Quantum Algorithms Now Scale Optimally with CI-Matrix Size

A new selected configuration interaction (QSCI) algorithm, formulated within the CI-matrix (CIM) framework, offers improved qubit scaling for simulating molecular systems. Vincent Graves and colleagues at the National Quantum Computing Centre, RAL, Oxfordshire, United Kingdom, present a single-bit flip error mitigation technique combined with a stochastic approximate Trotterization evolution, achieving comparable accuracy to sample-based diagonalization (SQD) methods when applied to benchmark molecules. They further augmented QSCI with a selected heat-bath CI (QSHCI) variant to attain performance approaching that of classical heat-bath CI, representing a key step towards more effective quantum simulations despite a current preprocessing cost.

Reduced qubit scaling enables efficient molecular quantum simulations

A six-fold reduction in qubit scaling for quantum simulations of molecular systems has been achieved, moving from a requirement proportional to the size of the Fock space to ⌈log2(NCSF )⌉, where NCSF is the size of the CI-matrix. This breakthrough overcomes a fundamental limitation of previous quantum selected configuration interaction (QSCI) and sample-based quantum diagonalization (SQD) algorithms, constrained by inefficient qubit usage when formulated in Fock space. Previously, simulating even moderately sized molecules demanded an impractical number of qubits, hindering progress in areas like drug discovery and materials science. The Fock space representation expands exponentially with the number of electrons and orbitals, quickly exceeding the capabilities of near-term quantum devices. The CI-matrix, however, provides a more compact representation by focusing on the most important electronic configurations, significantly reducing the required Hilbert space dimension. This optimisation is crucial because the number of qubits needed for a quantum simulation is directly related to the size of the Hilbert space being explored.

The new CI-matrix (CIM)-QSCI algorithm delivers comparable accuracy to existing methods while sharply reducing the quantum resources needed for calculations. Nitrogen (N2) and naphthalene molecules were simulated on quantum hardware to demonstrate the effectiveness of the CIM-QSCI algorithm, achieving accuracy levels comparable to existing SQD methods. The choice of N2 and naphthalene serves as a good test case, representing a simple diatomic molecule and a larger polycyclic aromatic hydrocarbon, respectively, allowing for validation across different molecular complexities. Incorporating a single-bit flip error mitigation technique added only one qubit of overhead to the system, addressing inherent noise in current quantum computers. This error mitigation strategy is particularly important in the NISQ (Noisy Intermediate-Scale Quantum) era, where quantum computers are susceptible to various sources of noise that can corrupt the computation. Single-bit flip errors are a common type of error, and this technique effectively reduces their impact on the final result. Further refinement led to the development of quantum selected heat-bath CI (QSHCI). This variant matched the performance of classical heat-bath CI (HCI) by replacing classical sampling with quantum sampling from QSCI. Classical heat-bath CI is a widely used method in quantum chemistry for calculating excited states and other molecular properties. By achieving comparable performance with QSHCI, quantum computers have the potential to tackle problems currently solved using classical methods. Despite this advancement, constructing the CI-matrix and performing the necessary Pauli decomposition currently requires a preprocessing cost proportional to N squared log N, limiting scalability. This preprocessing step involves identifying the most important configurations and transforming the Hamiltonian into a form suitable for quantum computation, and its computational complexity currently poses a bottleneck for larger systems.

Reducing qubit requirements for complex molecular modelling

Quantum computers are edging closer to simulating molecular behaviour, a feat beyond the reach of even the most powerful supercomputers. Accurate molecular simulations are essential for understanding chemical reactions, designing new materials, and developing novel drugs. Classical computational methods, while powerful, struggle with the exponential complexity of solving the Schrödinger equation for many-body systems. The latest work presents a new quantum algorithm that cleverly organises calculations using a ‘CI-matrix’ framework, reducing the number of qubits needed to model complex molecules. This is a key step towards practical quantum chemistry, potentially revolutionising the field by enabling the simulation of systems previously intractable. The algorithm leverages the principles of configuration interaction, a cornerstone of quantum chemistry, but implements it in a way that is more amenable to quantum computation.

Comparable accuracy to established classical techniques with fewer quantum resources represents a major advance, opening the door to modelling molecules previously inaccessible to quantum simulation. This reduction in qubit requirements is particularly significant given the limited number of qubits available on current and near-future quantum devices. Efficient data handling refinements should unlock even greater potential for practical quantum chemistry applications. The ability to simulate larger and more complex molecules will accelerate the discovery of new materials with tailored properties and the design of more effective pharmaceuticals. This new approach combines quantum and classical techniques to model molecules, offering a pathway to more efficient simulations. The interplay between quantum computation for the core simulation and classical computation for preprocessing and data analysis is a promising strategy for harnessing the strengths of both paradigms. By shifting away from traditional methods operating in Fock space, a disordered representation of electron arrangements, the algorithm achieves optimal qubit scaling, vital for tackling larger molecules. The Fock space, while conceptually simple, leads to an exponential increase in the number of basis functions required to represent the electronic wavefunction. The QSCI algorithm, combined with the heat-bath variant, delivers accuracy matching established classical techniques while potentially reducing the quantum resources needed, although the initial processing requirement remains a known limitation. Future research will focus on developing more efficient algorithms for constructing the CI-matrix and reducing the preprocessing cost, paving the way for simulations of even larger and more complex molecular systems. Addressing this preprocessing bottleneck is crucial for realising the full potential of this new approach.