Block Encoding
Block-encoding operators are an essential approach in quantum signal processing, and a team of researchers from Fraunhofer IAO and Universität Stuttgart has revealed a resource-efficient way to create them. This development makes it possible to create larger, more intricate quantum algorithms and significantly lowers the computing complexity usually involved in creating these encodings. For a large variety of input matrices, the novel method assembles these fundamental quantum elements with almost optimal resource requirements, obtaining a parameter count that is close to the number of free parameters in those matrices. Optimisation for quantum systems with up to eight qubits is made possible by this important advancement.
You can also read Columbia Researchers Introduce Quantum HyperQ for Multi-User
Quantum signal processing (QSP) algorithms, which have become a dominant force in the development of quantum computer techniques, rely heavily on block-encoding. It enables the representation of non-unitary input matrices as sub-blocks of larger unitary operators that can be carried out on quantum computers, such as those modelling Hamiltonians in physics and quantum chemistry.
Even though conventional techniques like oracle query models and Linear Combinations of Unitarizes (LCU) can produce quantum advantages, they frequently have significant drawbacks for near-term quantum computers, such as high ancilla overheads, a high number of multi-controlled gates, and unfavourable scaling when working with dense and unstructured matrices. The total quantum computing cost of the entire quantum procedure is mostly determined by the gate complexity of these block-encoding operators.
You can also read European Quantum Industry Consortium Growth & Collaboration
The group has developed a technique known as Variational Block-Encoding (VBE) to overcome these obstacles. VBE promises significant advancements for quantum calculations on hardware with limited resources by utilising the enormous expressibility of hardware-efficient parameterised quantum circuits (PQCs) to encode matrices effectively.
This study’s main accomplishment is showing that VBE can do precise encoding with just one ancilla qubit, which is remarkable considering that conventional techniques sometimes require more. This is consistent with theoretical lower limitations on the number of circuit parameters, which are related to the target matrix’s degrees of freedom.
Adapting the circuit design to take into account the input matrix’s intrinsic characteristics such as whether it includes real or Hermitian values or displays particular symmetries is an important part of this study. Researchers can further minimise the number of encoding parameters required and improve performance by directly integrating these symmetries into the circuit architecture. Because it closely correlates with the quantum resources qubits and operations needed for the computation, this parameter minimisation is essential. For example, resource costs are greatly decreased by limiting circuits to real-valued or Hermitian targets.
You can also read Liquid Helium & Electron Interaction: Key to Charge Qubits
In order to construct realistic quantum computers that are less prone to errors, the research examined the expressibility and complexity of quantum circuits, concentrating on creating circuits that are both powerful enough to represent a given matrix and as simple as possible. The generators of these circuits are analysed using mathematical tools like Lie algebra and the Derivative Lie Algebra to comprehend their capabilities and ascertain the range of activities they can carry out.
The impressibility of the circuit is measured by the size of the basis set made up of these generators. The study discovered a correlation between the algebraic structure of the circuit generators and the complexity of these symmetry-restricted circuits, providing information for future encoding designs that are even more effective.
With the number of free parameters in the quantum circuit roughly matching the lower bound of independent parameters for the target matrix, numerical studies empirically show that VBE enables efficient encodings of dense input matrices. It was discovered that the optimisation landscapes in VBE were smooth, allowing for effective convergence using common classical optimisation methods such as the BFGS optimiser. In the overparameterized environment, where a global minimum can be attained from any point, the process is surprisingly robust.
You can also read UbiQD First Solar Partner To Use Quantum Dots In PV Panels
In comparison to current block-encoding techniques, VBE provides a significant decrease in resource overhead when benchmarked against known approaches. For example, VBE reduces the 2-qubit gate counts, a crucial quantum resource measure, by more than an order of magnitude when compared to LCU for systems up to five sites in the particular situation of Heisenberg Hamiltonians.
This benefit is applicable to systems with up to eight locations for permutation invariant circuits. The reason why 2-qubit gate counts for LCU could seem smaller for bigger systems is because the number of LCU terms increases polynomially, while VBE circuit sizes for the majority of ansatzes aside from the permutation invariant ansatz increase exponentially.
Notwithstanding these remarkable improvements, VBE’s main drawback at the moment is the substantial amount of classical computation needed to tune the variational parameters, which limits its use to systems with a maximum of eight qubits. The enormous dimensionality of optimization landscapes and the computational expense of matrix calculations both increase exponentially with system size.
According to the researchers, combining VBE with LCU (linear combination of unitarizes) is a viable near-term use case. Smaller matrix blocks could be effectively encoded using VBE, and the entire matrix could then be built using linear combinations, which would lower the overall resource needs.
Future studies will look into linkages to multivariate quantum signal processing and how other system-specific characteristics can further lower circuit resource requirements. These studies could result in techniques for figuring out circuit parameters without requiring complete optimization. Additionally, the method offers opportunities to enhance quantum machine learning applications and variational quantum eigensolvers.
You can also read QuiX Quantum invest €15M for single-photon Quantum computer
