QBIQ: Quantum solver for hard BInary Quadratic problems
The aim is to improve the performance of the algorithms by upgrading the branch-and-bound (B&B) method, extending existing solutions to new cases, using quantum algorithms to determine bounds, and integrating quantum and classical approaches.
Duration: from 1.1.2025 to 31.12.2027
Project coordinator: Rudolfovo - Novo mesto Science and Technology Centre
Consortium partners: the Jožef Stefan Institute; the University of Ljubljana, Faculty of Mechanical Engineering; Xlab, d.o.o.
Project leader at {Rudolfovo: Prof. Dr. Janez Povh (janez.povh@rudolfovo.eu)
Project description:
The project focuses on the development of a powerful solver for binary quadratic optimization problems with quadratic bounds (QBO), which will run on modern supercomputing and quantum infrastructure. The aim is to improve the performance of the algorithms by building on the branch-and-bound (B&B) method, extending existing solutions to new cases, using quantum algorithms to determine bounds, and integrating quantum and classical approaches. The solver will be thoroughly tested on supercomputers, integrated with existing quantum platforms and tailored to XLAB's specific industry challenges related to cryptography and automation. The project has the potential to advance scientific and industrial progress in the field of quantum and optimisation computing.
Project value: EUR 400.000
Structure of the project team with links to SICRIS data:
Dr Janez Povh, SICRIS No: 22649
Dr Janez Žerovnik, SICRIS No: 03430
Dr Roman Kužel, SICRIS No: 60286
Dr Omkar Bihani, SICRIS No: 59360
Dr Leon Kos, SICRIS No 12725
Dr Tadej Kanduč, SICRIS No 32770
Dr Gregor Papa, SICRIS No 18291
Dr Anton Biasizzo, SICRIS No 11983
Dr Rok Žitko, SICRIS No 23567
Dr. Jaka Vodeb, SICRIS: https://cris.cobiss.net/ecris/si/sl/researcher/46682
Dr. Daniel Vladušič, SICRIS No. 21305
Dr Matej Artač, SICRIS No 21307
Dr Justin Činkelj, SICRIS No 26224
Dr Nejc Bat, SICRIS No. 51337
Project phases and description of their implementation:
O1: Develop a high-performance accurate solver for QBO that will outperform existing accurate solvers and operate efficiently on state-of-the-art supercomputers in the EU.
O2: Develop a hybrid quantum-classical exact solver for QBO that will offload the most complex computations to available quantum computers (QCs).
O3: Explore the possibility of directly solving the QBO problem exclusively with quantum computers and develop a conceptual prototype of a purely quantum solver for QBO.
O4: To combine the code of a high-performance solver, a hybrid solver and a quantum solver into a single solver, called QBIQ, which will be able to solve QBO instances exactly or approximately. QBIQ will have a user interface that will serve as a central entry point for QBO instance submission.
A5: Use the developed QBIQ solver to solve as many instances as possible of real-world problems of particular relevance to the XLAB industrial partner.
The project has the following work packages:
DP1: High Performance QBO Solver (leader: RUD, collaborator: UL FS).In this work package we will develop a high performance QBO solver based on the B&B algorithm that will run efficiently on state-of-the-art supercomputers.
DP2: Hybrid quantum-classical solver for QBO (PI: RUD, Co-PI: IJS) Inthis work package, which is organised into four tasks, we will develop a hybrid solver that uses quantum algorithms to determine tighter upper and lower bounds for the B&B algorithm. The treatment of the two bounds will be separate, so we will develop the quantum algorithms to determine them separately.
DP3: Quantum solver for QBO (PI: IJS, Co-PI: RUD).In this work package we will investigate how to solve QBO directly using only quantum computers.
WP4: Development of QBIQ (PI: UL FS, Co-PI: RUD).In this WP, we will combine the codes of the high-performance solver, the hybrid solver and the quantum solver from WP1, WP2 and WP3 into a single QBIQ solver. The work will be organised into two tasks starting in the middle of the second year.
WP5: Solving IFP and JSSP (leader: XLAB, participants: all).In this WP, we will use the QBIQ solver to solve the integer splitting problem (IFP) and the task scheduling problem (JSSP). The work will be organised in two tasks, starting at the beginning of the second year and finishing at the end of the project.
WP6: Project management, communication and dissemination of results (leader: RUD, participants: all):This work package will include the management and dissemination tasks necessary to ensure the success and impact of the project.
