constr = And(
|
|
rct_inh_constr,
|
|
Implies(
|
|
self.v_param[inh_param_name][entity] > 0,
|
|
conc < self.v_param[inh_param_name][entity],
|
|
),
|
|
)
|
|
|
|
elif is_param(reactants) and (not is_param(inhibitors)):
|
|
rct_param_name = reactants.name
|
|
|
|
for entity, conc in inhibitors:
|
|
assert conc > 0, "Unexpected concentration level!"
|
|
rct_inh_constr = And(
|
|
rct_inh_constr, self.v_param[rct_param_name][entity] < conc
|
|
)
|
|
|
|
return rct_inh_constr
|
|
|
|
def enc_single_reaction(self, level, path_idx, reaction):
|
|
"""
|
|
Encodes a single reaction
|
|
|
|
For encoding the products we use intermediate variables:
|
|
|
|
* each reaction has its own product variables,
|
|
|
|
* those are meant to be used to compute the MAX concentration
|
|
|
|
"""
|
|
|
|
reactants, inhibitors, products = reaction
|
|
|
|
# we need reaction_id to find the intermediate product variable
|
|
reaction_id = self.rs.reactions.index(reaction)
|
|
|
|
# ** REACTANTS *******************************************
|
|
enc_reactants = True
|
|
if is_param(reactants):
|
|
param_name = reactants.name
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
enc_reactants = And(
|
|
enc_reactants,
|
|
Or(
|
|
self.v_param[param_name][entity] == 0,
|
|
self.path_v[path_idx][level][entity]
|
|
>= self.v_param[param_name][entity],
|
|
self.path_v_ctx[path_idx][level][entity]
|
|
>= self.v_param[param_name][entity],
|
|
),
|
|
)
|
|
else:
|
|
for entity, conc in reactants:
|
|
enc_reactants = And(
|
|
enc_reactants,
|
|
Or(
|
|
self.path_v[path_idx][level][entity] >= conc,
|
|
self.path_v_ctx[path_idx][level][entity] >= conc,
|
|
),
|
|
)
|
|
|
|
# ** INHIBITORS ******************************************
|
|
enc_inhibitors = True
|
|
if is_param(inhibitors):
|
|
param_name = inhibitors.name
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
enc_inhibitors = And(
|
|
enc_inhibitors,
|
|
Or(
|
|
self.v_param[param_name][entity] == 0,
|
|
And(
|
|
self.path_v[path_idx][level][entity]
|
|
< self.v_param[param_name][entity],
|
|
self.path_v_ctx[path_idx][level][entity]
|
|
< self.v_param[param_name][entity],
|
|
),
|
|
),
|
|
)
|
|
else:
|
|
for entity, conc in inhibitors:
|
|
enc_inhibitors = And(
|
|
enc_inhibitors,
|
|
And(
|
|
self.path_v[path_idx][level][entity] < conc,
|
|
self.path_v_ctx[path_idx][level][entity] < conc,
|
|
),
|
|
)
|
|
|
|
# ** PRODUCTS *******************************************
|
|
enc_products = True
|
|
if is_param(products):
|
|
param_name = products.name
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
enc_products = simplify(
|
|
And(
|
|
enc_products,
|
|
self.path_v_improd[path_idx][level + 1][reaction_id][entity]
|
|
== self.v_param[param_name][entity],
|
|
)
|
|
)
|
|
else:
|
|
for entity, conc in products:
|
|
enc_products = simplify(
|
|
And(
|
|
enc_products,
|
|
self.path_v_improd[path_idx][level + 1][reaction_id][entity]
|
|
== conc,
|
|
)
|
|
)
|
|
|
|
# Nothing is produced (when the reaction is disabled)
|
|
enc_no_prod = True
|
|
if is_param(products):
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
enc_no_prod = And(
|
|
enc_no_prod,
|
|
self.path_v_improd[path_idx][level + 1][reaction_id][entity] == 0,
|
|
)
|
|
else:
|
|
for entity, _ in products:
|
|
enc_no_prod = simplify(
|
|
And(
|
|
enc_no_prod,
|
|
self.path_v_improd[path_idx][level + 1][reaction_id][entity]
|
|
== 0,
|
|
)
|
|
)
|
|
|
|
#
|
|
# (R and I) iff P
|
|
#
|
|
enc_enabled = And(enc_reactants, enc_inhibitors) == enc_products
|
|
|
|
#
|
|
# ~(R and I) iff P_zero
|
|
#
|
|
enc_not_enabled = Not(And(enc_reactants, enc_inhibitors)) == enc_no_prod
|
|
|
|
enc_reaction = And(enc_enabled, enc_not_enabled)
|
|
|
|
return enc_reaction
|
|
|
|
def enc_general_reaction_enabledness(self, level, path_idx):
|
|
"""
|
|
General enabledness condition for reactions
|
|
|
|
The necessary condition for a reaction to be enabled is
|
|
that the state is not empty, i.e., at least one entity
|
|
is present in the current state.
