def _build_gate_inftau(g, q10tempadjustmentName, neuroml_dt):
assert len(g.time_courses) == 1 and len(g.steady_states) == 1
eqns = []
state_name = g.name
term_name_inf = "%s_%s" % (state_name, "inf")
term_name_tau = "%s_%s" % (state_name, "tau")
def remap_gate_eqnI(s):
s = re.sub(r"""\bV\b""", "__VGate__", s)
s = re.sub(r"""\bv\b""", "__VGate__", s)
return s
tc = SeqUtils.expect_single(g.time_courses)
ss = SeqUtils.expect_single(g.steady_states)
tc_eqn = "%s = ( %s) * (%s)" % (term_name_tau, remap_gate_eqnI(tc.getEqn()), neuroml_dt)
ss_eqn = "%s = %s" % (term_name_inf, remap_gate_eqnI(ss.getEqn()))
state_eqn = "%s' = (%s-%s)/(%s) " % (state_name, term_name_inf, state_name, term_name_tau)
# Add the equations
eqns.extend([tc_eqn, ss_eqn, state_eqn])
# Add the steddy-state
eqns.append("<=> INITIAL %s:%s" % (state_name, term_name_inf))
return eqns
def _build_gate_inftau(g, q10tempadjustmentName, neuroml_dt):
assert len(g.time_courses) == 1 and len(g.steady_states) == 1
eqns = []
state_name = g.name
term_name_inf = "%s_%s" % (state_name, 'inf')
term_name_tau = "%s_%s" % (state_name, 'tau')
def remap_gate_eqnI(s):
s = re.sub(r"""\bV\b""", '__VGate__', s)
s = re.sub(r"""\bv\b""", '__VGate__', s)
return s
tc = SeqUtils.expect_single(g.time_courses)
ss = SeqUtils.expect_single(g.steady_states)
tc_eqn = '%s = ( %s) * (%s)' % (term_name_tau, remap_gate_eqnI(
tc.getEqn()), neuroml_dt)
ss_eqn = '%s = %s' % (term_name_inf, remap_gate_eqnI(ss.getEqn()))
state_eqn = "%s' = (%s-%s)/(%s) " % (state_name, term_name_inf, state_name,
term_name_tau)
# Add the equations
eqns.extend([
tc_eqn,
ss_eqn,
state_eqn,
])
# Add the steddy-state
eqns.append("<=> INITIAL %s:%s" % (state_name, term_name_inf))
return eqns
def parse_expr(text,
parse_type,
start_symbol=None,
debug=False,
backend=None,
working_dir=None,
options=None,
library_manager=None,
name=None):
#debug=True
# Are a parsing a complex expression? Then we need a library manager:
if library_manager is None and ParseTypes is not ParseTypes.L1_Unit:
library_manager = LibraryManager(backend=backend,
working_dir=working_dir,
options=options,
name=name,
src_text=text)
#First, let preprocess the text:
text = preprocess_string(text, parse_type=parse_type)
# Now, we can parse the expression using PLY:
try:
pRes, library_manager = parse_eqn_block(
text,
parse_type=parse_type,
debug=debug,
library_manager=library_manager)
except:
print
print 'Error Parsing: %s' % text
print 'Parsing as', parse_type
raise
# If its a level-3 expression, we need to evaluate it:
if parse_type == ParseTypes.L3_QuantityExpr:
from neurounits.writers.writer_ast_to_simulatable_object import FunctorGenerator
ev = FunctorGenerator().visit(pRes)
pRes = ev()
# And return the correct type of object:
ret = {
ParseTypes.L1_Unit:
lambda: pRes,
ParseTypes.L2_QuantitySimple:
lambda: pRes,
ParseTypes.L3_QuantityExpr:
lambda: pRes,
ParseTypes.L4_EqnSet:
lambda: SeqUtils.expect_single(library_manager.eqnsets),
ParseTypes.L5_Library:
lambda: SeqUtils.expect_single(library_manager.libraries),
ParseTypes.L6_TextBlock:
lambda: library_manager,
}
return ret[parse_type]()
def _build_gate_alphabetainftau(g, q10tempadjustmentName, neuroml_dt):
if len(g.openstates) != 1 or len(g.closedstates) != 1 or len(g.transitions) != 2:
raise NeuroUnitsImportNeuroMLNotImplementedException("Non-Standard Gate/Transtions setup")
state_name = g.name
alphaTr = SeqUtils.filter_expect_single(g.transitions, lambda s: s.name == "alpha")
