def test_direct_memory_transfer():
h('''create soma''')
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(.5))
ic.delay = .5
ic.dur = 0.1
ic.amp = 0.3
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
v = h.Vector()
v.record(h.soma(.5)._ref_v, sec=h.soma)
i_mem = h.Vector()
i_mem.record(h.soma(.5)._ref_i_membrane_, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
vstd = v.cl()
tvstd = tv.cl()
i_memstd = i_mem.cl()
pc = h.ParallelContext()
h.stdinit()
pc.nrncore_run("-e %g" % h.tstop, 0)
assert (tv.eq(tvstd))
assert (v.cl().sub(vstd).abs().max() < 1e-10)
assert (i_mem.cl().sub(i_memstd).abs().max() < 1e-10)
def test_spikes():
ref_spikes = np.array([1.05])
h('''create soma''')
h.load_file("stdrun.hoc")
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(0.5))
ic.delay = 0.5
ic.dur = 0.1
ic.amp = 0.3
v = h.Vector()
v.record(h.soma(0.5)._ref_v, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
nc = h.NetCon(h.soma(0.5)._ref_v, None, sec=h.soma)
spikestime = h.Vector()
nc.record(spikestime)
h.run()
simulation_spikes = spikestime.to_python()
assert np.allclose(simulation_spikes, ref_spikes)
def corrected_experimental_protocol_with_linear_gpas_distr(
Ra=157.3621, gpas_soma=0.000403860792, k=0.001, cm=3., dt=0.1):
# -- Biophysics --
# Sec parameters and conductance
for sec in h.allsec():
sec.Ra = Ra # Ra is a parameter to infer
sec.cm = cm # parameter optimisation algorithm found this
sec.v = 0
sec.insert('pas')
sec.g_pas = gpas_soma # gpas is a parameter to infer
for seg in sec:
h('soma distance()')
dist = (h.distance(seg.x))
gpas = gpas_soma * (1 + k * dist)
if gpas < 0:
seg.g_pas = 0
print("WARNING!!! 'gpas' is in negative! Corrected to zero.")
else:
seg.g_pas = gpas
sec.e_pas = 0
# Print information
#h.psection()
# Stimulus
stim1 = h.IClamp(h.soma(0.01))
stim1.delay = 199.9
stim1.amp = 0.5
stim1.dur = 3.1
stim2 = h.IClamp(h.soma(0.01))
stim2.delay = 503
stim2.amp = 0.01
stim2.dur = 599.9
# Run simulation ->
# Set up recording Vectors
v_vec = h.Vector() # Membrane potential vector
t_vec = h.Vector() # Time stamp vector
v_vec.record(h.soma(0.5)._ref_v)
t_vec.record(h._ref_t)
# Simulation duration and RUN
h.tstop = (15000 - 1) * dt # Simulation end
h.dt = dt # Time step (iteration)
h.steps_per_ms = 1 / dt
h.v_init = 0
h.finitialize(h.v_init)
h.init()
h.run()
t = t_vec.to_python()
v = v_vec.to_python()
return t, v
def real_morphology_model_srsoma_rdend(stim,
gpas=0.0001,
Ra=100.,
cm=1.,
dt=0.1):
# -- Biophysics --
# Sec parameters and conductance
for sec in h.allsec():
sec.Ra = Ra # Ra is a parameter to infer
sec.cm = cm # parameter optimisation algorithm found this
sec.v = 0
sec.insert('pas')
sec.g_pas = gpas # gpas is a parameter to infer
sec.e_pas = 0
# Print information
# h.psection()
h.dt = dt # Time step (iteration)
h.steps_per_ms = 1 / dt
# Stimulus
h.tstop = len(stim) * dt
h.load_file("vplay.hoc")
vec = h.Vector(stim)
istim = h.IClamp(h.soma(0.5))
vec.play(istim._ref_amp, h.dt)
istim.delay = 0 # Just for Neuron
istim.dur = 1e9 # Just for Neuron
# Run simulation ->
# Set up recording Vectors
vs_vec = h.Vector() # Membrane potential vector
vd_vec = h.Vector()
t_vec = h.Vector() # Time stamp vector
vd_vec.record(h.apic[30](0.5)._ref_v)
vs_vec.record(h.soma(0.5)._ref_v)
t_vec.record(h._ref_t)
# Simulation duration and RUN
# h.tstop = 1200 # Simulation end
h.v_init = 0
h.finitialize(h.v_init)
h.init()
h.run()
t = t_vec.to_python()
v_soma = vs_vec.to_python()
v_dend = vd_vec.to_python()
v_soma = np.array(v_soma)
v_dend = np.array(v_dend)
v = np.concatenate((v_soma, v_dend))
return t, v
def run_cclamp(self):
print "- running model", self.name
rec_t = h.Vector()
rec_t.record(h._ref_t)
rec_v = h.Vector()
rec_v.record(h.soma(0.5)._ref_v)
h.stdinit()
dt = 0.025
h.dt = dt
h.steps_per_ms = 1 / dt
h.v_init = -65
h.celsius = 34
h.init()
h.tstop = 1600
h.run()
t = numpy.array(rec_t)
v = numpy.array(rec_v)
return t, v
def run_syn(self):
# initiate recording
rec_t = h.Vector()
rec_t.record(h._ref_t)
rec_v = h.Vector()
rec_v.record(h.soma(0.5)._ref_v)
rec_v_dend = h.Vector()
rec_v_dend.record(h.dendrite[self.ndend](self.xloc)._ref_v)
print "- running model", self.name
# initialze and run
#h.load_file("stdrun.hoc")
h.stdinit()
dt = 0.025
h.dt = dt
h.steps_per_ms = 1 / dt
h.v_init = -65
h.celsius = 34
h.init()
h.tstop = 300
h.run()
# get recordings
t = numpy.array(rec_t)
v = numpy.array(rec_v)
v_dend = numpy.array(rec_v_dend)
return t, v, v_dend
def get_recording_vecs(st):
# record voltage in every ais segment (between AP init to the soma)
vecs_ais = []
for sec in cell.h.allsec():
if sec.name() == 'axon[0]': # axon[0] is first section of the axon
no_segs = sec.nseg
if no_segs > 1:
centers = np.linspace(0, 1., no_segs-1)
centers += centers[1]/2.