|
|
|
|
This condition must be used when there are parametric reactions
|
|
because parameters could have all the entities set to zero and
|
|
that immediately allows for all the conditions on the reactants
|
|
to be fulfilled: (entity <= param) -> (0 <= 0)
|
|
"""
|
|
|
|
enc_cond = False
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
enc_cond = simplify(
|
|
Or(
|
|
enc_cond,
|
|
self.path_v[path_idx][level][entity] > 0,
|
|
self.path_v_ctx[path_idx][level][entity] > 0,
|
|
)
|
|
)
|
|
|
|
return enc_cond
|
|
|
|
def enc_rs_trans(self, level, path_idx):
|
|
"""Encodes the transition relation"""
|
|
|
|
#
|
|
# IMPORTANT NOTE
|
|
#
|
|
# We need to make sure we do something about the UNUSED ENTITIES
|
|
# that is, those that are never produced.
|
|
#
|
|
# They should have concentration levels set to 0.
|
|
#
|
|
# That needs to happen automatically (in the MAX encoding) -- for the parametric
|
|
# case it makes no sense to identify the entities that are never produced, unless
|
|
# we have no parameters as products (special case, so that could be an
|
|
# optimisation)
|
|
#
|
|
|
|
enc_trans = True
|
|
|
|
for reaction in self.rs.reactions:
|
|
enc_reaction = self.enc_single_reaction(level, path_idx, reaction)
|
|
enc_trans = simplify(And(enc_trans, enc_reaction))
|
|
|
|
# Next we encode the MAX concentration values:
|
|
# we collect those from the intermediate product variables
|
|
|
|
enc_max_prod = True
|
|
|
|
# Save all the intermediate product variables for a given level:
|
|
#
|
|
# - Intermediate products of (level+1) correspond to the next level
|
|
#
|
|
# {reactants & inhibitors}[level]
|
|
# =>
|
|
# {improd}[level+1]
|
|
# =>
|
|
# {products}[level+1]
|
|
#
|
|
current_v_improd_for_entities = self.path_v_improd_for_entities[path_idx][
|
|
level + 1
|
|
]
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
per_reaction_vars = current_v_improd_for_entities.get(entity, [])
|
|
enc_max_prod = simplify(
|
|
And(
|
|
enc_max_prod,
|
|
self.path_v[path_idx][level + 1][entity]
|
|
== self.enc_max(per_reaction_vars),
|
|
)
|
|
)
|
|
|
|
# make sure at least one entity is >0
|
|
enc_general_cond = self.enc_general_reaction_enabledness(level, path_idx)
|
|
|
|
enc_trans_with_max = simplify(And(enc_general_cond, enc_max_prod, enc_trans))
|
|
|
|
# print(enc_trans_with_max)
|
|
|
|
return enc_trans_with_max
|
|
|
|
def enc_max(self, elements):
|
|
enc = None
|
|
|
|
if elements == []:
|
|
enc = 0
|
|
|
|
elif len(elements) == 1:
|
|
enc = z3_max(0, elements[0])
|
|
|
|
elif len(elements) > 1:
|
|
enc = 0
|
|
for i in range(len(elements) - 1):
|
|
enc = z3_max(enc, z3_max(elements[i], elements[i + 1]))
|
|
|
|
return enc
|
|
|
|
def enc_automaton_trans(self, level, path_idx):
|
|
"""Encodes the transition relation for the context automaton"""
|
|
|
|
enc_trans = False
|
|
|
|
for src, ctx, dst in self.ca.transitions:
|
|
src_enc = self.path_ca_state[path_idx][level] == src
|
|
dst_enc = self.path_ca_state[path_idx][level + 1] == dst
|
|
|
|
all_ent = set(range(len(self.rs.background_set)))