betaTr = SeqUtils.filter_expect_single(g.transitions, lambda s: s.name == "beta")
term_name_alpha = "%s_%s" % (state_name, "alpha")
term_name_beta = "%s_%s" % (state_name, "beta")
term_name_inf = "%s_%s" % (state_name, "inf")
term_name_tau = "%s_%s" % (state_name, "tau")
def remap_gate_eqn(s):
# Remap Alpha/Beta terms into the appropriate Units
# when they are used in inf/tau equations:
alpha_repl = "((%s) * (%s) )" % (term_name_alpha, neuroml_dt)
beta_repl = "((%s) * (%s) )" % (term_name_beta, neuroml_dt)
s = s.replace("alpha", alpha_repl).replace("beta", beta_repl)
# Remap voltage terms to __VGATE__ (since they might be subject
# to offset later on:
s = re.sub(r"""\bV\b""", "__VGate__", s)
s = re.sub(r"""\bv\b""", "__VGate__", s)
return s
# Write the alpha/beta terms:
e1 = "%s = (%s) * (1/%s)" % (term_name_alpha, remap_gate_eqn(alphaTr.getEqn()), neuroml_dt)
e2 = "%s = (%s) * (1/%s)" % (term_name_beta, remap_gate_eqn(betaTr.getEqn()), neuroml_dt)
# Time courses should always be divided by rate_adjustment term!
if len(g.time_courses) != 0:
tc = SeqUtils.expect_single(g.time_courses)
tc_eqn = "%s = %s * (1/%s) *(%s)" % (
term_name_tau,
remap_gate_eqn(tc.getEqn()),
q10tempadjustmentName,
neuroml_dt,
)
else:
tc_eqn = "%s = 1/(%s* (%s+%s))" % (term_name_tau, q10tempadjustmentName, term_name_alpha, term_name_beta)
if len(g.steady_states) != 0:
ss = SeqUtils.expect_single(g.steady_states)
ss_eqn = "%s = %s" % (term_name_inf, remap_gate_eqn(ss.getEqn()))
else:
ss_eqn = "%s = %s/(%s+%s)" % (term_name_inf, term_name_alpha, term_name_alpha, term_name_beta)
# The state-equations
state_eqn = "%s' = (%s-%s)/(%s) " % (state_name, term_name_inf, state_name, term_name_tau)
# Set the initial value of the state-variable to be the same
# as the steady-state value:
initial_cond = "<=> INITIAL %s:%s" % (state_name, term_name_inf)
return [e1, e2, tc_eqn, ss_eqn, state_eqn, initial_cond]
def parse_expr(text, parse_type, start_symbol=None, debug=False, backend=None, working_dir=None, options=None,library_manager=None, name=None):
original_text = text
# Some initial preprocessing
# (This is a bit hacky)
if parse_type in [ParseTypes.L4_EqnSet, ParseTypes.L5_Library, ParseTypes.L6_TextBlock]:
text = "\n".join([ l.split("#")[0] for l in text.split("\n") ])
lines = []
for l in text.split('\n'):
if len(lines) != 0 and lines[-1].endswith('\\'):
assert l
lines[-1] = (lines[-1])[:-1] + l
else:
l = l.strip()
if not l:
continue
if not l[-1] in ('{', ';'):
l = l + ';'
lines.append(l)
text = '\n'.join(lines)
else:
text = text.strip()
if library_manager is None and ParseTypes is not ParseTypes.L1_Unit:
library_manager = LibraryManager(backend=backend, working_dir=working_dir, options=options, name=name, src_text=original_text )
pRes, library_manager = parse_eqn_block(text, parse_type=parse_type, debug=debug, library_manager=library_manager)
if parse_type==ParseTypes.L3_QuantityExpr:
from neurounits.writers.writer_ast_to_simulatable_object import FunctorGenerator
ev = FunctorGenerator().visit(pRes)
#functor = SeqUtils.expect_single( F.assignment_evaluators.values() )
#\ev = F.visit(pRes
pRes = ev()
ret = { ParseTypes.L1_Unit: lambda: pRes,
ParseTypes.L2_QuantitySimple: lambda: pRes,
ParseTypes.L3_QuantityExpr: lambda: pRes,
ParseTypes.L4_EqnSet: lambda: SeqUtils.expect_single(library_manager.eqnsets),
ParseTypes.L5_Library: lambda: SeqUtils.expect_single(library_manager.libraries),
