Accelerating Innovation Through Collaboration Fujitsu’s Quantum Simulator Challenge 2024

Fujitsu is one of the leaders both in developing tomorrows Quantum Computers and in developing todays Quantum simulators. Most recently Fujitsu launched a 40-qubit, CPU-based state vector quantum simulator. Built on a foundation of 1,024 FX700 nodes powered by A64FX processors, this system represents a leap in computational efficiency and scalability.

At the core of this innovation is Qulacs, one of the world’s fastest quantum simulation platforms, originally developed by Osaka University and QunaSys Corporation. Fujitsu has enhanced Qulacs to enable parallel execution and distributed processing across its cluster system. By leveraging Message Passing Interface (MPI) technology and optimizing memory bandwidth with Scalable Vector Extension (SVE) operations, Fujitsu has significantly improved performance and scalability.

Recognizing the importance of a robust development ecosystem, Fujitsu has also introduced a software development kit (SDK) with partial compatibility with Qiskit, a widely used quantum computing framework. This move lowered the barriers for developers, enabling them to seamlessly build and test quantum applications within a familiar environment.

Fujitsu's latest version of the quantum simulator features two new key technological advancements: VQE acceleration and QDD technology.

This innovative technology speeds up quantum-classical hybrid algorithms running on a quantum simulator by up to a factor of 200. This dramatically reduces processing time, unlocking new potential for large-scale simulations in a range of quantum chemistry and materials science.

The key to this innovation is the simultaneous processing of numerous quantum circuit computations across multiple groups. Traditionally, the computational amount of quantum-classical hybrid algorithm simulation increases exponentially with problem complexity, creating bottlenecks in solving large-scale challenges. Fujitsu’s approach can potentially help reduce the computational time linearly along with the number of computational nodes.,. This technology combines node-parallel computing technology using Message Passing Interface (MPI) and distributed processing technology, as well as technology that reduces Hamiltonian terms while suppressing accuracy degradation. It achieves 32QubitVQE calculations 200 times faster than conventional methods.

QDD implements large-scale, high-speed quantum simulations using graph structures. QDD has the advantage of running FTQC algorithms and can run Shor and Grover algorithms on a scale that causes memory failures in explicit state vector-based simulators. (References: S. Li et al., "Parallelizing Quantum Simulation with Decision Diagrams,".

As many of you know Quantum simulations are modeled using vectors and matrices in a Hilbert space, which grows exponentially with the number of qubits. Traditional simulation methods using statevectors are memory intensive. Decision diagrams, such as Binary Decision Diagrams (BDDs) and their variants like QMDDs, provide a more efficient representation by exploiting redundancy.

Fujitsu Research has discovered that ring communication is effective for decision diagram-based simulator in a multi-node HPC environment. The proposed strategies significantly enhance simulation speed, making them a viable alternative to statevector-based methods.

Building on the success of last year Fujitsu Quantum Simulator Challenge this year’s competition again attracted a wide range of entrants from both indicatory, academia keen to test the capabilities of the latest Fujitsu Quantum Simulator technology.

The quality of entrant’s submissions was extremely high, but after careful deliberation, the judges found submission from the Delft University of Technology team in the Netherlands to be the overall winners of the second Quantum Simulator Challenge in a close competition. With their innovative quantum algorithm for Industrial Shift Scheduling.

In second place came Technische Universität Ilmenau from Germany with their innovative quantum algorithm to accelerate Particle Image Velocimetry (PIV) and in third place came a team form QunaSys Inc in Japan with their large-scale simulation of quantum phase estimation algorithms.

We would again like to extend our heartfelt thanks to all participants for their remarkable contributions and active participation in the Quantum Simulator challenge and contribution to improving Quantum simulations . We eagerly anticipate future collaborations and continued engagement with you all.

Technische Universität Ilmenau (Germany)
QuPIV - Quantum cross-correlation for Particle Image Velocimetry

This paper proposes a quantum algorithm to accelerate Particle Image Velocimetry (PIV), a technique used to measure fluid flow velocities. PIV currently relies on computationally expensive cross-correlation calculations. The paper detailed improvements to quantum amplitude amplification and efficient state preparation for sparse images. The algorithm's applicability is demonstrated through its use in PIV, highlighting its potential impact on various fields including wind tunnel testing and turbine design. The broader business applicability stems from the fact that cross-correlation is a fundamental operation in many signal processing applications, from AI and image processing to telecommunications.

QunaSys Inc, (Japan)
Large-scale simulation of quantum phase estimation algorithms

QunaSys performed large-scale simulations of quantum phase estimation (QPE) algorithms, a key method for quantum chemistry calculations. They simulated both qubitization and Trotter QPE methods. The qubitization method, while requiring fewer gates, demands many ancilla qubits; their simulation on the H2 molecule used up to 39 qubits, a significant achievement. The Trotter method simulation, applied to a truncated 32-qubit Hamiltonian of C2H6, reached up to 38 qubits, exceeding the capabilities of classical Full-CI calculations on standard laptops. These simulations represent, the largest-scale simulations of their kind, for each QPE method, demonstrating progress towards practical quantum chemistry applications.

Quantum research is advancing at an unprecedented pace, and with it, the capabilities of quantum simulations to solve real-world challenges are rapidly expanding. For organizations looking to gain a competitive edge, the moment to explore quantum technology is now. Fujitsu stands at the forefront of this revolution, offering advanced quantum simulations that can drive transformative benefits for your business today.

These sectors are also the most vulnerable to disruption if they wait for fully functional quantum computers before exploring quantum-based solutions. Early adopters will lead their industries, while those who delay risk falling behind.

Quantum computers will not replace classical systems; rather, they will complement them by tackling complex problems that are beyond the reach of traditional computing. Harnessing the power of quantum technology has the potential to unlock value in five key areas:

1.Advanced data analysis

Quantum computing enables faster and more sophisticated data analysis, allowing organizations to process massive datasets in real-time. This leads to sharper insights and more agile decision-making. For example, quantum algorithms can uncover hidden patterns in customer behavior, driving more effective product development and marketing strategies.

Quantum technology is reshaping industries and creating unprecedented opportunities for innovation and growth. Organizations can already access many of these advantages today using Fujitsu’s advanced quantum simulator.

The pace of development is accelerating. For forward-thinking leaders who want to gain a competitive advantage from early quantum adoption, the time to act is now. Connect with Fujitsu to discover how our quantum simulations can solve your challenges today and how our quantum computers will transform your business tomorrow.

Why not talk to Fujitsu now to see how we can help your organization?