|
|
|
|
incl_ctx = set([e for e, c in ctx])
|
|
excl_ctx = all_ent - incl_ctx
|
|
|
|
ctx_enc = True
|
|
|
|
for e, c in ctx:
|
|
ctx_enc = simplify(
|
|
And(ctx_enc, self.path_v_ctx[path_idx][level][e] == c)
|
|
)
|
|
|
|
for e in excl_ctx:
|
|
ctx_enc = simplify(
|
|
And(ctx_enc, self.path_v_ctx[path_idx][level][e] == 0)
|
|
)
|
|
|
|
cur_trans = simplify(And(src_enc, ctx_enc, dst_enc))
|
|
enc_trans = simplify(Or(enc_trans, cur_trans))
|
|
|
|
return enc_trans
|
|
|
|
def enc_exact_state(self, level, state):
|
|
"""Encodes the state at the given level with the exact concentration values"""
|
|
|
|
raise RuntimeError("Should not be used with RSC")
|
|
|
|
def enc_min_state(self, level, state):
|
|
"""Encodes the state at the given level with the minimal required concentration levels"""
|
|
|
|
enc = True
|
|
for ent, conc in state:
|
|
e_id = self.rs.get_entity_id(ent)
|
|
enc = And(enc, self.v[level][e_id] >= conc)
|
|
|
|
return simplify(enc)
|
|
|
|
def enc_state_with_blocking(self, level, prop):
|
|
"""Encodes the state at the given level with blocking certain concentrations"""
|
|
|
|
required, blocked = prop
|
|
|
|
enc = True
|
|
for ent, conc in required:
|
|
e_id = self.rs.get_entity_id(ent)
|
|
enc = And(enc, self.v[level][e_id] >= conc)
|
|
|
|
for ent, conc in blocked:
|
|
e_id = self.rs.get_entity_id(ent)
|
|
enc = And(enc, self.v[level][e_id] < conc)
|
|
|
|
return simplify(enc)
|
|
|
|
def decode_witness(self, max_level, path_idx, print_model=False):
|
|
"""
|
|
Decodes the witness
|
|
|
|
Also decodes the parameters
|
|
"""
|
|
|
|
m = self.solver.model()
|
|
|
|
if print_model:
|
|
print(m)
|
|
|
|
for level in range(max_level + 1):
|
|
print("\n{: >70}".format("[ level=" + repr(level) + " ]"))
|
|
|
|
print(" State: {", end=""),
|
|
for var_id in range(len(self.path_v[path_idx][level])):
|
|
var_rep = repr(m[self.path_v[path_idx][level][var_id]])
|
|
if not var_rep.isdigit():
|
|
raise RuntimeError(
|
|
"unexpected: representation is not a positive integer"
|
|
)
|
|
if int(var_rep) > 0:
|
|
print(" " + self.rs.get_entity_name(var_id) + "=" + var_rep, end="")
|
|
print(" }")
|
|
|
|
if level != max_level:
|
|
print(" Context set: ", end="")
|
|
print("{", end="")
|
|
for var_id in range(len(self.path_v[path_idx][level])):
|
|
var_rep = repr(m[self.path_v_ctx[path_idx][level][var_id]])
|
|
if not var_rep.isdigit():
|
|
raise RuntimeError(
|
|
"unexpected: representation is not a positive integer"
|
|
)
|
|
if int(var_rep) > 0:
|
|
print(
|
|
" " + self.rs.get_entity_name(var_id) + "=" + var_rep,
|
|
end="",
|
|
)
|
|
print(" }")
|
|
|
|
print()
|
|
|
|
def get_enc_formulae(self, encoder, formulae_list):
|
|
enc_form = []
|
|
for formula in formulae_list:
|
|
path_idx = formulae_list.index(formula)
|
|
|
|
print_info(
|
|
"Generating the encoding for {:s} ({:d} of {:d})".format(
|
|
str(formula), path_idx + 1, len(formulae_list)
|
|
)
|
|
)
|
|
|
|
encoder.load_variables(
|
|
var_rs=self.path_v[path_idx],
|
|
var_ctx=self.path_v_ctx[path_idx],
|
|
