04 · E · Quantum-inspired optimization
Quantum Optimization Lab
Select the best set of partnerships under a portfolio size constraint and conflict penalties, formulated as a QUBO. Quantum advantage is not assumed — every run is benchmarked against classical baselines.
maxx [ Σij sijxij − λ Σk Penaltyk(x) ]
x_ij ∈ {0,1} selects a partnership. Penalties enforce K partnerships and forbid conflicting pairs. Translatable to QUBO for annealers or QAOA.
Exact search (2¹² states)
3.80
✓ optimal
Simulated annealing
3.74
gap 0.06
Greedy baseline
3.80
✓ optimal
Annealing convergence (best objective)
Partnerships K5
Penalty λ1.5
Optimal portfolioexact
● Helix ↔ Tyndall0.83
○ Helix ↔ OptiParts0.42
○ Photon ↔ UCC0.67
○ Nova AI ↔ UCD0.44
● Atlas Bio ↔ U. Galway0.86
● Cobalt ↔ Precision Eng.0.69
● Strata Q ↔ UCD0.69
○ Lumen ↔ Tyndall0.5
○ Kinetic ↔ UL0.56
● Meridian ↔ CleanRoom0.73
○ Orbit ↔ TalentBridge0.52
○ Nova AI ↔ Strata Q0.45
Conflicts: 1×8, 4×7, 2×6, 7×12