Cav2p1_MA2024_PC_Frozen

Cav2p1_MA2024_PC_Frozen#

class braincell.channel.Cav2p1_MA2024_PC_Frozen(size, g_max=Quantity(0.00022, 'cm / s'), V_sh=Quantity(0., 'mV'), temp=Quantity(296.15, 'K'), name=None)#

Purkinje cell Cav2.1 with the GHK drive frozen for autodiff.

A standalone Purkinje-cell Cav2.1 mechanism that stops the gradient through the membrane potential where it enters the constant-field (GHK) flux term. Its kinetics and its named parameter block are those of Cav2p1_RI2021_SC, and its attribution is that of Cav2p1_MA2024_PC: the Cav2.1 kinetics built from the dissociated Purkinje neuron recordings of (Swensen & Bean, 2005) [1] and published in (Anwar, Hong & De Schutter, 2012) [2], imported for the human Purkinje cell model of (Masoli et al., 2024) [3]. It inherits none of that code, and its forward current is not identical to the unfrozen class’s – see Notes.

Parameters:
  • size (int | Sequence[int] | integer | Sequence[integer]) – Channel state shape.

  • g_max (Array | ndarray | bool | number | bool | int | float | complex | Quantity | Callable) – Calcium permeability entering the GHK flux (the mod file’s pcabar), despite the g_max name. Defaults to 2.2e-4 cm/s.

  • V_sh (Array | ndarray | bool | number | bool | int | float | complex | Quantity | Callable) – Voltage shift subtracted from \(V\) before every rate and before the GHK term; the mod file’s vshift. Defaults to 0.0 mV.

  • temp (Array | ndarray | bool | number | bool | int | float | complex | Quantity) – Absolute temperature. Enters both the gate’s Q10 factor and the GHK flux. Defaults to 23 degrees Celsius.

  • name (str | None) – Optional channel name.

See also

Cav2p1_MA2024_PC

The unfrozen Purkinje-cell import of the same mechanism, and this class’s base.

Cav2p1_RI2021_SC

Where the equation set, the permeability discrepancy and the mod header corrections are documented in full.

Cav2p1_RI2021_SC_Frozen

Stellate-cell frozen variant, built the same way on top of Cav2p1_RI2021_SC.

Cav2p1_MA2025_BC_Frozen

Basket-cell frozen variant, likewise built on top of Cav2p1_MA2025_BC.

Notes

The class body is the two declarations _FrozenCav3p1Ghk supplies. Every gate, constant and rate function comes from Cav2p1_MA2024_PC, so a change to Cav2p1_RI2021_SC propagates here. The two Cav2.1 frozen siblings, Cav2p1_RI2021_SC_Frozen and Cav2p1_MA2025_BC_Frozen, are built the same way over their own unfrozen classes rather than over this one.

“Frozen” describes the gradient path, not the channel states. The m gate keeps integrating exactly as in the unfrozen class – nothing stops evolving. freeze_drive_gradient routes the membrane potential through freeze_gradient() before it reaches the GHK helper, so the drive’s explicit voltage dependence contributes nothing to reverse-mode gradients while the value it computes is unchanged. The unfrozen V still reaches conductance_factor(), but that method ignores its voltage argument and reads only the gate states, so the distinction affects neither value nor gradient there.

The freezing is not the only difference from the unfrozen class. _FrozenCav3p1Ghk also declares ghk as the module-level _cav3p1_nmodl_ghk_flux helper, whereas Cav2p1_RI2021_SC calls the shared ghk_flux(). The two helpers use different physical constants: the shared one uses CODATA values and the temperature as passed, while the NMODL helper uses F = 9.6485e4 C/mol, R = 8.3145 J/(K mol) and a 273.19 + celsius Kelvin conversion carried as _CAV3P1_NMODL_TEMP_OFFSET. Those are exactly the constants Cav2p1_MA24_PC.mod itself declares, so the helper’s _cav3p1_ name – inherited from the Cav3.1 import, whose mod file declares the same values – understates its applicability here. The consequence is that this class’s forward current differs from Cav2p1_MA2024_PC’s by of order \(10^{-4}\) in relative terms, and that this class is the closer of the two to the mod file. Nothing in the mod file corresponds to the freezing itself: it is a BrainCell autodiff facility with no NMODL counterpart, so it is not an import deviation and changes nothing a NEURON comparison would observe.

As in Cav2p1_RI2021_SC, g_max is a permeability in cm/s rather than a conductance density, the mod file’s unused vhalfh/cvh are dropped, the Q10 is declared on the Gate because the mod file divides the whole tau by it, and current() negates NEURON’s outward-positive ica to match BrainCell’s inward-positive convention. The permeability default is the deposit’s tuned pcabar, not a value reported by either origin paper.

References