else:
centers = [0.5]
for seg_idx in range(0,no_segs-2):
# don't take two segments on the edges (neuron specs)
temp_axon_v= h.Vector()
temp_axon_v.record(eval('h.'+sec.name()+'('+str(centers[seg_idx])+')._ref_v'), sec = eval('h.'+sec.name()))
vecs_ais.append(temp_axon_v)
vec_soma = h.Vector()
vec_soma.record(h.soma(0.5)._ref_v, sec=h.soma) # vector for recording voltage at thes soma
vec_st = h.Vector()
vec_st.record(st._ref_i) # vector for recording current
return vecs_ais, vec_soma, vec_st
def exp_model2(Ra=157.3621, gpas=0.000403860792, cm=7.849480, dt=0.1):
# -- Biophysics --
# Sec parameters and conductance
for sec in h.allsec():
sec.Ra = Ra # Ra is a parameter to infer
sec.cm = cm # parameter optimisation algorithm found this
sec.v = 0
sec.insert('pas')
sec.g_pas = gpas # gpas is a parameter to infer
sec.e_pas = 0
# Print information
#h.psection()
# Stimulus
stim1 = h.IClamp(h.soma(0.01))
stim1.delay = 199.9
stim1.amp = 0.5
stim1.dur = 3.1
stim2 = h.IClamp(h.soma(0.01))
stim2.delay = 503
stim2.amp = 0.01
stim2.dur = 599.9
# Run simulation ->
# Set up recording Vectors
v_vec = h.Vector() # Membrane potential vector
t_vec = h.Vector() # Time stamp vector
v_vec.record(h.soma(0.5)._ref_v)
t_vec.record(h._ref_t)
# Simulation duration and RUN
h.tstop = (15000 - 1) * dt # Simulation end
h.dt = dt # Time step (iteration)
h.steps_per_ms = 1 / dt
h.v_init = 0
h.finitialize(h.v_init)
h.init()
h.run()
t = t_vec.to_python()
v = v_vec.to_python()
return t, v
def test_simple_sim():
h('''create soma''')
h.load_file("stdrun.hoc")
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(0.5))
ic.delay = 0.5
ic.dur = 0.1
ic.amp = 0.3
v = h.Vector()
v.record(h.soma(0.5)._ref_v, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
assert v[0] == -65.0
def run(mode):
pc.set_maxstep(10)
h.stdinit()
if mode == 0:
pc.psolve(h.tstop)
elif mode == 1:
while h.t < h.tstop:
pc.psolve(h.t + 1.0)
else:
while h.t < h.tstop:
h.continuerun(h.t + 0.5)
pc.psolve(h.t + 0.5)
tran = [h.t, h.soma(0.5).v, h.soma(0.5).hh.m]
assert tv.eq(tvstd)
assert (
v.cl().sub(vstd).abs().max() < 1e-10
) # usually v == vstd, some compilers might give slightly different results
assert i_mem.cl().sub(i_memstd).abs().max() < 1e-10
assert h.Vector(tran_std).sub(h.Vector(tran)).abs().max() < 1e-10
def test_simple_sim():
h("""create soma""")
h.load_file("stdrun.hoc")
h.soma.L = 5.6419
h.soma.diam = 5.6419
expect_hocerr(h.ion_register, ("na", 2))
assert h.ion_charge("na_ion") == 1.0
expect_hocerr(h.ion_register, ("ca", 3))
assert h.ion_charge("ca_ion") == 2.0
h.soma.insert("hh")
ic = h.IClamp(h.soma(0.5))
ic.delay = 0.5
ic.dur = 0.1
ic.amp = 0.3
v = h.Vector()
v.record(h.soma(0.5)._ref_v, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
assert v[0] == -65.0
def get_membrane_potential_soma(self, tstop):
# to supress hoc output from Jupyter notebook
save_stdout = sys.stdout
sys.stdout = open('/dev/stdout', 'w')
t_vec = h.Vector()
v_vec = h.Vector()
t_vec.record(h._ref_t)
v_vec.record(h.soma(0.5)._ref_v)
h.tstop = tstop
h.run()
sys.stdout = save_stdout #setting output back to before
return [t_vec.to_python(), v_vec.to_python()]
def test_direct_memory_transfer():
h('''create soma''')
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(.5))
ic.delay = .5
ic.dur = 0.1
ic.amp = 0.3
#for testing external mod file
h.soma.insert("Sample")
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
v = h.Vector()
v.record(h.soma(.5)._ref_v, sec=h.soma)
i_mem = h.Vector()
i_mem.record(h.soma(.5)._ref_i_membrane_, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
vstd = v.cl()
tvstd = tv.cl()
i_memstd = i_mem.cl()
# Save current (after run) value to compare with transfer back from coreneuron
tran_std = [h.t, h.soma(.5).v, h.soma(.5).hh.m]
from neuron import coreneuron
coreneuron.enable = True
coreneuron.gpu = bool(os.environ.get('CORENRN_ENABLE_GPU', ''))
pc = h.ParallelContext()
h.stdinit()
pc.psolve(h.tstop)
tran = [h.t, h.soma(.5).v, h.soma(.5).hh.m]
assert (tv.eq(tvstd))
assert (
v.cl().sub(vstd).abs().max() < 1e-10
) # usually v == vstd, some compilers might give slightly different results
# This check is disabled on GPU because fast imem is not implemented on
# GPU: https://github.com/BlueBrain/CoreNeuron/issues/197
assert (coreneuron.gpu or i_mem.cl().sub(i_memstd).abs().max() < 1e-10)
assert (h.Vector(tran_std).sub(h.Vector(tran)).abs().max() < 1e-10)
def test_direct_memory_transfer():
h('''create soma''')
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(.5))
