How dreamsproxy/lisnn works
LiSNN’s active product is an experimental, vectorized spiking-neuron substrate; the legacy tree is explicitly frozen. The user-facing path is constructing a homogeneous or mixed population, then—where desired—stepping it with fixed-weight causal propagation. Population layout, model dispatch, sparse topology, and validation form the supporting layers. Spatial utilities and smoke diagnostics are separate capabilities. Runtime source was only partly sampled, so edges are limited to README statements and references visible in the SOURCE INDEX; the graph does not imply task-driven learning, which the README says is not implemented.
Main entry points: runtime.py Fixed-weight runtime · core.py SNN constructor · propagation.py Current propagation · registry.py Neuron registry
Components 19
Network construction
- Construction example
main.py - SNN constructor
lisnn/network/core.py - Population specification
lisnn/network/spec.py - Population builder
lisnn/neurons/population.py
Neuron substrate
- Population layout
lisnn/neurons/layout.py - Neuron registry
lisnn/neurons/registry.py - Vectorized kernels
lisnn/neurons/kernels.py - Izhikevich adapters
lisnn/neurons/adapters.py
Connectivity and runtime
- Sparse topology
lisnn/synapses/core.py - Current propagation
lisnn/synapses/propagation.py - Fixed-weight runtime
lisnn/network/runtime.py - Numeric validation
lisnn/validation.py
Spatial and diagnostics
- Spatial morphology
lisnn/spatial/morphology.py - Spatial placement
lisnn/spatial/placement.py - Spatial volume
lisnn/spatial/volume.py - Smoke-suite runner
lisnn/debugging/runner.py - Population diagnostics
lisnn/debugging/neuron_smoke.py - Synapse diagnostics
lisnn/debugging/synapse_smoke.py
Other components
- Engineer
Connections 19
- Engineer → SNN constructor: constructs
- Construction example → SNN constructor: demonstrates
main.py - SNN constructor → Population specification: uses
lisnn/network/core.py - SNN constructor → Neuron registry: resolves models
lisnn/network/core.py - SNN constructor → Population builder: builds
- SNN constructor → Population layout: uses
- Neuron registry → Vectorized kernels: registers steps
lisnn/neurons/registry.py - Fixed-weight runtime → SNN constructor: steps
lisnn/network/runtime.py - Fixed-weight runtime → Neuron registry: resolves steps
lisnn/network/runtime.py - Fixed-weight runtime → Vectorized kernels: runs kernels
lisnn/network/runtime.py - Fixed-weight runtime → Izhikevich adapters: adapts units
lisnn/network/runtime.py - Fixed-weight runtime → Sparse topology: reads edges
lisnn/network/runtime.py - Fixed-weight runtime → Current propagation: propagates spikes
lisnn/network/runtime.py - Fixed-weight runtime → Numeric validation: validates inputs
lisnn/network/runtime.py - Current propagation → Sparse topology: validates edges
lisnn/synapses/propagation.py - Current propagation → Numeric validation: validates values
lisnn/synapses/propagation.py - Spatial volume → Spatial placement: uses placement
lisnn/spatial/volume.py - Smoke-suite runner → Population diagnostics: runs
lisnn/debugging/runner.py - Smoke-suite runner → Synapse diagnostics: runs
lisnn/debugging/runner.py