Changelog¶
Changelog¶
All notable changes to NEVO will be documented in this file.
The format follows Keep a Changelog. This project adheres to Semantic Versioning.
[Unreleased]¶
[0.1.1] - 2026-03-23¶
Added¶
nm_dualoperator mode: hard winner-take-all switching betweenNeuromorphicExplorationEnsembleandNeuromorphicExploitationEnsemble.nm_softmixoperator mode: softmax-blended populations with configurabletemperatureandconcentration.NeuromorphicExplorationEnsemble: Nengo LIF ensemble with Poisson spike generation, fast synapses (τ=5 ms), and broad encoders.NeuromorphicExploitationEnsemble: Nengo LIF ensemble with slow synapses (τ=20 ms), sharp tuning, and attractor dynamics.Modular TD learning system (
nevo/core/td_learning.py):SimpleTDRule,DecayingTDRule,ConservativeTDRule,AdaptiveTDRule,LinearValueModel,BoundedValueModel,EligibilityTraceManager,TemporalDifferenceLearner.TD(λ) support via eligibility traces.
Dynamic rule/model switching at runtime:
opt.bg_selector.set_learning_rule(),set_value_model(),set_td_lambda().Pass-through helpers on
NEVOptimiser:set_td_learning_rule(),set_td_value_model(),set_td_lambda().Nine additional standard operators:
RandomSearch,GravitationalSearch,FireflyAlgorithm,CentralForce,GeneticCrossover,GeneticMutation,LocalRandomWalk,SimulatedAnnealing,TabuSearch.OPERATOR_REGISTRYandget_operator()factory innevo/operators/__init__.py.Comprehensive integration test suite (
tests/test_td_integration.py) coveringnm_dualandnm_softmixmodes.
Fixed¶
All documentation operator mode strings updated to match code (
"trad","nm_dual","nm_softmix").td_learning_examples.py: replaced non-existentoptimize()andbasal_ganglia_selectorwithrun()andbg_selector.
[0.1.0] — 2025-11-19¶
Added¶
Initial project structure.
NEVOptimisermain class with Nengo-based simulation loop.BasalGangliaSelectorfor neuromorphic operator selection.StateFeaturesfor feature extraction (diversity, improvement rate, convergence).Four core operators:
LevyFlight,DifferentialEvolution,ParticleSwarm,SpiralOptimisation.v-space normalisation (
trs2o()) and competitive memory archive.Visualisation utilities:
plot_optimisation_results(),plot_operator_statistics().IOHexperimenter integration.
Basic example script and README.
MIT Licence.