Namespace DotFly.Core.Models

Classes

SignPolicy

Maps a neurotransmitter label to the sign of the presynaptic neuron's outgoing weights. The policy is applied once by the checkpoint builder and recorded in the checkpoint's provenance; kernels only ever see signed contact counts.

Structs

LifStepConstants

Per-step constants of the exact LIF integrator for a fixed timestep, as consumed by kernels. Produced by Discretize(double); all values are in mV / ms / steps.

NeuronModel

Parameters of the leaky integrate-and-fire (LIF) point-neuron model used by dotFly. Defaults reproduce Shiu et al. 2024 (philshiu/Drosophila_brain_model@91bdd1e7, model.py):

dv/dt = (v_0 - v + g) / t_mbr   (unless refractory)
dg/dt = -g / tau                (unless refractory)
threshold: v > v_th;  reset: v = v_rst, g = 0;  refractory t_rfc
synapse:   g += w (after delay t_dly), w = signed contact count × w_syn

All voltages are in millivolts and all times in milliseconds so the descriptor is unit-free at the kernel boundary. Use Discretize(double) to obtain the per-step constants of the exact (matrix-exponential) integrator.

Enums

Neurotransmitter

Predicted neurotransmitter of a neuron, as labelled in the connectome releases (MaleCNS consensusNt, FlyWire nt_type). A single label per neuron is a modelling convention, not receptor-resolved physiology.