Hardware-agnostic optimisation, benchmarked honestly
DOMINO-Q formulates resilience planning as a discrete combinatorial problem, solves it exactly on CPU today, and translates it into forms that future quantum or quantum-inspired backends could accept.
All 2^12 = 4096 portfolios evaluated; provably optimal for this proxy model (≤14 actions).
Seeded classical simulated annealing over feasible bit-flips. Runs in your browser on request.
No QPU is connected. No quantum execution has occurred and no quantum advantage is claimed.
1 · Graph cascade calculation
Local failure probability per asset and scenario:
pᵢ⁰ = min(0.99, baseᵢ × hazardₛ × Πₐ(1 − effectₐ,ᵢ))Then three propagation sweeps through dependency edges j→i of strength wⱼᵢ:
pᵢᵗ⁺¹ = 1 − (1 − pᵢ⁰) × Πⱼ (1 − wⱼᵢ × pⱼᵗ)Scenario loss in synthetic disruption points (not money):
Lₛ = Σᵢ impactᵢ × pᵢ³Objective with risk weight λ = and α = 0.8:
min E[L] + λ · CVaR₈₀(L)"Three sweeps" is a modelling iteration count, not a time step or measured hours.
2 · Solver benchmark on current inputs
Exact optimum vs baselines – all computed live
4 · Benchmark-based backend routing (planned)
Constrained binary model from graph + actions
n ≤ 14 → exact enumeration; larger → MILP / heuristics
Run candidate backends on held-out instances, record gap & time
Pick backend by measured evidence; re-score with exact evaluator
Quantum backends would be eligible for routing only after they demonstrate measured benefit on benchmarks. Currently: Quantum hardware: NOT CONFIGURED.