var_loop_pos=self.path_loop_position[path_idx],
|
|
)
|
|
|
|
enc_form.append(encoder.get_encoding(formula, self.current_level))
|
|
ncalls = encoder.get_ncalls()
|
|
|
|
print_info(
|
|
"Cache hits: {:d}, encode calls: {:d} (approx: {:d})".format(
|
|
encoder.get_cache_hits(), ncalls[0], ncalls[1]
|
|
)
|
|
)
|
|
|
|
encoder.flush_cache()
|
|
|
|
return enc_form
|
|
|
|
def print_witness(self, formulae_list):
|
|
for formula in formulae_list:
|
|
path_idx = formulae_list.index(formula)
|
|
print("\n{:=^70}".format("[ WITNESS ]"))
|
|
print("\n Witness for: {:s}".format(str(formula)))
|
|
self.decode_witness(self.current_level, path_idx)
|
|
|
|
self.print_parameter_valuations()
|
|
|
|
def print_parameter_valuations(self):
|
|
m = self.solver.model()
|
|
|
|
print("\n Parameters:\n")
|
|
for param_name in sorted(self.rs.parameters.keys()):
|
|
print("{: >6}: ".format(param_name), end="")
|
|
print("{", end="")
|
|
|
|
params = self.v_param[param_name]
|
|
|
|
for entity in self.rs.set_of_bgset_ids:
|
|
var_rep = repr(m[params[entity]])
|
|
if not var_rep.isdigit():
|
|
raise RuntimeError(
|
|
"unexpected: representation is not a positive integer"
|
|
)
|
|
if int(var_rep) > 0:
|
|
print(
|
|
" " + str(self.rs.get_entity_name(entity)) + "=" + str(var_rep),
|
|
end="",
|
|
)
|
|
print(" }")
|
|
|
|
print()
|
|
|
|
def enc_concentration_levels_assertions_for_paths(self, level, num_of_paths):
|
|
additional_assertions = []
|
|
for path_idx in range(num_of_paths):
|
|
additional_assertions.append(
|
|
self.enc_concentration_levels_assertion(level, path_idx)
|
|
)
|
|
additional_assertions.append(self.enc_param_concentration_levels_assertion())
|
|
|
|
return additional_assertions
|
|
|
|
def enc_transition_relation_for_paths(self, level, num_of_paths):
|
|
enc_trans = []
|
|
for path_idx in range(num_of_paths):
|
|
enc_trans.append(self.enc_transition_relation(level, path_idx))
|
|
return enc_trans
|
|
|
|
def print_level(self):
|
|
print("{:->70}".format("[ level=" + str(self.current_level) + " done ]"))
|
|
|
|
def check_rsltl(
|
|
self,
|
|
formulae_list,
|
|
print_witness=True,
|
|
print_time=True,
|
|
print_mem=True,
|
|
max_level=None,
|
|
cont_if_sat=False,
|
|
param_constr=None,
|
|
):
|
|
"""
|
|
Bounded Model Checking for rsLTL properties
|
|
|
|
* print_witness -- prints the decoded witness
|
|
* print_time -- prints the time consumed
|
|
* print_mem -- prints the memory consumed
|
|
* max_level -- if not None, the methods
|
|
stops at the specified level
|
|
* cont_if_sat -- if True, then the method
|
|
continues up until max_level is
|
|
reached (even if sat found)
|
|
* param_constr -- constraints on parameters
|
|
"""
|
|
|
|
if not isinstance(formulae_list, (list, tuple)):
|
|
print_error("Expected a list of formulae")
|
|
|
|
self.reset()
|
|
|
|
num_of_paths = len(formulae_list)
|
|
|
|
print_info("Running rsLTL bounded model checking")
|
|
print_info("Tested formulae:")
|
|
for form in formulae_list:
|
|
print_info(" " * 4 + str(form))
|
|
|
|
print_info("INITIALISING...")
