QuBRA Benchmarking Framework CI

qubrabench is a Python library that aims to enable domain experts to estimate quantum speedups provided by future quantum computers for realistic use cases and data sets, in advance of large-scale quantum computers being available.

To this end, the library provides classical implementations of algorithms and subroutines (e.g., search or max) that can be accelerated on a quantum computer. These implementations are instrumented in such a way that when they are run, they track not only classical cost metrics, but also quantum cost metrics, i.e., costs that would be incurred if one were running the same algorithm with the same data on a quantum computer. These collected cost metrics can then be compared to obtain insight into when a quantum advantage might be achieved for a particular problem size or data set. This approach builds on the pioneering work of Cade et al (2022).

A long-term goal of this project is to also allow the execution of subroutines on quantum backends.

Installation

To install and use this project, download or check out this repository, and install it with the package manager pip (using a virtual environment is advised):

pip install .

We recommend the use of Python 3.10 or 3.11.

Usage

The qubrabench library offers subroutines and algorithms, which record classical and quantum cost metrics that can then be evaluated. For example, consider the following code which iterates over a list of users to find one with a particular name:

from dataclasses import dataclass

@dataclass
class User:
    name: str
    
def find_shor(users: list[User]):
    for user in users:
        if user.name == "Peter Shor":
            return user

We can equivalently write this using the search function in qubrabench:

import qubrabench as qb

def is_shor(user: User):
    return user.name == "Peter Shor"

def find_shor(users: list[User]):
    return qb.search(users, key=is_shor, max_fail_probability=1e-5)

To determine whether a quantum search algorithm can provide any advantage over classically iterating over the list, we add some minimal annotations to tell qubrabench which object we are interested in counting query stats for:

@qb.oracle
def is_shor(user: User):
    return user.name == "Peter Shor"

maybe_shor = find_shor([User("Lov Grover"), User("Peter Shor")])
print(maybe_shor)
print(is_shor.get_stats())

To further familiarize yourself with this approach and workflow, we recommend looking at the examples.

Contributing & Development

Please see the Development Guide for information on how to contribute to this project.

This work has been supported by the German Ministry for Education and Research (BMBF) through project “Quantum Methods and Benchmarks for Resource Allocation” (QuBRA).

API

Index