spark.nn.components.somas

spark.nn.components.somas#

Submodules#

Classes#

`Soma`	Abstract soma model.
`SomaOutput`	Generic soma model output spec.
`LeakySoma`	Leaky soma model.
`LeakySomaConfig`	LeakySoma model configuration class.
`RefractoryLeakySoma`	Leaky soma with refractory time model.
`RefractoryLeakySomaConfig`	RefractoryLeakySoma model configuration class.
`StrictRefractoryLeakySoma`	Leaky soma with strict refractory time model.
`StrictRefractoryLeakySomaConfig`	StrictRefractoryLeakySoma model configuration class.
`AdaptiveLeakySoma`	Adaptive leaky soma model.
`AdaptiveLeakySomaConfig`	AdaptiveLeakySoma model configuration class.
`ExponentialSoma`	Exponential soma model.
`ExponentialSomaConfig`	ExponentialSoma model configuration class.
`RefractoryExponentialSoma`	Exponential soma with refractory time model.
`RefractoryExponentialSomaConfig`	RefractoryExponentialSoma model configuration class.
`AdaptiveExponentialSoma`	Adaptive Exponential soma model.
`AdaptiveExponentialSomaConfig`	AdaptiveExponentialSoma model configuration class.
`SimplifiedAdaptiveExponentialSoma`	Simplified Adaptive Exponential soma model. This model drops the subthreshold adaptation.
`SimplifiedAdaptiveExponentialSomaConfig`	SimplifiedAdaptiveExponentialSoma model configuration class.
`IzhikevichSoma`	Izhikevich soma model.
`IzhikevichSomaConfig`	IzhikevichSoma model configuration class.

Package Contents#

class spark.nn.components.somas.Soma(config=None, **kwargs)[source]#

Bases: spark.nn.components.base.Component, Generic[ConfigT]

Abstract soma model.

Parameters:: config (ConfigT | None)

config: ConfigT[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])

reset()[source]#: Resets neuron states to their initial values.

__call__(current)[source]#

Update neuron’s states and compute spikes.

Parameters:: current (spark.core.payloads.CurrentArray)
Return type:: SomaOutput

class spark.nn.components.somas.SomaOutput[source]#

Bases: TypedDict

Generic soma model output spec.

Initialize self. See help(type(self)) for accurate signature.

spikes: spark.core.payloads.SpikeArray[source]#

potential: spark.core.payloads.PotentialArray[source]#

class spark.nn.components.somas.LeakySoma(config=None, **kwargs)[source]#

Bases: spark.nn.components.somas.base.Soma

Leaky soma model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 1.3 Integrate-And-Fire Models https://neuronaldynamics.epfl.ch/online/Ch1.S3.html

Parameters:: config (LeakySomaConfig | None)

config: LeakySomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

class spark.nn.components.somas.LeakySomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: spark.nn.components.somas.base.SomaConfig

LeakySoma model configuration class.

Parameters:: __skip_validation__ (bool)

potential_rest: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

potential_reset: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

potential_tau: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

resistance: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

threshold: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.RefractoryLeakySoma(config=None, **kwargs)[source]#

Bases: LeakySoma

Leaky soma with refractory time model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 1.3 Integrate-And-Fire Models https://neuronaldynamics.epfl.ch/online/Ch1.S3.html

Parameters:: config (RefractoryLeakySomaConfig | None)

config: RefractoryLeakySomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

reset()[source]#

Resets component state.

Return type:: None

class spark.nn.components.somas.RefractoryLeakySomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: LeakySomaConfig

RefractoryLeakySoma model configuration class.

Parameters:: __skip_validation__ (bool)

cooldown: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.StrictRefractoryLeakySoma(config=None, **kwargs)[source]#

Bases: RefractoryLeakySoma

Leaky soma with strict refractory time model. Note: This model is here mostly for didactic/historical reasons.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 1.3 Integrate-And-Fire Models https://neuronaldynamics.epfl.ch/online/Ch1.S3.html

Parameters:: config (RefractoryLeakySomaConfig | None)

config: StrictRefractoryLeakySomaConfig[source]#

class spark.nn.components.somas.StrictRefractoryLeakySomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: RefractoryLeakySomaConfig

StrictRefractoryLeakySoma model configuration class.

Parameters:: __skip_validation__ (bool)

class spark.nn.components.somas.AdaptiveLeakySoma(config=None, **kwargs)[source]#

Bases: RefractoryLeakySoma

Adaptive leaky soma model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array threshold_tau: float | jax.Array threshold_delta: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 5.1 Thresholds in a nonlinear integrate-and-fire model https://neuronaldynamics.epfl.ch/online/Ch5.S1.html

Parameters:: config (AdaptiveLeakySomaConfig | None)

config: AdaptiveLeakySomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

reset()[source]#

Resets component state.

Return type:: None

class spark.nn.components.somas.AdaptiveLeakySomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: RefractoryLeakySomaConfig

AdaptiveLeakySoma model configuration class.