|
|
|
|
if print_time:
|
|
start_time = resource.getrusage(resource.RUSAGE_SELF).ru_utime
|
|
|
|
self.prepare_all_variables(num_of_paths)
|
|
|
|
self.load_varset_for_path(0)
|
|
|
|
# initial states for all the paths
|
|
initial_states = []
|
|
for path_idx in range(num_of_paths):
|
|
initial_states.append(self.enc_init_state(0, path_idx))
|
|
self.solver_add(initial_states)
|
|
|
|
self.current_level = 0
|
|
|
|
# assertions for all the paths and parameters
|
|
self.solver_add(
|
|
self.enc_concentration_levels_assertions_for_paths(0, num_of_paths)
|
|
)
|
|
self.solver_add(self.enc_param_concentration_levels_assertion())
|
|
self.solver_add(self.enc_param_sanity_for_reactions())
|
|
|
|
encoder = rsLTL_Encoder(self)
|
|
|
|
if param_constr:
|
|
param_contr_encoder = ParamConstr_Encoder(self)
|
|
enc_param_constr = param_contr_encoder.encode(param_constr)
|
|
self.solver_add(enc_param_constr)
|
|
|
|
if self.optimise:
|
|
self.assert_param_optimisation()
|
|
|
|
print_info("STARTING TO ITERATE...")
|
|
|
|
while True:
|
|
print(
|
|
"\n{:-^70}".format(
|
|
"[ Working at level=" + str(self.current_level) + " ]"
|
|
)
|
|
)
|
|
|
|
# reachability test:
|
|
self.solver.push()
|
|
|
|
# Encoding all the formulae
|
|
enc_form = self.get_enc_formulae(encoder, formulae_list)
|
|
print_info("Adding the formulae to the solver...")
|
|
self.solver_add(enc_form)
|
|
|
|
# Loops encoding
|
|
print_info("Adding the encoding for the loops...")
|
|
self.solver_add(self.get_loop_encodings())
|
|
|
|
print_info("Testing satisfiability...")