ic.delay = .5
ic.dur = 0.1
ic.amp = 0.3
#for testing external mod file
h.soma.insert("Sample")
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
v = h.Vector()
v.record(h.soma(.5)._ref_v, sec=h.soma)
i_mem = h.Vector()
i_mem.record(h.soma(.5)._ref_i_membrane_, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
vstd = v.cl()
tvstd = tv.cl()
i_memstd = i_mem.cl()
# Save current (after run) value to compare with transfer back from coreneuron
tran_std = [h.t, h.soma(.5).v, h.soma(.5).hh.m]
from neuron import coreneuron
coreneuron.enable = True
pc = h.ParallelContext()
h.stdinit()
pc.psolve(h.tstop)
tran = [h.t, h.soma(.5).v, h.soma(.5).hh.m]
assert (v.eq(vstd))
assert (tv.eq(tvstd))
assert (v.cl().sub(vstd).abs().max() < 1e-10)
assert (i_mem.cl().sub(i_memstd).abs().max() < 1e-10)
assert (h.Vector(tran_std).sub(h.Vector(tran)).abs().max() < 1e-10)
def test_spikes(use_mpi4py=False):
if use_mpi4py:
from mpi4py import MPI
h('''create soma''')
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
h.soma.nseg = 3
ic = h.IClamp(h.soma(.25))
ic.delay = .1
ic.dur = 0.1
ic.amp = 0.3
ic2 = h.IClamp(h.soma(.75))
ic2.delay = 5.5
ic2.dur = 1
ic2.amp = 0.3
h.tstop = 10
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
pc = h.ParallelContext()
pc.set_gid2node(pc.id() + 1, pc.id())
myobj = h.NetCon(h.soma(0.5)._ref_v, None, sec=h.soma)
pc.cell(pc.id() + 1, myobj)
# NEURON spikes run
nrn_spike_t = h.Vector()
nrn_spike_gids = h.Vector()
pc.spike_record(-1, nrn_spike_t, nrn_spike_gids)
h.run()
nrn_spike_t = nrn_spike_t.to_python()
nrn_spike_gids = nrn_spike_gids.to_python()
# CORENEURON spike_record(-1) / spike_record(gidlist):
from neuron import coreneuron
coreneuron.enable = True
coreneuron.verbose = 0
h.stdinit()
corenrn_all_spike_t = h.Vector()
corenrn_all_spike_gids = h.Vector()
pc.spike_record(-1 if pc.id() == 0 else (pc.id()), corenrn_all_spike_t,
corenrn_all_spike_gids)
pc.psolve(h.tstop)
corenrn_all_spike_t = corenrn_all_spike_t.to_python()
corenrn_all_spike_gids = corenrn_all_spike_gids.to_python()
# check spikes match
assert (len(nrn_spike_t)) # check we've actually got spikes
assert (len(nrn_spike_t) == len(nrn_spike_gids))
# matching no. of gids
assert (nrn_spike_t == corenrn_all_spike_t)
assert (nrn_spike_gids == corenrn_all_spike_gids)
h.quit()
def test_spikes(use_mpi4py=False, use_nrnmpi_init=False, file_mode=False):
# mpi4py needs tp be imported before importing h
if use_mpi4py:
from mpi4py import MPI
from neuron import h, gui
# without mpi4py we need to call nrnmpi_init explicitly
elif use_nrnmpi_init:
from neuron import h, gui
h.nrnmpi_init()
# otherwise serial execution
else:
from neuron import h, gui
h('''create soma''')
h.soma.L=5.6419
h.soma.diam=5.6419
h.soma.insert("hh")
h.soma.nseg = 3
ic = h.IClamp(h.soma(.25))
ic.delay = .1
ic.dur = 0.1
ic.amp = 0.3
ic2 = h.IClamp(h.soma(.75))
ic2.delay = 5.5
ic2.dur = 1
ic2.amp = 0.3
h.tstop = 10
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
pc = h.ParallelContext()
pc.set_gid2node(pc.id()+1, pc.id())
myobj = h.NetCon(h.soma(0.5)._ref_v, None, sec=h.soma)
pc.cell(pc.id()+1, myobj)
# NEURON run
nrn_spike_t = h.Vector()
nrn_spike_gids = h.Vector()
# rank 0 record spikes for all gid while others
# for specific gid. this is for better test coverage.
pc.spike_record(-1 if pc.id() == 0 else (pc.id()+1), nrn_spike_t, nrn_spike_gids)
h.run()
nrn_spike_t = nrn_spike_t.to_python()
nrn_spike_gids = nrn_spike_gids.to_python()
# CORENEURON run
from neuron import coreneuron
coreneuron.enable = True
coreneuron.file_mode = file_mode
coreneuron.verbose = 0
h.stdinit()
corenrn_all_spike_t = h.Vector()
corenrn_all_spike_gids = h.Vector()
pc.spike_record(-1, corenrn_all_spike_t, corenrn_all_spike_gids )
pc.psolve(h.tstop)
corenrn_all_spike_t = corenrn_all_spike_t.to_python()
corenrn_all_spike_gids = corenrn_all_spike_gids.to_python()
# check spikes match
assert(len(nrn_spike_t)) # check we've actually got spikes
assert(len(nrn_spike_t) == len(nrn_spike_gids)); # matching no. of gids
assert(nrn_spike_t == corenrn_all_spike_t)
assert(nrn_spike_gids == corenrn_all_spike_gids)
h.quit()
for event in events:
I_vcs_events[-1].append(event)
IS_firing_times[-1].append(event)
I_vcs[-1].play(I_vcs_events[-1])
h("access {}".format(sec))
placeGABA(loc, 0.001)
h.pop_section()
IS_ind += 1
firing_times = np.zeros(shape=(int(number_of_E_synapses), int(h.tstop) + 100))
for ind in range(len(spine_firing_times)):
firing_times[ind, spine_firing_times[ind]] = 1
print('Running...')
