OPTIMIZATION & SWARMS PRODUCTION READY LOCAL AIR-GAPPED
HyperSwarm Optimization Suite & Benchmarks
Execute matrix sweeps, dynamic Director authority switching, and evaluate 35 optimizer modes.
HyperSwarm Optimization Suite & Benchmarks
The HyperSwarm Optimization Suite is a high-assurance metaheuristic framework containing 35 optimizer modes (12 core swarm models, 14 hybrids, and 9 advanced/multi-objective engines) with compiled Intel Level Zero GPU kernels.
35 Optimizer Modes Breakdown
┌────────────────────────────────────────────────────────────────────────┐
│ 35 Optimizer Modes │
├───────────────────────┬────────────────────────┬───────────────────────┤
│ Core Swarms (12) │ Hybrid Engines (14) │ Advanced & MO (9) │
│ • Particle Swarm (PSO)│ • PSO-GA Hybrid │ • NSGA-II Multi-Obj │
│ • Ant Colony (ACO) │ • Bat-Firefly Hybrid │ • MO-PSO Pareto │
│ • Artificial Bee (ABC)│ • Gravitational-PSO │ • Of Bats & Beacons │
│ • Grey Wolf (GWO) │ • Cuckoo-Differential │ • TuRBO Trust-Region │
│ • Whale (WOA) │ • Harmony-Nelder-Mead │ • Quantum Delta PSO │
│ • Harris Hawks (HHO) │ • Cultural Annealing │ • Chaos Levy Swarm │
│ • Firefly & Bat Swarm │ • Biogeography-Swarm │ • Dynamic Director │
└───────────────────────┴────────────────────────┴───────────────────────┘Director Dynamic Ensembles
The Director is an adaptive meta-controller that orchestrates sub-swarm fleets. It tracks swarm diversity, velocity stagnation, and fitness gradients, dynamically reallocating compute budgets and authority across competing optimizers during search execution.
Benchmark Runner Invocations
1. Matrix Sweep across All 35 Optimizers
dotnet run --project src/HFLabs/Products/HyperSwarmOptimizationSuite/BenchmarkRunner.csproj -c Release -- matrix-sweep2. Multi-Objective Pareto Benchmark
dotnet run --project src/HFLabs/Products/HyperSwarmOptimizationSuite/BenchmarkRunner.csproj -c Release -- mo-bench --objectives 3 --pop 2003. GPU Offload Verification on Intel Level Zero
dotnet run --project src/HFLabs/Products/HyperSwarmOptimizationSuite/BenchmarkRunner.csproj -c Release -- test-gpu --device-id 0Output Telemetry
Benchmark runs emit structured JSON summaries recording:
- Best Fitness & Convergence Iteration
- Hypervolume Indicator (for Multi-Objective Pareto fronts)
- Spatial Diversity Index & Velocity Entropy
- GPU Kernel Execution Time vs Host Marshaling Overhead