Model: Izhikevich 2003
The Izhikevich (2003) model is a computationally efficient spiking neuron with four firing modes from different parameter sets.
Reference: NeuroML2 LEMS_NML2_Ex2_Izh.xml
1. Define Network in TVBO
from tvbo import SimulationExperiment
exp = SimulationExperiment.from_string("""
label: "NeuroML Ex2: Izhikevich"
dynamics:
name: izBurst
iri: neuroml:izhikevichCell
network:
dynamics:
izBurst:
name: izBurst
iri: neuroml:izhikevichCell
parameters:
v0: {value: -70, unit: mV}
thresh: {value: 30, unit: mV}
a: {value: 0.02}
b: {value: 0.2}
c: {value: -50}
d: {value: 2}
izTonic:
name: izTonic
iri: neuroml:izhikevichCell
parameters:
v0: {value: -70, unit: mV}
thresh: {value: 30, unit: mV}
a: {value: 0.02}
b: {value: 0.2}
c: {value: -65}
d: {value: 6}
izMixed:
name: izMixed
iri: neuroml:izhikevichCell
parameters:
v0: {value: -70, unit: mV}
thresh: {value: 30, unit: mV}
a: {value: 0.02}
b: {value: 0.2}
c: {value: -55}
d: {value: 4}
izClass1:
name: izClass1
iri: neuroml:izhikevichCell
parameters:
v0: {value: -60, unit: mV}
thresh: {value: 30, unit: mV}
a: {value: 0.02}
b: {value: -0.1}
c: {value: -55}
d: {value: 6}
i0:
name: i0
iri: neuroml:pulseGeneratorDL
parameters:
delay: {value: 22, unit: ms}
duration: {value: 2000, unit: ms}
amplitude: {value: 15}
i1:
name: i1
iri: neuroml:pulseGeneratorDL
parameters:
delay: {value: 20, unit: ms}
duration: {value: 2000, unit: ms}
amplitude: {value: 14}
i2:
name: i2
iri: neuroml:pulseGeneratorDL
parameters:
delay: {value: 20, unit: ms}
duration: {value: 2000, unit: ms}
amplitude: {value: 10}
rg0:
name: rg0
iri: neuroml:rampGeneratorDL
parameters:
delay: {value: 30, unit: ms}
duration: {value: 170, unit: ms}
startAmplitude: {value: -32}
finishAmplitude: {value: 50}
baselineAmplitude: {value: -32}
nodes:
- {id: 0, dynamics: izBurst}
- {id: 1, dynamics: izTonic}
- {id: 2, dynamics: izMixed}
- {id: 3, dynamics: izClass1}
- {id: 10, dynamics: i0, record: false}
- {id: 11, dynamics: i1, record: false}
- {id: 12, dynamics: i2, record: false}
- {id: 13, dynamics: rg0, record: false}
edges:
- {source: 10, target: 0}
- {source: 11, target: 1}
- {source: 12, target: 2}
- {source: 13, target: 3}
integration:
method: euler
step_size: 0.005
duration: 200.0
time_scale: ms
""")
print(f"Network dynamics: {list(exp.network.dynamics.keys())}")
Network dynamics: ['i0', 'i1', 'i2', 'izBurst', 'izClass1', 'izMixed', 'izTonic', 'rg0']
* Owlready2 * Warning: ignoring cyclic subclass of/subproperty of, involving:
http://uri.interlex.org/tgbugs/uris/readable/atlas/Space
2. Render LEMS XML
xml = exp.render("lems")
print(xml[:2000])
<Lems>
<!-- Tell jLEMS/jNeuroML which component is the simulation entry point. -->
<Target component="sim_NeuroML_Ex2__Izhikevich"/>
<!-- ════════════════════════════════════════════════════════════════
Standard NeuroML2 type includes for network mode.
Provides all standard dimensions, units, synapse types, input
types, network infrastructure, and simulation types.
════════════════════════════════════════════════════════════════ -->
<Include file="Cells.xml"/>
<Include file="Networks.xml"/>
<Include file="Simulation.xml"/>
<!-- ════════════════════════════════════════════════════════════════
Dynamics ComponentType & Component instances
════════════════════════════════════════════════════════════════ -->
<!-- ── ComponentType: izBurst ── -->
<ComponentType name="izBurst" extends="baseCellMembPot">
<Parameter name="a" dimension="none"/>
<Parameter name="b" dimension="none"/>
<Parameter name="c" dimension="none"/>
<Parameter name="d" dimension="none"/>
<Parameter name="thresh" dimension="voltage"/>
<Parameter name="I_amp" dimension="none"/>
<Parameter name="pulse_delay" dimension="time"/>
<Parameter name="pulse_duration" dimension="time"/>
<Parameter name="v0" dimension="voltage"/>
<Parameter name="refract" dimension="time"/>
<Parameter name="v_0" dimension="none" />
<Parameter name="U_0" dimension="none" />
<Constant name="SEC" dimension="time" value="1ms"/>
<Exposure name="U" dimension="none" />
<!-- Dynamically attached synapses/inputs from network connections -->
<Attachments name="synapses" type="basePointCurrent"/>
<Dynamics>
<StateVariable name="v" dimension="none" exposure="v"/>
<StateVariable name="U" dimension="none" exposure="U"/>
<StateVariable name="lastSpikeTime" dimension="time"/>
<ConditionalDerivedVariable name="I_ext" dimension="none">
<Case condition="pulse_delay .leq. t .and. t .lt. pul
3. Run Reference
from tvbo.adapters.neuroml import run_lems_example
ref_outputs = run_lems_example("LEMS_NML2_Ex2_Izh.xml")
for name, arr in ref_outputs.items():
print(f" {name}: shape={arr.shape}")
auto.dat: shape=(40001, 10)
4. Run TVBO
result = exp.run("neuroml")
da = result.integration.data
print(f"TVBO: {da.dims}, shape={da.shape}")
pyNeuroML >>> 13:07:09 - INFO - Loading LEMS file: tvbo_lems_sim.xml and running with jNeuroML
pyNeuroML >>> 13:07:09 - INFO - Executing: (java -Xmx400M -Djava.awt.headless=true -jar "/Users/leonmartin_bih/tools/tvbo/.venv/lib/python3.12/site-packages/pyneuroml/utils/./../lib/jNeuroML-0.14.0-jar-with-dependencies.jar" tvbo_lems_sim.xml -nogui -I '') in directory: /var/folders/ym/9kw1g21j1nd7kwfn8c0z3st40000gn/T/tmpz8cp6qru
pyNeuroML >>> 13:07:10 - INFO - Command completed successfully!
TVBO: ('time', 'quantity'), shape=(40001, 4)
5. Compare & Plot
from tvbo.adapters.neuroml import plot_lems_comparison
plot_lems_comparison("LEMS_NML2_Ex2_Izh.xml", ref_outputs, result.integration.data, title_prefix="Ex2")