ParseTypes.L6_TextBlock: lambda: library_manager,
}
return ret[parse_type]()
def _res_assignments(self, o, **kwargs):
removed = []
for aKey in o._eqn_assignment.keys():
a = o._eqn_assignment[aKey]
alhs = a.lhs
fixed_value = self.visit(a.rhs_map)
if fixed_value:
sym_suffix = '_as_symconst'
sym_suffix = ''
s = ast.SymbolicConstant(symbol=aKey.symbol
+ sym_suffix, value=fixed_value)
#ReplaceNode(a.lhs, s).visit(o)
ReplaceNode.replace_and_check(srcObj=a.lhs, dstObj=s, root = o)
o._symbolicconstants[aKey.symbol] = s
from neurounits.misc import SeqUtils
old_ass = SeqUtils.filter_expect_single( o._eqn_assignment, lambda o:o.symbol == aKey.symbol )
del o._eqn_assignment[ old_ass ] #o.get_terminal_obj(aKey.symbol) ]
#del o._eqn_assignment[ o.get_terminal_obj(aKey.symbol) ]
removed.append(alhs)
# Double check they have gone:
for a in removed:
nc = EqnsetVisitorNodeCollector()
nc.visit(o)
assert not a in nc.all()
def VisitRegimeDispatchMap(self,o,**kwargs):
rt_graph = o.get_rt_graph()
rhs_functors = dict([(regime, self.visit(rhs)) for (regime,rhs) in o.rhs_map.items()])
from neurounits.misc import SeqUtils
try:
default = SeqUtils.filter_expect_single(rhs_functors.keys(), lambda r:r.name == None)
assert not None in rhs_functors
rhs_functors[None] = rhs_functors[default]
except ValueError:
pass
def f3(state_data, **kw):
regime_states = state_data.rt_regimes
#print 'Getting regime_state for RT graph:', repr(rt_graph)
curr_state = regime_states[rt_graph]
if curr_state in rhs_functors:
rhs_functor = rhs_functors[curr_state]
else:
rhs_functor = rhs_functors[None]
return rhs_functor(state_data=state_data, **kw)
return f3
assert len(o.rhs_map) == 1
return self.visit(o.rhs_map.values()[0])
def get_library(self, libname):
# print 'Searching for library: ' % libname
lib = SeqUtils.expect_single([l for l in chain(self.libraries,
self._stdlib_cache.libraries) if l.name == libname])
return lib
def get_library(self, libname):
lib = SeqUtils.expect_single([
l for l in chain(self.libraries, self._stdlib_cache.libraries)
if l.name == libname
])
return lib
def get(self, name, include_stdlibs=True):
if LibraryManager._stdlib_cache_loading:
include_stdlibs = False
if include_stdlibs:
srcs = chain(self.eqnsets, self.libraries, self._stdlib_cache.libraries)
else:
srcs = chain(self.eqnsets, self.libraries)
return SeqUtils.expect_single([l for l in srcs if l.name == name])
def get(self, name, include_stdlibs=True):
if LibraryManager._stdlib_cache_loading:
include_stdlibs = False
if include_stdlibs:
srcs = chain(self.eqnsets, self.libraries,
self._stdlib_cache.libraries)
else:
srcs = chain(self.eqnsets, self.libraries)
return SeqUtils.expect_single([l for l in srcs if l.name == name])
def parse_expr(text, parse_type, start_symbol=None, debug=False, backend=None, working_dir=None, options=None,library_manager=None, name=None):
#debug=True
# Are a parsing a complex expression? Then we need a library manager:
if library_manager is None and ParseTypes is not ParseTypes.L1_Unit:
library_manager = LibraryManager(backend=backend, working_dir=working_dir, options=options, name=name, src_text=text )
#First, let preprocess the text:
text = preprocess_string(text, parse_type=parse_type)
# Now, we can parse the expression using PLY:
try:
pRes, library_manager = parse_eqn_block(text, parse_type=parse_type, debug=debug, library_manager=library_manager)
except:
print
print 'Error Parsing: %s' % text
print 'Parsing as', parse_type
raise
# If its a level-3 expression, we need to evaluate it:
if parse_type==ParseTypes.L3_QuantityExpr:
from neurounits.writers.writer_ast_to_simulatable_object import FunctorGenerator
ev = FunctorGenerator().visit(pRes)
pRes = ev()
# And return the correct type of object:
ret = { ParseTypes.L1_Unit: lambda: pRes,
ParseTypes.L2_QuantitySimple: lambda: pRes,
ParseTypes.L3_QuantityExpr: lambda: pRes,
ParseTypes.L4_EqnSet: lambda: SeqUtils.expect_single(library_manager.eqnsets),
ParseTypes.L5_Library: lambda: SeqUtils.expect_single(library_manager.libraries),
ParseTypes.L6_TextBlock: lambda: library_manager,
}
return ret[parse_type]()
def get_eqnset(self, libname):
eqnset = SeqUtils.expect_single(
[l for l in self.eqnsets if l.name == libname])
return eqnset
def get_eqnset(self, libname):
eqnset = SeqUtils.expect_single([l for l in self.eqnsets if l.name == libname])
return eqnset
def _build_gate_alphabetainftau(g, q10tempadjustmentName, neuroml_dt):
if len(g.openstates) != 1 or len(g.closedstates) != 1 or len(
g.transitions) != 2:
raise NeuroUnitsImportNeuroMLNotImplementedException(
'Non-Standard Gate/Transtions setup')
state_name = g.name
alphaTr = SeqUtils.filter_expect_single(g.transitions,
lambda s: s.name == "alpha")
betaTr = SeqUtils.filter_expect_single(g.transitions,
lambda s: s.name == "beta")
term_name_alpha = "%s_%s" % (state_name, 'alpha')
term_name_beta = "%s_%s" % (state_name, 'beta')
term_name_inf = "%s_%s" % (state_name, 'inf')
term_name_tau = "%s_%s" % (state_name, 'tau')
def remap_gate_eqn(s):
# Remap Alpha/Beta terms into the appropriate Units
# when they are used in inf/tau equations:
alpha_repl = "((%s) * (%s) )" % (term_name_alpha, neuroml_dt)
beta_repl = "((%s) * (%s) )" % (term_name_beta, neuroml_dt)
s = s.replace('alpha', alpha_repl).replace('beta', beta_repl)
# Remap voltage terms to __VGATE__ (since they might be subject
# to offset later on:
s = re.sub(r"""\bV\b""", '__VGate__', s)
s = re.sub(r"""\bv\b""", '__VGate__', s)
return s
# Write the alpha/beta terms:
e1 = '%s = (%s) * (1/%s)' % (term_name_alpha,
remap_gate_eqn(alphaTr.getEqn()), neuroml_dt)
e2 = '%s = (%s) * (1/%s)' % (term_name_beta,
remap_gate_eqn(betaTr.getEqn()), neuroml_dt)
# Time courses should always be divided by rate_adjustment term!
if len(g.time_courses) != 0:
tc = SeqUtils.expect_single(g.time_courses)
tc_eqn = '%s = %s * (1/%s) *(%s)' % (
term_name_tau, remap_gate_eqn(
tc.getEqn()), q10tempadjustmentName, neuroml_dt)
else:
tc_eqn = '%s = 1/(%s* (%s+%s))' % (term_name_tau,
q10tempadjustmentName,
term_name_alpha, term_name_beta)
if len(g.steady_states) != 0:
ss = SeqUtils.expect_single(g.steady_states)
ss_eqn = '%s = %s' % (term_name_inf, remap_gate_eqn(ss.getEqn()))
else:
ss_eqn = '%s = %s/(%s+%s)' % (term_name_inf, term_name_alpha,
term_name_alpha, term_name_beta)
# The state-equations
state_eqn = "%s' = (%s-%s)/(%s) " % (state_name, term_name_inf, state_name,
term_name_tau)
# Set the initial value of the state-variable to be the same
# as the steady-state value:
initial_cond = "<=> INITIAL %s:%s" % (state_name, term_name_inf)
return [e1, e2, tc_eqn, ss_eqn, state_eqn, initial_cond]
def EqnSet(cls, text, **kwargs):
library_manager = cls.File(text=text, **kwargs )
eqnset_name = SeqUtils.expect_single( library_manager.get_eqnset_names() )
return library_manager.get_eqnset(eqnset_name)
