Anglia Ruskin Research Online (ARRO)
Browse

An optimization of traditional CPU emulation techniques for execution on a quantum computer

Download (1.51 MB)
journal contribution
posted on 2024-12-20, 14:22 authored by James Fitzjohn, George Wilson, Domenico Vicinanza, Adrian Winckles

The use and adoption of quantum computers by the wider computing community is diminished by the need to adopt new programming techniques. These techniques involve moving from a high-level language where the programmer can define and manipulate objects, to a quantum model where the programmer defines and configures the circuits at a gate level. Previous work by the authors aimed to ease this transition through the use of a software development kit (Qx86 SDK) that emulates a traditional CPU for execution on a quantum computer, but only delivered a raw capability. The current work now presents a number of new methods that extends and improves the SDK's capability. These methods include optimizing traditional logic gate emulation, multiple gate simplification methods, reducing the number of required qubits and alternative optimized techniques for many CPU instructions. A quantum machine code mapping method is described that enhances the emulation of a traditional/quantum hybrid CPU prototype. While still orders of magnitude slower than the performance of a traditional CPU in terms of arithmetic, logic and bitwise operations, execution speed is shown to be markedly improved (in some cases by more than 1,000%) and without introducing any unrealistic requirements (that is, all execution can be performed utilizing less than 32 qubits). The usefulness of the SDK has now been enhanced as a reference guide, where the programmer/researcher can contrast traditional methods versus multiple quantum methods of execution.

History

Refereed

  • Yes

Volume

23

Issue number

10

Publication title

Quantum Information Processing

ISSN

1570-0755

Publisher

Springer Science and Business Media LLC

File version

  • Published version

Language

  • eng

Item sub-type

Journal Article

Affiliated with

  • School of Computing and Information Science Outputs

Usage metrics

    ARU Outputs

    Categories

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC