Quantum Calculations Become Far Simpler with New Circuit Design

Researchers at University College London, in collaboration with University of Amsterdame, have developed a new method for preparing quantum states and verifying circuits utilising tree tensor network compilation. This innovative approach decomposes matrix product states into quantum circuits exhibiting logarithmic depth, a crucial advancement for practical implementation on near-term quantum hardware. The technique facilitates a controlled trade-off between fidelity and circuit depth, potentially unlocking more complex quantum algorithms such as quantum phase estimation and quantum-selected configuration interaction. Furthermore, the team’s work extends to matrix product operators, enabling the creation of circuits for calculating overlaps and offering a novel interpretation as verifier circuits for device calibration and performance assessment.

Logarithmic scaling of quantum circuits unlocks complex algorithm implementation

Circuit depths required for loading matrix product states have been significantly reduced to O (log N) from the previously standard O(N) for N qubits, representing a substantial leap in computational efficiency. This improvement addresses a critical limitation that previously hindered the practical implementation of sophisticated quantum algorithms, including quantum phase estimation and quantum-selected configuration interaction, on currently available near-term quantum hardware. Quantum phase estimation, a cornerstone algorithm for determining the eigenvalues of unitary operators, is vital for applications in quantum chemistry and materials science. Quantum-selected configuration interaction, a method for approximating the ground state of many-body quantum systems, promises improvements in molecular simulations. The new method leverages tree tensor network renormalisation, a systematic technique for simplifying the representation of quantum states. This simplification allows the creation of circuits with logarithmic scaling, enabling a tunable balance between fidelity and circuit depth, crucial for optimisation on devices constrained by limited qubit counts and coherence times. The logarithmic scaling implies that as the number of qubits increases, the circuit depth grows much more slowly, making it feasible to implement larger and more complex quantum computations.

The technique also extends beyond simple state preparation to encompass matrix product operators. This allows for the construction of quantum circuits capable of calculating the overlap between different quantum states, a fundamental operation in many quantum algorithms. Importantly, this capability provides a novel approach to verifying the calibration of quantum circuits. Accurate calibration is paramount for reliable quantum computation, as even small errors in gate operations can accumulate and significantly degrade the final result. By constructing circuits that measure the overlap between a known, ideal state and the state produced by a quantum device, researchers can assess the device’s performance and identify sources of error. In particular, the controllable trade-off between circuit depth and fidelity allows for optimisation tailored to the specific characteristics of near-term quantum devices. Fidelity, in this context, refers to the accuracy with which the quantum state is prepared or manipulated. While these logarithmic scalings currently assume ideal conditions, they do not yet fully account for the substantial overhead introduced by real-world transpilation and gate errors inherent in existing hardware. Transpilation is the process of mapping a logical quantum circuit to the physical constraints of a specific quantum device. Tensor networks, graphical representations of quantum systems utilising interconnected tensors or multidimensional arrays, provide a nuanced balance between computational cost and result accuracy, allowing for efficient representation of complex quantum states.

Balancing accuracy and complexity in near-term quantum state preparation

Efficiently loading quantum states onto a quantum computer is vital for realising the transformative potential of quantum computation in diverse fields such as materials science, drug discovery, and financial modelling. Previously, circuit depth scaled linearly with the number of qubits, rapidly becoming impractical for even moderately sized systems. This simplification, achieved through tree tensor network decomposition, bypasses a significant hurdle for implementing complex quantum algorithms on current, limited hardware. Reducing the complexity of quantum circuits is paramount for near-term quantum computers, and even a slightly imperfect result obtained from a feasible circuit is often more valuable than a perfectly accurate result that remains computationally unattainable. A method for constructing quantum circuits with logarithmic depth from matrix product states, a compact and efficient representation of quantum systems, is now available. Matrix product states are particularly well-suited for representing ground states of one-dimensional quantum systems, but their applicability extends to higher dimensions with appropriate approximations.

Achieving this simplification not only enables the efficient preparation of quantum states but also allows for calculating overlaps between them and creates circuits interpretable as verifier circuits for calibrating quantum devices, offering a pathway to improved hardware reliability and performance. The verifier circuits function by comparing the output of a known quantum state with the output of the device under test, providing a quantitative measure of its accuracy. This is particularly important in the context of noisy intermediate-scale quantum (NISQ) devices, where errors are prevalent. The logarithmic depth scaling achieved by this method represents a significant step towards overcoming the limitations of NISQ hardware and paving the way for more complex and impactful quantum computations. Further research will focus on mitigating the effects of real-world noise and optimising the method for specific quantum architectures, ultimately aiming to unlock the full potential of quantum computation for solving challenging scientific and technological problems. The ability to efficiently represent and manipulate quantum states with reduced circuit complexity is a key enabler for advancing the field of quantum information science.

The researchers successfully decomposed matrix product states into quantum circuits with logarithmic depth, a significant reduction in computational complexity. This matters because it allows for the efficient preparation of quantum states on near-term hardware, where circuit depth is a limiting factor. The method also enables the calculation of overlaps between states and creates circuits that function as verifier circuits for quantum device calibration, improving hardware reliability. Future work intends to address real-world noise and optimise the method for specific quantum architectures.