HybridCSP

Overview

Constraint satisfaction problems are widely used in scheduling, logistics, and planning systems. Classical solvers handle most instances efficiently, but hard instances trigger exponential search, causing unpredictable runtime and energy spikes. This project investigates whether a hybrid quantum–classical architecture can bound this tail risk. I developed HybridCSP, a solver combining AC-3 constraint propagation, MRV/LCV heuristics, Grover's quantum search algorithm, and a Random Forest ML partitioner, evaluated across 5,150 measured experiments in four CSP domains plus a financial collateral allocation application. Results show the hybrid solver increases median energy on easy instances due to quantum overhead, but eliminates catastrophic worst-case blowup — reducing maximum observed energy consumption by 293×. The quantum circuit activated on 32.3% of hybrid runs, confirming the architecture is applicable well beyond narrow worst-case regimes.

Awards (2)

• Bronze Medal
• Selected for CWSF 2026

Competition history

• CWSF 2026 Digital Technology

Related projects

Closest projects by meaning, across every fair and year in the corpus.