|
|
result = self.solver.check()
|
|
if result == sat:
|
|
print_positive(
|
|
green_str("SAT at level={:d}".format(self.current_level))
|
|
)
|
|
# print(self.solver.model())
|
|
if print_witness:
|
|
self.print_witness(formulae_list)
|
|
if not cont_if_sat:
|
|
break
|
|
else:
|
|
print_info("UNSAT")
|
|
self.solver.pop()
|
|
|
|
if not max_level is None and self.current_level > max_level:
|
|
print_info("As requested, stopping at level=" + str(max_level))
|
|
break
|
|
|
|
self.prepare_all_variables(num_of_paths)
|
|
|
|
# assertions for all the paths
|
|
self.solver_add(
|
|
self.enc_concentration_levels_assertions_for_paths(
|
|
self.current_level + 1, num_of_paths
|
|
)
|
|
)
|
|
|
|
print_info("Unrolling the transition relation")
|
|
self.solver_add(
|
|
self.enc_transition_relation_for_paths(self.current_level, num_of_paths)
|
|
)
|
|
|
|
self.print_level()
|
|
|
|
self.current_level += 1
|
|
|
|
if print_time:
|
|
self.print_time(start_time)
|
|
if print_mem:
|
|
self.print_mem()
|
|
|
|
def print_time(self, start):
|
|
stop = resource.getrusage(resource.RUSAGE_SELF).ru_utime
|
|
self.verification_time = stop - start
|
|
print()
|
|
print_info("{: >60}".format(" Time: " + repr(self.verification_time) + " s"))
|
|
|
|
def print_mem(self):
|
|
print_info(
|
|
"{: >60}".format(
|
|
" Memory: "
|
|
+ repr(
|
|
resource.getrusage(resource.RUSAGE_SELF).ru_maxrss / (1024 * 1024)
|
|
)
|
|
+ " MB"
|
|
)
|
|
)
|
|
|
|
def dummy_unroll(self, levels):
|
|
"""Unrolls the variables for testing purposes"""
|
|
|
|
self.current_level = -1
|
|
for i in range(levels + 1):
|
|
self.prepare_all_variables()
|
|
self.current_level += 1
|
|
|
|
print(C_MARK_INFO + " Dummy Unrolling done.")
|
|
|
|
def state_equality(self, level_A, level_B):
|
|
"""Encodes equality of two states at two different levels"""
|
|
|
|
eq_enc = True
|
|
|
|
for e_i in range(len(self.rs.background_set)):
|
|
e_i_equality = self.v[level_A][e_i] == self.v[level_B][e_i]
|
|
eq_enc = simplify(And(eq_enc, e_i_equality))
|
|
|
|
eq_enc_ctxaut = self.ca_state[level_A] == self.ca_state[level_B]
|
|
eq_enc = simplify(And(eq_enc, eq_enc_ctxaut))
|
|
|
|
return eq_enc
|
|
|
|
def get_loop_encodings(self):
|
|
k = self.current_level
|
|
loop_var = self.loop_position
|
|
|
|
loop_enc = True
|
|
|
|
"""
|
|
(loop_var == i) means that there is a loop taking back to the state (i-1)
|
|
|
|
Therefore, the encoding starts at 1, not at 0.
|
|
"""
|
|
|
|
for i in range(1, k + 1):
|
|
loop_enc = simplify(
|
|
And(loop_enc, Implies(loop_var == i, self.state_equality(i - 1, k)))
|
|
)
|
|
|
|
return loop_enc
|
|
|
|
def solver_add(self, expression):
|
|
"""
|
|
This is a solver.add() wrapper
|
|
"""
|
|
|
|
if expression == True:
|
|
return
|
|
|
|
if expression == False:
|
|
raise RuntimeError("Trying to assert False.")
|
|
|
|
self.solver.add(expression)
|
|
|
|
def check_reachability(
|
|
self, state, print_witness=True, print_time=True, print_mem=True, max_level=1000
|
|
):
|
|
"""Main testing function"""
|
|
|
|
self.reset()
|
|
|
|
if print_time:
|
|
# start = time()
|
|
start = resource.getrusage(resource.RUSAGE_SELF).ru_utime
|
|
|
|
self.prepare_all_variables()
|
|
self.solver_add(self.enc_init_state(0))
|
|
self.current_level = 0
|
|
|
|
self.prepare_all_variables()
|
|
|
|
self.solver_add(self.enc_concentration_levels_assertion(0))
|
|
|
|
while True:
|
|
self.prepare_all_variables()
|
|
self.solver_add(
|
|
self.enc_concentration_levels_assertion(self.current_level + 1)
|
|
)
|
|
|
|
print(
|
|
"\n{:-^70}".format(
|
|
"[ Working at level=" + str(self.current_level) + " ]"
|
|
)
|
|
)
|
|
stdout.flush()
|
|
|
|
# reachability test:
|
|
print("[" + colour_str(C_BOLD, "i") + "] Adding the reachability test...")
