Ngclearn

— — — — — — — — -
* rewrite of framework to adhere to revised ngc-sim-lib
* simple starting point for lava support
* new adex, wtas, and event-based stdp components integrated
* all current tutorials written
* clean-up of utility sub-package files
* minor clean-up, revisions, bug fixes and doc-string updates
* model museum updates reflect current state of ngc-learn/sim-lib


=======
History
=======

— — — — — — — — -
* backend migration/revision finished; ngc-learn now depends on ngc-sim-lib (ngcsimlib)
* tutorials updated reflected
* STDP is now first-class citizen

— — — — — — — — -
* ngc-learn backend re-engineered to be pure Jax
* LIF, quad-LIF spiking cells, as well as a set of graded neuronal cells, integrated
* exponential STDP synaptic plasticity integrated
* power-law STDP synaptic plasticity integrated
* docs rebuilts/revised to adhere to new ngc-learn nodes-and-cables simulator/controller backend -- ngcsimlib

— — — — — — — — -
* Spiking nodes (LIF and ELIF) support implemented as core nodes
* networkx support provided in experimental auxiliary folder for op-graph visualization
* Walkthrough 7 written to demonstrate how to train a spiking network with error-Hebbian feedback

— — — — — — — — -
* Harmonium written into model museum
* "demonstrations" have been renamed to "walkthroughs"
* tutorial 1 written/provided
* nodes-and-cables system upgraded/polished; incremental simulation centrally supported
* infinite-mnist data generating process integrated into experimental

— — — — — — — — -
* NGCGraph .compile() further tweaked to use an injection/clamping look-up
system to allow for dynamic changes to occur w/in a static graph compiled
simulated NGC system
* Cable API slightly modified to increase flexiblity (demonstrations and
tests modified to reflect updated API)
* Demonstration 6 released showcasing how to use ngc-learn to construct/fit a
restricted Boltzmann machine