soma_voltageVector = h.Vector()
soma_voltageVector.record(h.soma(0.5)._ref_v)
timeVector = h.Vector()
timeVector.record(h._ref_t)
E_densities = []
I_densities = []
h.stdinit()
h.run()
voltage_vectors = np.array(voltage_vectors)
timeVector = np.array(timeVector)
def measure_width(trace):
trace = np.array(trace)
points = np.where(trace > (trace[15000] + 2))[0]
points = points[np.logical_and(points >= 20000, points <= 30000)]
width = dt * (points[-1] - points[0])
return (width)
h.load_file('init_models_with_ca/init_model2.hoc')
default_sca = h.tuft.gbar_sca
# Set up stimulus
stim = h.Ipulse2(h.soma(.5))
stim.amp = 2
stim.dur = 3
stim.delay = 500
stim.per = 10
stim.num = 3
# Set up recording vectors
soma_v_vec = h.Vector()
soma_v_vec.record(h.soma(.5)._ref_v)
apical_end_v_vec = h.Vector()
apical_end_v_vec.record(h.apical(1)._ref_v)
tuft_v_vec = h.Vector()
tuft_v_vec.record(h.tuft(.5)._ref_v)
stim_vec = h.Vector()
stim_vec.record(stim._ref_i)
def pyloop(ratio1=1.1, ratio2=2.2, loc1=128, loc2=128, simiter=0, subth=False):
print '\n%s starting run' % (gethostname())
iotim = 0
tic = time()
if subth:
h5fn = 'paper-subth-%dum_%dum_run%d.h5' % (l1, l2, simiter)
h.tstop = 200
h.maxsyn2 = 20
else:
h5fn = 'paper-supth-%dum_%dum_run%d.h5' % (l1, l2, simiter)
h.tstop = 500
savd = './data/'
filters = pyt.Filters(complevel=9,
complib='zlib',
shuffle=True,
fletcher32=True)
h5f = pyt.openFile(os.path.join(savd, h5fn), 'w', filters=filters)
maxsyn1 = int(h.maxsyn1) + 1
maxsyn2 = int(h.maxsyn2) + 1
h.tsamp = h.tstop / h.dt + 1
tsamp = int(h.tsamp)
shape = (maxsyn1, maxsyn2, tsamp)
cshape = (1, 1, shape[-1])
r = h.Random(simiter + h.luckyoffset)
r.negexp(h.meanisi)
# stimulated branch
sec = h.a10_11
syns = [[], []]
for ind, loc in enumerate([loc1, loc2]):
if subth:
for delay, ratio in zip([2, 22], [ratio1, ratio2]):
syn = h.synsuper(.5, r, sec=sec)
syns[ind].append(syn)
syndict = dict(sloc=loc, ratio=ratio, e1del=delay, e1flag=1)
for name, value in syndict.iteritems():
setattr(syn.syn, name, value)
else:
syn = h.synsuper(.5, r, sec=sec)
syns[ind].append(syn)
syndict = dict(sloc=loc, ratio=ratio1, e1flag=0)
for name, value in syndict.iteritems():
setattr(syn.syn, name, value)
# initialize nseg with two 'active' branches
theseclist = [h.a10_11, h.a1_111]
sl2 = h.SectionList()
for sec in theseclist:
sl2.append(sec=sec)
poppedsecs = sl2.unique()
h.refreshnseg(h.makeactivelist(sl2))
print 'nseg: %d' % (h.nsegcnt())
h.cvode.cache_efficient(1)
h.cvode_active(0)
# somatic voltage recordings, initialize after nseg is set
v = h.Vector(tsamp)
v.record(h.soma(.5)._ref_v)
trash = v.label(h.soma.name())
# voltage recording dictionary
vd = {'s': v}
# dendritic voltage recording
h.distance(0, h.soma_con_pt, sec=h.soma)
for n, sec in enumerate(theseclist):
d = h.distance(0, sec=sec)
# location 1 synapses
locx1 = (loc1 - d) / sec.L
v = h.Vector(tsamp)
v.record(sec(locx1)._ref_v)
trash = v.label(sec.name())
vd.update({'dp' + str(n): v})
# location 2 synapses
locx2 = (loc2 - d) / sec.L
v = h.Vector(tsamp)
v.record(sec(locx2)._ref_v)
trash = v.label(sec.name())
vd.update({'dd' + str(n): v})
if subth:
h.poisson = 0
else:
h.poisson = 1
# 'background' guassian current injection
npy.random.seed(1)
ic = h.IClamp(0.5, sec=h.soma)
ic.dur = h.tstop
icmean = .75
icstd = 1
icvals = icmean + icstd * npy.random.randn(h.tstop / h.dt + 1)
icrand = h.Vector(tsamp)
for i in xrange(tsamp):
icrand.x[i] = icvals[i]
icrand.play(ic._ref_amp, h.dt)
h5d = {}
for key in vd.keys():
h5d.update({
key:
h5f.createCArray(h5f.root,
key,
shape=shape,
atom=pyt.Float64Atom(),
title=vd[key].label(),
chunkshape=cshape)
})
synpairs = [(i, j) for i in xrange(maxsyn1) for j in xrange(maxsyn2)]
if not subth:
# subsample stimulus configuration space to speed up simulations
oddpairs = [(i, j) for i in npy.mgrid[1:maxsyn1:2]
for j in npy.mgrid[1:maxsyn2:2]]
synpairs = list(set(synpairs) - set(oddpairs))
# subset of input configurations needed for figure 7 traces
realrunsdist = [3, 6, 9]
realrunsprox = [17, 21, 25]
iisi = simiter * (maxsyn1 + maxsyn2) # inter-iteration seed interval
for runcnt, (nsyn1, nsyn2) in enumerate(synpairs):
# configure s(t)imulation
if not subth:
if ((nsyn1 in realrunsprox and nsyn2 == 0)
or (nsyn2 in realrunsdist and nsyn1 == 0)):
h.fakerun = 0
else:
h.fakerun = 1
for ssyn in syns[0]:
ssyn.syn.nsyn = nsyn1
seed1 = float(iisi + nsyn1 + h.luckyoffset)
r1 = h.Random(seed1)
r1.negexp(h.meanisi)
ssyn.setrand(r1)
for ssyn in syns[1]:
