Quantum Computers Now Share the Load to Solve Complex Equations

Scientists Tong Shen, have developed a novel variational quantum algorithm designed to circumvent the limitations inherent in the capacity of individual quantum computers. The approach involves partitioning a large matrix across multiple noisy intermediate-scale quantum (NISQ) computers, thereby enabling a scalable solution where the effective system size grows linearly with the number of processors employed. Quantum simulations indicate a promising pathway towards solving linear systems that are currently intractable for single-quantum-computer architectures.

Distributed quantum algorithms unlock scalable linear equation solutions

Numerical quantum simulations can now solve linear systems scaling with the number of quantum computers, achieving a 2n × 2n matrix size. Previously, single quantum computers restricted solutions to considerably smaller dimensions. A new distributed variational quantum algorithm (VQA) developed by Tong Shen partitions large matrices across multiple noisy intermediate-scale quantum (NISQ) devices, coordinating solutions via classical communication between computers. This partitioning is crucial, as solving linear equations of the form Ax = b often requires computational resources proportional to the cube of the matrix dimension, making large systems particularly challenging.

This breakthrough addresses a fundamental barrier in quantum linear equation solvers, enabling calculations with system sizes unattainable by current single-processor architectures. Several quantum processors, when combined, bypass the qubit limitations of individual machines and pave the way for more complex simulations. The algorithm integrates a modified variational quantum linear solver on each computer with distributed classical optimisation, allowing for parallel evaluation of expectation values and reducing the computational burden on any single processor. The variational quantum linear solver (VQLS) itself typically involves preparing a trial quantum state parametrised by a set of variational parameters, then optimising these parameters to minimise a cost function related to the residual error of the linear system. Distributing this process requires careful consideration of how information is exchanged and aggregated across the network of quantum computers.

Valuable insights into designing effective distributed algorithms for complex problems are provided by the derivation of the quantum cost function. Although current simulations rely on idealised conditions, they do not yet demonstrate performance with real-world quantum hardware, where qubit coherence and gate errors remain significant obstacles to scaling beyond relatively small problem sizes. The success of this method is also contingent on the specific arrangement of quantum computers, requiring ‘row- and column-neighbour graphs’ for communication, which may prove limiting given current hardware constraints. Specifically, each NISQ computer only communicates with other quantum computers located in the same row and column of the block partition, creating a sparse communication network. This arrangement minimises communication overhead but necessitates a specific topology for the quantum computer network.

Distributed quantum computation overcomes limitations of single-processor systems

Countless scientific and engineering challenges, ranging from computational fluid dynamics and materials science to financial modelling and machine learning, depend on solving large linear equations. These equations frequently arise in the discretisation of partial differential equations, the analysis of complex networks, and the optimisation of high-dimensional parameters. This new distributed approach offers a potential route beyond the limitations of today’s quantum processors, demonstrating that distributing the computational load across multiple, smaller quantum processors is possible. It bypasses the need for a single, enormously powerful machine, offering a viable pathway towards tackling complex problems presently beyond reach. The ability to scale the solution size with the number of available quantum processors is particularly significant, as it suggests a path towards quantum advantage for problems that are currently intractable for classical computers.

Partitioning a large matrix A into smaller square block submatrices, each of which is known only to a single noisy intermediate-scale quantum (NISQ) device, allows the algorithm to overcome limitations imposed by the capacity of any single machine. This partitioning strategy effectively decomposes the original problem into a set of smaller, more manageable subproblems. Classical communication between these devices coordinates the solution process, enabling the system’s effective size to scale with the number of processors involved. This coordinated approach allows for parallel processing and reduces the computational demands on each individual unit. The classical optimisation component is responsible for coordinating the updates to the variational parameters across all the quantum processors, ensuring that the overall solution converges towards the correct answer. The algorithm’s performance is heavily influenced by the efficiency of this classical communication and optimisation loop.

Furthermore, the algorithm’s design is particularly relevant in the context of current quantum hardware limitations. Building a single quantum computer with a sufficient number of qubits to solve large-scale problems is a formidable engineering challenge. This distributed approach offers an alternative strategy, leveraging the collective power of multiple smaller quantum computers. While challenges remain in terms of maintaining coherence and minimising errors in a distributed system, the potential benefits in terms of scalability and computational power are substantial. Future research will likely focus on developing more robust communication protocols, error mitigation techniques, and optimisation algorithms tailored to the specific characteristics of distributed quantum computing architectures. The 2n × 2n scaling achieved in simulations represents a significant step towards realising the full potential of this approach, but further advancements are needed to translate these results into practical applications.

The researchers successfully developed a distributed quantum algorithm capable of solving large-scale linear equations by partitioning the problem across multiple noisy intermediate-scale quantum computers. This approach overcomes the limitations of single quantum computers by dividing a large matrix into smaller blocks, each processed by a separate device. The algorithm integrates quantum computation with classical optimisation, allowing the system’s size to scale with the number of processors used. Simulations demonstrated the algorithm could solve linear systems scaling to 2n × 2n, and the authors intend to focus on improving communication protocols and error mitigation in future work.