|
|
self.solver.push()
|
|
|
|
self.solver_add(self.enc_state_with_blocking(self.current_level, state))
|
|
|
|
result = self.solver.check()
|
|
if result == sat:
|
|
print(
|
|
"["
|
|
+ colour_str(C_BOLD, "+")
|
|
+ "] "
|
|
+ colour_str(C_GREEN, "SAT at level=" + str(self.current_level))
|
|
)
|
|
if print_witness:
|
|
print("\n{:=^70}".format("[ WITNESS ]"))
|
|
self.decode_witness(self.current_level)
|
|
break
|
|
else:
|
|
self.solver.pop()
|
|
|
|
print("[" + colour_str(C_BOLD, "i") + "] Unrolling the transition relation")
|
|
self.solver_add(self.enc_transition_relation(self.current_level))
|
|
|
|
print("{:->70}".format("[ level=" + str(self.current_level) + " done ]"))
|
|
self.current_level += 1
|
|
|
|
if self.current_level > max_level:
|
|
print("Stopping at level=" + str(max_level))
|
|
break
|
|
|
|
if print_time:
|
|
# stop = time()
|
|
stop = resource.getrusage(resource.RUSAGE_SELF).ru_utime
|
|
self.verification_time = stop - start
|
|
print()
|
|
print(
|
|
"\n[i] {: >60}".format(" Time: " + repr(self.verification_time) + " s")
|
|
)
|
|
|
|
if print_mem:
|
|
print(
|
|
"[i] {: >60}".format(
|
|
" Memory: "
|
|
+ repr(
|
|
resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
|
|
/ (1024 * 1024)
|
|
)
|
|
+ " MB"
|
|
)
|
|
)
|
|
|
|
def get_verification_time(self):
|
|
return self.verification_time
|
|
|
|
def show_encoding(
|
|
self,
|
|
state,
|
|
print_witness=True,
|
|
print_time=False,
|
|
print_mem=False,
|
|
max_level=100,
|
|
):
|
|
"""Encoding debug function"""
|
|
|
|
self.reset()
|
|
|
|
self.prepare_all_variables()
|
|
init_s = self.enc_init_state(0)
|
|
print(init_s)
|
|
self.solver_add(init_s)
|
|
self.current_level = 0
|
|
|
|
self.prepare_all_variables()
|
|
|
|
while True:
|
|
self.prepare_all_variables()
|
|
|
|
print("-----[ Working at level=" + str(self.current_level) + " ]-----")
|
|
stdout.flush()
|
|
|
|
# reachability test:
|
|
print("[i] Adding the reachability test...")
|
|
self.solver.push()
|
|
|
|
s = self.enc_min_state(self.current_level, state)
|
|
print("Test: ", s)
|
|
|
|
self.solver_add(s)
|
|
|
|
result = self.solver.check()
|
|
if result == sat:
|
|
print(
|
|
"\n[+] "
|
|
+ colour_str(C_RED, "SAT at level=" + str(self.current_level))
|
|
)
|
|
if print_witness:
|
|
self.decode_witness(self.current_level)
|
|
break
|
|
else:
|
|
self.solver.pop()
|
|
|
|
print("[i] Unrolling the transition relation")
|
|
t = self.enc_transition_relation(self.current_level)
|
|
print(t)
|
|
self.solver_add(t)
|
|
|
|
print("-----[ level=" + str(self.current_level) + " done ]")
|
|
self.current_level += 1
|
|
|
|
if self.current_level > max_level:
|
|
print("Stopping at level=" + str(max_level))
|
|
break
|
|
else:
|
|
x = input("Next level? ")
|
|
x = x.lower()
|
|
if not (x == "y" or x == "yes"):
|
|
break
|
|
|
|
|
|
# EOF
|