ssyn.syn.nsyn = nsyn2
seed2 = float(iisi + maxsyn1 + nsyn2 + h.luckyoffset)
r2 = h.Random(seed2)
r2.negexp(h.meanisi)
ssyn.setrand(r2)
# run simulation
h.run()
# get results
iotic = time()
if not h.fakerun:
for key in h5d.keys():
h5d[key][nsyn1, nsyn2] = npy.array(vd[key]).reshape(cshape)
iotim += time() - iotic
iotic = time()
h5f.close()
iotim += time() - iotic
print '%s running %d runs took %d seconds, of which %d seconds was I/O' % (
gethostname(), runcnt + 1, time() - tic, iotim)
def test_direct_memory_transfer():
h("""create soma""")
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(0.5))
ic.delay = 0.5
ic.dur = 0.1
ic.amp = 0.3
# for testing external mod file
h.soma.insert("Sample")
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
v = h.Vector()
v.record(h.soma(0.5)._ref_v, sec=h.soma)
i_mem = h.Vector()
i_mem.record(h.soma(0.5)._ref_i_membrane_, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
vstd = v.cl()
tvstd = tv.cl()
i_memstd = i_mem.cl()
# Save current (after run) value to compare with transfer back from coreneuron
tran_std = [h.t, h.soma(0.5).v, h.soma(0.5).hh.m]
from neuron import coreneuron
coreneuron.enable = True
coreneuron.verbose = 0
coreneuron.gpu = enable_gpu
pc = h.ParallelContext()
def run(mode):
pc.set_maxstep(10)
h.stdinit()
if mode == 0:
pc.psolve(h.tstop)
elif mode == 1:
while h.t < h.tstop:
pc.psolve(h.t + 1.0)
else:
while h.t < h.tstop:
h.continuerun(h.t + 0.5)
pc.psolve(h.t + 0.5)
tran = [h.t, h.soma(0.5).v, h.soma(0.5).hh.m]
assert tv.eq(tvstd)
assert (
v.cl().sub(vstd).abs().max() < 1e-10
) # usually v == vstd, some compilers might give slightly different results
assert i_mem.cl().sub(i_memstd).abs().max() < 1e-10
assert h.Vector(tran_std).sub(h.Vector(tran)).abs().max() < 1e-10
for mode in [0, 1, 2]:
run(mode)
cnargs = coreneuron.nrncore_arg(h.tstop)
h.stdinit()
assert tv.size() == 1 and tvstd.size() != 1
pc.nrncore_run(cnargs, 1)
assert tv.eq(tvstd)
# print warning if HocEvent on event queue when CoreNEURON starts
def test_hoc_event():
print("in test_hoc_event() at t=%g" % h.t)
if h.t < 1.001:
h.CVode().event(h.t + 1.0, test_hoc_event)
fi = h.FInitializeHandler(2, test_hoc_event)
coreneuron.enable = False
run(0)
coreneuron.enable = True
run(0)
def set_cclamp(self, amp):
self.stim = h.IClamp(h.soma(0.5))
self.stim.amp = amp
self.stim.delay = 500
self.stim.dur = 1000
from neuron import h, gui
import sys
h('''create soma''')
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
ic = h.IClamp(h.soma(.5))
ic.delay = .5
ic.dur = 0.1
ic.amp = 0.3
# for testing external mod file
h.soma.insert("CaDynamics_E2")
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
v = h.Vector()
v.record(h.soma(.5)._ref_v, sec=h.soma)
i_mem = h.Vector()
i_mem.record(h.soma(.5)._ref_i_membrane_, sec=h.soma)
tv = h.Vector()
tv.record(h._ref_t, sec=h.soma)
h.run()
vstd = v.cl()
tvstd = tv.cl()
i_memstd = i_mem.cl()
#h.CoreNeuronRun[0].run()
pc = h.ParallelContext()
h.soma.diam = 10
h.soma.Ra = 100
h.soma.insert('pas') # add passive properties
h.soma.g_pas = 1/10000 # set the specific membrane resistance to 10000 ohm*cm^2
# add active conductances (the channels [mod files] are from Mainen and Sejnowski 1996)
h.soma.insert('kv') # add potassium channel
h.soma.gbar_kv = 2000 # set the potassium conductance
h.soma.insert('na') # add sodium channel
h.soma.gbar_na = 8000 # set the sodium conductance
h.celsius = 30
# current clamp
stim = h.IClamp(h.soma(0.5))
stim.amp = 0.007446
stim.delay = 250
stim.dur = 1000
# record voltage of some and injected current
# and the time
soma_v = h.Vector()
soma_v.record(h.soma(0.5)._ref_v)
stim_current = h.Vector()
stim_current.record(stim._ref_i)
t = h.Vector()
t.record(h._ref_t)
points = np.where(trace > (trace[15000] + 2))[0]
points = points[np.logical_and(points >= 20000, points <= 30000)]
width = dt * (points[-1] - points[0])
return width
h.load_file('init_models_with_ca/init_model2.hoc')
locations = [0, 0.1, 0.3, 0.5, 0.7, 0.9, 1]
probes = [0] * len(locations)
# Set up stimulus
stim = h.Ipulse2(h.soma(.5))
stim.amp = 2
stim.dur = 3
stim.delay = 500
stim.per = 10
stim.num = 1
# Set up recording vectors
for i in range(len(probes)):
probes[i] = h.Vector()
probes[i].record(h.apical(locations[i])._ref_v)
t_vec = h.Vector()
t_vec.record(h._ref_t)
dt = 0.025
h.tstop = 1000 - dt
from neuron import h, gui
from matplotlib import pyplot as plt
import numpy as np
import csv
from os import path
h.load_file('init_models_with_ca/init_model2.hoc')
