def go_bmc(force=False):
"""
Performs all the necessary steps to use loaded model and be able to perform
bmc related operations on it.
:raises NuSMVNoReadModelError: if no module was read (:see:`glob.load`)
before this method was called.
"""
cmp = glob.global_compile_cmps()
if not _compile.cmp_struct_get_read_model(cmp):
raise NuSMVNoReadModelError("No read model.")
# Check cmps and perform what is needed
if not _compile.cmp_struct_get_flatten_hrc(cmp):
glob.flatten_hierarchy()
if not _compile.cmp_struct_get_encode_variables(cmp):
glob.encode_variables()
if not _compile.cmp_struct_get_build_bool_model(cmp):
glob.build_boolean_model(force=force)
if not _compile.cmp_struct_get_bmc_setup(cmp):
bmc_setup(force=force)
def mas(agents=None):
"""
Return (and compute if needed) the multi-agent system represented by
the currently read SMV model.
If agents is not None, the set of agents (and groups) of the MAS is
determined by agents.
Otherwise, every top-level module instantiation is considered an agent
where
- her actions are the inputs variables prefixed by her name;
- her observable variables are composed of
* the state variables of the system prefixed by her name;
* the state variables provided as an argument to the module
instantiation.
Note: if the MAS is already computed, agents argument has no effect.
agents -- a set of agents.
"""
global __mas
if __mas is None:
# Check cmps
if not nscompile.cmp_struct_get_read_model(nscompile.cvar.cmps):
raise NuSMVNoReadModelError("Cannot build MAS; no read file.")
if agents is None:
# Get agents names
tree = nsparser.cvar.parsed_tree
main = None
while tree is not None:
module = nsnode.car(tree)
if (nsnode.sprint_node(nsnode.car(nsnode.car(module))) ==
"main"):
main = module
tree = nsnode.cdr(tree)
if main is None:
print("[ERROR] No main module.")
return # TODO Error, cannot find main module
arguments = _get_instances_args_for_module(main)
# arguments is a dict instancename(str)->listofargs(node)
agents = arguments.keys()
# Compute the model
_compute_model()
st = symb_table()
# Flatten arguments and filter on variables
argvars = _flatten_and_filter_variable_args(arguments)
# Get agents observable variables (locals + module parameters)
localvars = _get_variables_by_instances(agents)
#localvars is a dict instancename(str)->listofvars(node)
inputvars = _get_input_vars_by_instances(agents)
# Merge instance variable arguments and local variables
variables = {key: ((key in argvars and argvars[key] or []) +
(key in localvars and localvars[key] or []))
for key in
list(argvars.keys())+list(localvars.keys())}
# Compute epistemic relation
singletrans = {}
for agent in variables:
transexpr = None
for var in variables[agent]:
var = nsnode.sprint_node(var)
transexpr = nsnode.find_node(nsparser.AND,
_get_epistemic_trans(var),
transexpr)
singletrans[agent] = transexpr
# Process variables to get strings instead of nodes
observedvars = {ag: {nsnode.sprint_node(v) for v in variables[ag]}
for ag in variables.keys()}
inputvars = {ag: {nsnode.sprint_node(v)
for v in inputvars[ag]}
for ag in inputvars.keys()}
groups = None
else:
_compute_model()
# observedvars: a dictionary of agent name -> set of observed vars
observedvars = {str(agent.name): {str(var)
for var in agent.observables}
for agent in agents}
# inputsvars: a dictionary of agent name -> set of inputs vars
inputvars = {str(agent.name): {str(ivar)
for ivar in agent.actions}
for agent in agents}
# groups:
# a dictionary of group name -> names of agents of the group
groups = {str(group.name): {str(agent.name)
for agent in group.agents}
for group in agents if isinstance(group, Group)}
# singletrans: a dictionary of agent name -> epistemic transition
singletrans = {}
for agent in agents:
name = str(agent.name)
transexpr = None
for var in observedvars[name]:
transexpr = nsnode.find_node(nsparser.AND,
_get_epistemic_trans(var),
transexpr)
singletrans[name] = transexpr
# Create the MAS
fsm = _prop_database().master.bddFsm
__mas = MAS(fsm._ptr, observedvars, inputvars, singletrans,
groups=groups, freeit=False)
return __mas
def mas(agents=None, initial_ordering=None):
"""
Return (and compute if needed) the multi-agent system represented by
the currently read SMV model.