Parameters:: __skip_validation__ (bool)

threshold_tau: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

threshold_delta: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.ExponentialSoma(config=None, **kwargs)[source]#

Bases: spark.nn.components.somas.base.Soma

Exponential soma model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: How Spike Generation Mechanisms Determine the Neuronal Response to Fluctuating Inputs Nicolas Fourcaud-Trocmé, David Hansel, Carl van Vreeswijk, and Nicolas Brunel The Journal of Neuroscience, December 17, 2003 https://www.jneurosci.org/content/23/37/11628 Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 5.2 Exponential Integrate-and-Fire Model https://neuronaldynamics.epfl.ch/online/Ch5.S2.html

Parameters:: config (ExponentialSomaConfig | None)

config: ExponentialSomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

class spark.nn.components.somas.ExponentialSomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: spark.nn.components.somas.base.SomaConfig

ExponentialSoma model configuration class.

Parameters:: __skip_validation__ (bool)

potential_rest: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

potential_reset: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

potential_tau: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

resistance: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

threshold: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

rheobase_threshold: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

spike_slope: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.RefractoryExponentialSoma(config=None, **kwargs)[source]#

Bases: ExponentialSoma

Exponential soma with refractory time model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: How Spike Generation Mechanisms Determine the Neuronal Response to Fluctuating Inputs Nicolas Fourcaud-Trocmé, David Hansel, Carl van Vreeswijk, and Nicolas Brunel The Journal of Neuroscience, December 17, 2003 https://www.jneurosci.org/content/23/37/11628 Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 5.2 Exponential Integrate-and-Fire Model https://neuronaldynamics.epfl.ch/online/Ch5.S2.html

Parameters:: config (RefractoryExponentialSomaConfig | None)

config: RefractoryExponentialSomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

reset()[source]#

Resets component state.

Return type:: None

class spark.nn.components.somas.RefractoryExponentialSomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: ExponentialSomaConfig

RefractoryExponentialSoma model configuration class.

Parameters:: __skip_validation__ (bool)

cooldown: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.AdaptiveExponentialSoma(config=None, **kwargs)[source]#

Bases: ExponentialSoma

Adaptive Exponential soma model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array threshold_tau: float | jax.Array threshold_delta: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray

Reference:

Adaptive Exponential Integrate-and-Fire Model as an Effective Description of Neuronal Activity. Romain Brette and Gerstner Wulfram Gerstner W, Kistler WM, Naud R, Paninski L. Journal of Neurophysiology vol. 94, no. 5, pp. 3637-3642, 2005 https://doi.org/10.1152/jn.00686.2005 Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 5.2 Exponential Integrate-and-Fire Model https://neuronaldynamics.epfl.ch/online/Ch5.S2.html

Parameters:: config (AdaptiveExponentialSomaConfig | None)

config: AdaptiveExponentialSomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

reset()[source]#: Resets neuron states to their initial values.

class spark.nn.components.somas.AdaptiveExponentialSomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: ExponentialSomaConfig

AdaptiveExponentialSoma model configuration class.

Parameters:: __skip_validation__ (bool)

adaptation_tau: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

adaptation_delta: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

adaptation_subthreshold: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.SimplifiedAdaptiveExponentialSoma(config=None, **kwargs)[source]#

Bases: RefractoryExponentialSoma

Simplified Adaptive Exponential soma model. This model drops the subthreshold adaptation.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array potential_tau: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array cooldown: float | jax.Array threshold_tau: float | jax.Array threshold_delta: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray

Reference:

Adaptive Exponential Integrate-and-Fire Model as an Effective Description of Neuronal Activity. Romain Brette and Gerstner Wulfram Gerstner W, Kistler WM, Naud R, Paninski L. Journal of Neurophysiology vol. 94, no. 5, pp. 3637-3642, 2005 https://doi.org/10.1152/jn.00686.2005 Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition. Gerstner W, Kistler WM, Naud R, Paninski L. Chapter 5.2 Exponential Integrate-and-Fire Model https://neuronaldynamics.epfl.ch/online/Ch5.S2.html

Parameters:: config (SimplifiedAdaptiveExponentialSomaConfig | None)

config: SimplifiedAdaptiveExponentialSomaConfig[source]#

build(input_specs)[source]#

Build method.

Parameters:: input_specs (dict[str, spark.core.specs.PortSpecs])
Return type:: None

reset()[source]#

Resets component state.

Return type:: None

class spark.nn.components.somas.SimplifiedAdaptiveExponentialSomaConfig(__skip_validation__=False, **kwargs)[source]#

Bases: RefractoryExponentialSomaConfig

SimplifiedAdaptiveExponentialSoma model configuration class.

Parameters:: __skip_validation__ (bool)

threshold_tau: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

threshold_delta: float | jax.Array | spark.nn.initializers.base.Initializer[source]#

class spark.nn.components.somas.IzhikevichSoma(config=None, **kwargs)[source]#

Bases: spark.nn.components.somas.base.Soma

Izhikevich soma model.

Init:: units: tuple[int, …] potential_rest: float | jax.Array potential_reset: float | jax.Array resistance: float | jax.Array threshold: float | jax.Array recovery_timescale: float | jax.Array recovery_sensitivity: float | jax.Array recovery_update: float | jax.Array
Input:: in_spikes: SpikeArray
Output:: out_spikes: SpikeArray
Reference:: Simple Model of Spiking Neurons Eugene M. Izhikevich IEEE Transactions on Neural Networks, vol. 14, no. 6, pp. 1569-1572, Nov. 2003 https://doi.org/10.1109/TNN.2003.820440