Maryland Team Solves Poisson Equations Using One Quantum Circuit

A new variational quantum linear solver (VQLS) solves the Poisson equation using an improved method for encoding information about mathematical operators into quantum circuits. Unlike existing VQLS approaches that require multiple distinct circuits to evaluate solutions, this technique needs only one circuit, simplifying calculations sharply. Simulations involving heat conduction and fluid dynamics reveal how classical optimizers impact optimisation progress during computation.

This new technique simplifies complex calculations by sharply reducing the number of steps needed compared with current approaches. BosonQ Psi Corp. and University of Maryland scientists achieved this using a variational quantum linear solver (VQLS), employing an efficient way to represent mathematical operations within quantum circuits. The team has developed a new approach to solving equations governing many physical processes; it tackles problems like heat transfer and fluid movement with greater efficiency than existing methods.

Their work centres around a VQLS, an algorithm designed to find approximate solutions to linear equations on a quantum computer, but streamlines calculations by requiring only one quantum circuit where others need multiple. This simplification is achieved through ‘block encoding’, representing data as instructions for the quantum processor much like converting into machine code for conventional computers.

Benchmarking their VQLS on simulations including heat conduction and fluid dynamics reveals how different classical optimizers affect computational progress; however, questions remain regarding its scalability and whether it can truly outperform established techniques, details of this new method follow.

Quantum algorithm reduces complexity of solving partial differential equations

The variational quantum linear solver (VQLS) sharply reduced required circuits, decreasing cost function evaluation from scaling with mathcalO(L²), where L represents terms in the Laplacian decomposition, down to needing only one circuit. This breakthrough surpasses limitations of previous methods that demanded numerous distinct calculations and prevented simulations requiring high precision or large problem sizes. Previously, tackling complex physical models necessitated exponentially increasing computational effort as grid resolution increased; this new approach circumvents that barrier by streamlining data processing within the quantum computer.

A significant benefit is its ability to handle problems previously intractable due to their immense computational demands. Benchmarking involved three distinct physical scenarios: simulating heat transfer from a localised source, modelling sinusoidal pressure distributions and analysing two-dimensional fluid dynamics within a lid-driven cavity. Simulations were run across grid resolutions ranging from 2×2 to 32×32 interior points, a total of ten different sizes up to 1024 unknown variables, allowing direct comparison against classically computed results obtained via the Successive Over-Relaxation method.

An exact block encoding technique represents mathematical operators as instructions for the quantum processor; this enables accurate solutions on physically motivated benchmarks. The team created an exact block encoding for their discrete Laplacian operator, meaning that mathematical operations are translated into instructions directly executable by quantum processors without introducing approximation errors during translation. This allows efficient tackling of the Poisson equation, a fundamental component in modelling phenomena like heat transfer and fluid movement, while also providing insights into how these methods scale with increasing problem size.

Single quantum circuit solves linear systems for advanced simulations

This new variational quantum linear solver (VQLS) offers a potential pathway towards simulating intricate physical systems, from predicting heat flow to modelling fluid dynamics, which currently strain even the most powerful supercomputers. Forum and ASCEND acknowledge a key dependency on classical optimisers; these algorithms guide the quantum computation but are notoriously sensitive to their initial settings and can easily stall before finding an optimal solution.

However, acknowledging that classical optimisation algorithms can hinder performance does not diminish this work’s significance, it highlights a vital area needing further attention within hybrid quantum-classical computation. Unlike earlier VQLS methods relying on Laplacian decompositions demanding increasing numbers of computational circuits as problem complexity grew, calculations with this new method are streamlined by converting data into instructions directly executable by a quantum processor, similar to how conventional computers translate into machine code.

The researchers developed a variational quantum linear solver capable of solving the Poisson equation using only one circuit for each cost evaluation. This contrasts with previous approaches which required more circuits and thus greater computational demand. The team demonstrated its effectiveness across three benchmarks including heat conduction and fluid flow problems, utilising an exact block encoding technique to represent mathematical operations for the quantum processor. Their work highlights that performance is affected by the classical optimiser used alongside the quantum computation, suggesting this area requires further investigation.

More like this

Quantum AlgorithmsStanford team confirms at Physical Superintelligence’s proof of quantum speedup limits

Quantum Research NewsQuantum algorithms sidestep readout limits with a new block-encoding method

Quantum AlgorithmsA quantum simulation method avoids extra qubits for observable estimation

Quantum Research NewsA new algorithm breaks down project scheduling into 30-qubit daily batches