soma_v_vec = h.Vector()
soma_v_vec.record(h.soma(0.5)._ref_v)
stim = h.IClamp(h.soma(0.5))
stim_vec = h.Vector()
stim_vec.record(stim._ref_i)
t_vec = h.Vector()
t_vec.record(h._ref_t)
dt = 0.025
h.tstop = 1500 - dt
stim.dur = 500
stim.delay = 500
lengths = np.array([200, 300, 400, 500, 600])
fig, axes = plt.subplots(2,
lengths.size,
squeeze=False,
sharex='all',
sharey='row',
figsize=(16, 8))
def pyrun(ratio=0, loc=0, nsyn=0, sec=h.a10_11, icamps=[], dendrec=True):
print '\n%s starting run' % (gethostname())
iotim = 0
tic = time()
h.tstop = 500
h.tsamp = h.tstop / h.dt + 1
h.synrec = 1
tsamp = int(h.tsamp)
r = h.Random(h.luckyoffset)
r.negexp(h.meanisi)
# stimulated branch
syn = h.synsuper(.5, r, sec=sec)
syndict = dict(sloc=loc, ratio=ratio, e1flag=0)
for name, value in syndict.iteritems():
setattr(syn.syn, name, value)
# initialize nseg with 'active' branches
seclist = [h.a1_111, h.a10_11]
sl2 = h.SectionList()
for sec in seclist:
sl2.append(sec=sec)
poppedsecs = sl2.unique()
h.refreshnseg(h.makeactivelist(sl2))
print 'nseg: %d' % (h.nsegcnt())
h.cvode.cache_efficient(1)
h.cvode_active(0)
# somatic voltage recordings, after nseg initialization
v = h.Vector(tsamp)
v.record(h.soma(.5)._ref_v)
trash = v.label(h.soma.name())
# voltage recording dictionary
vd = {'s': v}
if dendrec:
# dendritic voltage recording
h.distance(0, h.soma_con_pt, sec=h.soma)
d = h.distance(0, sec=sec)
locx = (loc - d) / sec.L
v = h.Vector(tsamp)
v.record(sec(locx)._ref_v)
trash = v.label(sec.name())
vd.update({'d': v})
h.poisson = 1
# 'background' current injection
ic = h.IClamp(0.5, sec=h.soma)
ic.dur = h.tstop
p_tstart = 100 # plot tstart, crop initial rising phase of current injection
ind_tstart = p_tstart / h.dt
t = npy.arange(0, h.tstop - p_tstart + h.dt, h.dt)
fh = newfig(figsize=(4, 8))
for runcnt, icamp in enumerate(icamps):
ic.amp = icamp
syn.syn.nsyn = nsyn
seed1 = float(686)
r1 = h.Random(seed1)
r1.negexp(h.meanisi)
syn.setrand(r1)
# run simulation
h.run()
postrunrecgather(vd)
# plot voltage and synaptic current
ax = fh.add_subplot(6, 1, runcnt + 3)
ax.plot(t, npy.array(vd['d'])[ind_tstart:], c='k', lw=1)
ax = fh.add_subplot(6, 1, runcnt + 5)
ax.plot(t,
-(vd['nis'].sum(0) + vd['ais'].sum(0))[ind_tstart:],
c='r',
lw=1)
ax = fh.add_subplot(6, 1, 1)
for ind, ras in enumerate(vd['rasind']):
ticks = ras[ras >= p_tstart]
ax.scatter(ticks - p_tstart,
npy.ones(ticks.shape) * ind,
marker=[[[0, 0], [0, 1]], 0])
ax.axis((0, h.tstop - p_tstart, 0, ind + 1))
zero = npy.zeros(t.size)
ax = fh.add_subplot(6, 1, 2)
ax.fill_between(t,
zero, (vd['ngs'] / vd['nbs']).sum(0)[ind_tstart:],
lw=2,
color='k')
ax.fill_between(t, zero, vd['ags'].sum(0)[ind_tstart:], lw=2, color='r')
ax.set_xlim([0, h.tstop - p_tstart])
fh.savefig('figs/Figure4.png')
print '%s running %d runs took %d seconds' % (gethostname(), runcnt + 1,
time() - tic)
h.dend.L = 500 # um
h.dend.diam = 1 # um
h.dend.Ra = 100 # ohm*cm
h.dend.insert('pas')
h.dend.g_pas = 1 / 10000
h.dend.connect(h.soma, 1,
0) #connect the end of the soma to the start of the dendrite
h("forall { nseg = int((L/(0.1*lambda_f(100))+0.9)/2)*2 + 1 }")
#########################
# # Set up experiment # #
#########################
stim = h.IClamp(h.soma(0.5)) # add a current clamp the the middle of the soma
stim.delay = 10 # ms
stim.dur = 100 # ms
stim.amp = 0.1 # nA
soma_v = h.Vector() # set up a recording vector
soma_v.record(h.soma(0.5)._ref_v) # record voltage at the middle of the soma
# Record voltage from all segments in the dendrite
dend_vs = []
for seg in h.dend:
dend_vs.append(h.Vector())
dend_vs[-1].record(seg._ref_v)
t = h.Vector()
t.record(h._ref_t) #record time.
off_path_inhibition.e = v_rest
#create a NetStim that will activate the synapses at t=1000
inhibition_netstim = h.NetStim(0.5, sec=h.dend)
inhibition_netstim.number = 1 # number of synaptic activation
inhibition_netstim.start = 1000 # activation start time
inhibition_netstim.noise = 0 # randomness
on_path_netcon = h.NetCon(inhibition_netstim, on_path_inhibition)
off_path_netcon = h.NetCon(inhibition_netstim, off_path_inhibition)
#########################
# # set up recording # #
#########################
soma_v = h.Vector() # set up a recording vector
soma_v.record(h.soma(0.5)._ref_v) # record voltage at the middle of the soma
# Record voltage from all segments in the dendrite
dend_vs = []
for seg in h.dend:
dend_vs.append(h.Vector())
dend_vs[-1].record(seg._ref_v)
t = h.Vector()
t.record(h._ref_t) #record time.