If agents is not None, the set of agents (and groups) of the MAS is
determined by agents.
Otherwise, every top-level module instantiation is considered an agent
where
- her actions are the inputs variables prefixed by her name;
- her observable variables are composed of
* the state variables of the system prefixed by her name;
* the state variables provided as an argument to the module
instantiation.
If initial_ordering is not None, it must be the path to a variables
ordering file. It is used as the initial ordering for variables of the
model.
Note: if the MAS is already computed, agents and initial_ordering arguments
have no effect.
agents -- a set of agents.
"""
global __mas
if __mas is None:
# Check cmps
if not nscompile.cmp_struct_get_read_model(nscompile.cvar.cmps):
raise NuSMVNoReadModelError("Cannot build MAS; no read file.")
if agents is None:
# Get agents names
tree = nsparser.cvar.parsed_tree
main = None
while tree is not None:
module = nsnode.car(tree)
if (nsnode.sprint_node(nsnode.car(
nsnode.car(module))) == "main"):
main = module
tree = nsnode.cdr(tree)
if main is None:
print("[ERROR] No main module.")
return # TODO Error, cannot find main module
arguments = _get_instances_args_for_module(main)
# arguments is a dict instancename(str)->listofargs(node)
agents = arguments.keys()
# Compute the model
_compute_model(variables_ordering=initial_ordering)
st = symb_table()
# Flatten arguments and filter on variables
argvars = _flatten_and_filter_variable_args(arguments)
# Get agents observable variables (locals + module parameters)
localvars = _get_variables_by_instances(agents)
#localvars is a dict instancename(str)->listofvars(node)
inputvars = _get_input_vars_by_instances(agents)
# Merge instance variable arguments and local variables
variables = {
key: ((key in argvars and argvars[key] or []) +
(key in localvars and localvars[key] or []))
for key in list(argvars.keys()) + list(localvars.keys())
}
# Compute epistemic relation
singletrans = {}
for agent in variables:
transexpr = None
for var in variables[agent]:
var = nsnode.sprint_node(var)
transexpr = nsnode.find_node(nsparser.AND,
_get_epistemic_trans(var),
transexpr)
singletrans[agent] = transexpr
# Process variables to get strings instead of nodes
observedvars = {
ag: {nsnode.sprint_node(v)
for v in variables[ag]}
for ag in variables.keys()
}
inputvars = {
ag: {nsnode.sprint_node(v)
for v in inputvars[ag]}
for ag in inputvars.keys()
}
groups = None
else:
_compute_model(variables_ordering=initial_ordering)
# observedvars: a dictionary of agent name -> set of observed vars
observedvars = {
str(agent.name): [str(var) for var in agent.observables]
for agent in agents
}
# inputsvars: a dictionary of agent name -> set of inputs vars
inputvars = {
str(agent.name): [str(ivar) for ivar in agent.actions]
for agent in agents
}
# groups:
# a dictionary of group name -> names of agents of the group
groups = {
str(group.name): [str(agent.name) for agent in group.agents]
for group in agents if isinstance(group, Group)
}
# singletrans: a dictionary of agent name -> epistemic transition
singletrans = {}
for agent in agents:
name = str(agent.name)
transexpr = None
for var in observedvars[name]:
transexpr = nsnode.find_node(nsparser.AND,
_get_epistemic_trans(var),
transexpr)
singletrans[name] = transexpr
# Create the MAS
fsm = _prop_database().master.bddFsm
__mas = MAS(fsm._ptr,
observedvars,
inputvars,
singletrans,
groups=groups,
freeit=False)
return __mas
