norse.torch.functional.tsodyks_makram module

class norse.torch.functional.tsodyks_makram.TsodyksMakramParameters(tau_f_inv: float = 20.0, tau_s_inv: float = 50.0, tau_d_inv: float = 1.3333333333333333, U: float = 0.45)

Bases: tuple

Parameters of the Tsodyks-Makram Model

Parameters

• tau_f_inv (float) – facilitation time constant (in ms).

• tau_s_inv (float) – synaptic time constant (in ms).

• tau_d_inv (float) – depressing time constant (in ms).

• U (float) – size of the jump in the utilization variable.

Create new instance of TsodyksMakramParameters(tau_f_inv, tau_s_inv, tau_d_inv, U)

U: float

Alias for field number 3

tau_d_inv: float

Alias for field number 2

tau_f_inv: float

Alias for field number 0

tau_s_inv: float

Alias for field number 1

class norse.torch.functional.tsodyks_makram.TsodyksMakramState(u: torch.Tensor, x: torch.Tensor)

Bases: tuple

State of the Tsodyks-Makram Model, note that we are tracking the input current state separately.

Parameters

• u (torch.Tensor) – utilization parameter.

• x (torch.Tensor) – fraction of remaining available resources.

Create new instance of TsodyksMakramState(u, x)

u: torch.Tensor

Alias for field number 0

x: torch.Tensor

Alias for field number 1

norse.torch.functional.tsodyks_makram.stp_step(z, s, p, dt=0.001)

Euler integration step for Tsodyks Makram model of STP.

Reference: http://www.scholarpedia.org/article/Short-term_synaptic_plasticity

Parameters

• z (torch.Tensor) – Input spikes

• s (TsodyksMakramState) – State of the Tsodyks-Makram model

• p (TsodyksMakramParameters) – Parameters of the Tsodyks-Makram model

• dt (float) – Euler integration timestep

Return type

Tuple[Tensor, TsodyksMakramState]