h.v_init = v_rest # set starting voltage
h.tstop = 5000 # set simulation time
###########################################
# Run simulation with proximal inhibition #
###########################################
pulses_amp=1.5,
pulses_num=0,
pulses_period=10):
stim.amp, stim.dur, stim.delay = Is, Is_dur, Is_onset
apical_stim.amp, apical_stim.dur, apical_stim.delay = Ia, Ia_dur, Ia_onset
tuft_stim.amp, tuft_stim.dur, tuft_stim.delay = It, It_dur, It_onset
epsp.imax, epsp.tau0, epsp.tau1, epsp.onset = Id_max, Id_rise, Id_decay, Id_onset
pulses.delay, pulses.dur, pulses.per, pulses.num, pulses.amp = pulses_onset, pulses_dur, pulses_period, pulses_num, pulses_amp
h.tstop = simdur - dt
h.run()
h.load_file('init_models_with_ca/init_model2.hoc')
# Set up stimulus
stim = h.IClamp(h.soma(0.5))
epsp = h.epsp(h.tuft(0.5))
pulses = h.Ipulse2(h.soma(0.5))
apical_stim = h.IClamp(h.apical(.5))
tuft_stim = h.IClamp(h.tuft(.5))
# Set up recording vectors
soma_v_vec = h.Vector()
soma_v_vec.record(h.soma(0.5)._ref_v)
apical_v_vec = h.Vector()
apical_v_vec.record(h.apical(0.5)._ref_v)
tuft_v_vec = h.Vector()
tuft_v_vec.record(h.tuft(0.5)._ref_v)
stim_vec = h.Vector()
stim_vec.record(stim._ref_i)
epsp_vec = h.Vector()
def test_spikes(
use_mpi4py=mpi4py_option,
use_nrnmpi_init=nrnmpi_init_option,
file_mode=file_mode_option,
):
print(
"test_spikes(use_mpi4py={}, use_nrnmpi_init={}, file_mode={})".format(
use_mpi4py, use_nrnmpi_init, file_mode))
h("""create soma""")
h.soma.L = 5.6419
h.soma.diam = 5.6419
h.soma.insert("hh")
h.soma.nseg = 3
ic = h.IClamp(h.soma(0.25))
ic.delay = 0.1
ic.dur = 0.1
ic.amp = 0.3
ic2 = h.IClamp(h.soma(0.75))
ic2.delay = 5.5
ic2.dur = 1
ic2.amp = 0.3
h.tstop = 10
h.cvode.use_fast_imem(1)
h.cvode.cache_efficient(1)
pc = h.ParallelContext()
pc.set_gid2node(pc.id() + 1, pc.id())
myobj = h.NetCon(h.soma(0.5)._ref_v, None, sec=h.soma)
pc.cell(pc.id() + 1, myobj)
# NEURON run
nrn_spike_t = h.Vector()
nrn_spike_gids = h.Vector()
# rank 0 record spikes for all gid while others
# for specific gid. this is for better test coverage.
pc.spike_record(-1 if pc.id() == 0 else (pc.id() + 1), nrn_spike_t,
nrn_spike_gids)
h.run()
nrn_spike_t = nrn_spike_t.to_python()
nrn_spike_gids = nrn_spike_gids.to_python()
# CORENEURON run
from neuron import coreneuron
coreneuron.enable = True
coreneuron.gpu = enable_gpu
coreneuron.file_mode = file_mode
coreneuron.verbose = 0
corenrn_all_spike_t = h.Vector()
corenrn_all_spike_gids = h.Vector()
pc.spike_record(-1, corenrn_all_spike_t, corenrn_all_spike_gids)
pc.set_maxstep(10)
def run(mode):
h.stdinit()
if mode == 0:
pc.psolve(h.tstop)
elif mode == 1:
while h.t < h.tstop:
pc.psolve(h.t + 1.0)
else:
while h.t < h.tstop:
h.continuerun(h.t + 0.5)
pc.psolve(h.t + 0.5)
corenrn_all_spike_t_py = corenrn_all_spike_t.to_python()
corenrn_all_spike_gids_py = corenrn_all_spike_gids.to_python()
# check spikes match
assert len(nrn_spike_t) # check we've actually got spikes
assert len(nrn_spike_t) == len(nrn_spike_gids) # matching no. of gids
if nrn_spike_t != corenrn_all_spike_t_py:
print(mode)
print(nrn_spike_t)
print(nrn_spike_gids)
print(corenrn_all_spike_t_py)
print(corenrn_all_spike_gids_py)
print([
corenrn_all_spike_t[i] - nrn_spike_t[i]
for i in range(len(nrn_spike_t))
])
assert nrn_spike_t == corenrn_all_spike_t_py
assert nrn_spike_gids == corenrn_all_spike_gids_py
if file_mode is False:
for mode in [0, 1, 2]:
run(mode)
else:
run(0)
return h
def test():
rall = 113 # axial resistance
cap = 1.1 # membrane capacitance
Rm = 45000.0 # membrane resistance
## INSERT ION CHANNELS:
for sec in h.allsec():
sec.insert("pas")
sec.e_pas = Epas
sec.g_pas = 1/Rm
sec.Ra = rall
sec.cm = cap
sec.insert("Cad")
sec.insert("it2")
sec.insert("hh2")
sec.ena = 50 # Reversal potential for sodium
sec.ek = -90 # Reversal potential for potassium
##################################################################
# Channel densities
gna = 0.04 # S/cm2
gkdr = 0.04
gcat = 0.0001
###################################################################
## INSERT STIMULATION ELECTRODES
stim = h.IClamp(h.soma(0.5))
# stim = h.IClamp(h.dend[99](0.5))
stim.delay = 1000
stim.dur = 10
stim.amp = 0#.25 #nA
###################################################################
## INSERT SYNAPSES
syn = h.Exp2Syn(h.dend[99](1)) # Sum of exp's
syn.tau1 = 0.5 # Rise (ms)
syn.tau2 = 2 # Decay (ms)
syn.e = 10 # Reversal # pot.
