Ms Sreedevi Bhaskaran Potty Thulasi Lakshmi1
1Deggendorf institute of technology, Deggendorf, Germany
Biography:
Sreedevi B T holds a Master’s degree in Physics and is currently pursuing a second MSc in Quantum Computing and High-Performance Computing at Deggendorf Institute of Technology, Germany. Her research bridges hybrid Quantum-Inspired Optimization techniques, High-Energy and Particle Physics, and Computational Fluid Dynamics. As CTO of a startup, she is currently developing a Copilot focused on AI- and quantum-driven models for hydrogen energy systems. Sreedevi was awarded Best Poster at the FuT3ch Symposium conducted by the Institute For Future Technologies at the Deggendorf Institute of Technology. Sreedevi is passionate about applying advanced computing to solve real-world scientific and industrial challenges.
Abstract:
Efficient resource allocation is a critical challenge in research computing and applied industrial systems. This project focuses on solving the classical Knapsack Problem using quantum-inspired techniques applicable to real-world optimization tasks. The objective is to address a real-world logistics challenge known as the Lock Filling Problem, which resembles a constrained version of the 0-1 Knapsack Problem, where the goal is to optimally select and pack ships into a lock without exceeding its capacity.
The problem is modelled using Quadratic Unconstrained Binary Optimization (QUBO) and solved using the Quantum Approximate Optimization Algorithm (QAOA), a leading algorithm in near-term quantum computing. The model was implemented using myQLM, a quantum programming framework, and encoded using AQASM (Atos Quantum Assembly), enabling compatibility with quantum-inspired solvers and emulators. We applied the Layered Recursive Quantum Approximate Optimization Algorithm (LR-QAOA) to iteratively explore and refine solutions. Using QAOA, we evaluated the algorithm’s effectiveness in navigating complex solution spaces under realistic constraints.
Our results demonstrate that QUBO-based formulations can effectively model discrete optimization problems like multi-knapsack with competitive solution quality. The QAOA-based approach showed promise, particularly in identifying near-optimal solutions in reduced time compared to classical exhaustive methods.
This work highlights the potential of quantum-inspired optimization methods in addressing computational bottlenecks in resource planning, scheduling, and energy systems. It also contributes a practical use case for integrating hybrid models in emerging research infrastructure. The method is extensible to multi-objective and multi-knapsack variants, making it relevant for broader eResearch applications.