s = h.NetStim(0.5)
s.start = 1000 # start for distal synapses
s.number = 1
s.noise = 0
nc = h.NetCon(s,syn,sec = sec)
nc.weight[0] = 0.015
# syn1 = h.Exp2Syn(h.dend[83](1)) # Sum of exp's
# syn1.tau1 = 0.5 # Rise (ms)
# syn1.tau2 = 2 # Decay (ms)
# syn1.e = 10 # Reversal # pot.
# s1 = h.NetStim(0.5)
# s1.start = 1100 # start for distal synapses
# s1.number = 1
# s1.noise = 0
# nc1 = h.NetCon(s1,syn1,sec = sec)
# nc1.weight[0] = 0.05
#################################################################
# Split the morphology up in a suitable number of segments
freq = 50
h.geom_nseg(freq)
tot = 0
for sec in h.allsec():
tot += sec.nseg
h.distance()
print "total # of segments (50Hz):", tot
##################################################################
# INITIALIZE
gcat_func = lambda dist : 0.1054*gcat*(1 + 0.04*dist)
def initialize():
global Epas
h.celsius = celsius
for sec in h.soma:
h.distance()
for sec in h.allsec():
sec.v = Epas
sec.e_pas = Epas
sec.insert("pas")
sec.e_pas = Epas
sec.g_pas = 1/Rm
sec.Ra = rall
sec.cm = cap
sec.gnabar_hh2 = 0
sec.gkbar_hh2 = 0
dist = h.distance(0.5)
# sec.gcabar_it2 = gcat_func(dist)
sec.gcabar_it2 = gcat
for sec in h.soma:
sec.gnabar_hh2 = gna
sec.gkbar_hh2 = gkdr
# sec.gcabar_it2 = 0.1054*gcat
sec.gcabar_it2 = gcat
h.finitialize()
h.fcurrent()
cvode.re_init()
initialize()
vec ={}
for var in 't', 'd_sec', 'd_seg', 'diam_sec','gc','diam_seg','stim_curr':
vec[var] = h.Vector()
for var in 'V_sec', 'V_seg', 'CaConc_sec','CaConc_seg':
vec[var] = h.List()
def create_lists(vec):
for sec in h.allsec():
vec['d_sec'].append(h.distance(1))
vec['diam_sec'].append(sec.diam)
rec0 = h.Vector()
rec0.record(sec(0.5)._ref_v)
vec['V_sec'].append(rec0)
rec_Ca = h.Vector()
rec_Ca.record(sec(0.5)._ref_Cai)
vec['CaConc_sec'].append(rec_Ca)
for seg in sec:
vec['d_seg'].append(h.distance(0) + sec.L * seg.x)
vec['diam_seg'].append(seg.diam)
vec['gc'].append(seg.gcabar_it2)
rec = h.Vector()
rec.record(seg._ref_v)
vec['V_seg'].append(rec)
rec1 = h.Vector()
rec1.record(seg._ref_Cai)
vec['CaConc_seg'].append(rec1)
return vec
#####################################
create_lists(vec)
# run the simulation
vec['t'].record(h._ref_t)
# vec['current'].record(VC_patch._ref_i)
vec['stim_curr'].record(stim._ref_i)
h.load_file("stdrun.hoc")
h.tstop = 2500 # Simulation time
h.t = -500
h.run()
return vec
def pythonsim(simid, rundir, inputdir, modeldir, outputdir, currentstr, vinitstr, delaystr, stimdurationstr, timestepstr, simstopstr, recordintervalstr):
import neuron
from neuron import h
Section = h.Section
# All the mods are in the simulator directory
# Apparently this is already done by the import, if we use the "-dll"
# option in the call to nrniv. Since the import behavior seems to be
# inconsistent without the "-dll" option, we do it this way.
#neuron.load_mechanisms(".") # NOT USING LOAD_MECHANISMS HERE
# local auto-modified version for no gui and control over certain ion
# channels is in the run dir
h.xopen(rundir + '/' + simid + '.hoc')
timefilepath = outputdir + '/' + 'timedata.txt'
timefile = h.File()
timefile.wopen(timefilepath, 'w')
voltagefilepath = outputdir + '/' + 'voltagedata.txt'
voltagefile = h.File()
voltagefile.wopen(voltagefilepath, 'w')
current = float(currentstr)
vinit = float(vinitstr)
delay = float(delaystr)
stimduration = float(stimdurationstr)
timestep = float(timestepstr)
simstop = float(simstopstr)
recordinterval = float(recordintervalstr)
# ----- Current Injection
stim = h.IClamp(0.5, sec=h.soma)
stim.amp = current # nA
stim.delay = delay # msec
stim.dur = stimduration # msec
# ----- Simulation Control (now mostly done through input arguments)
h.dt = timestep
# Preallocate record vectors for speed
# Requires recordinterval to be an exact multiple of tstep.
recordlength = simstop/recordinterval + 1
testt = h.Vector(recordlength)
testv = h.Vector(recordlength)
# Recording at the soma
testt.record(h._ref_t, recordinterval)
testv.record(h.soma(0.5)._ref_v, recordinterval)
# Initialize
h.finitialize(vinit)
h.fcurrent()
# Integrate
while h.t <= simstop:
v = h.soma.v
h.fadvance()
# Shutdown
testt.printf(timefile, '%f\n')
testv.printf(voltagefile, '%f\n')
timefile.close()
voltagefile.close()
return(0)
