Quantum Computing News

The launch of the Quantum Optimization Benchmarking Library (QOBLIB) aims to speed up the hunt for quantum advantage.

The Quantum Optimization Working Group has released the Quantum Optimization Benchmarking Library (QOBLIB), a major step towards comprehending and attaining quantum advantage in the difficult subject of combinatorial optimization. The “intractable decathlon,” a set of 10 carefully chosen problem classes, is presented in this new open-source repository and the paper that goes with it. Its purpose is to provide a demanding environment for testing both quantum and classical optimization techniques. In order to speed up development and pinpoint areas where quantum computers can perform better than their classical counterparts for real-world issues, the program seeks to encourage cooperation between researchers and practitioners.

The goal of the mathematical discipline of combinatorial optimization is to identify the best option among a limited number of options. These methods are essential for resolving several worthwhile issues in a variety of fields of science and industry. Nonetheless, it is still quite challenging to find the best solution for a lot of real-world optimization situations. Thus, many popular quantum and classical optimization approaches are heuristics, which are procedures that use intuition-based “rules of thumb” to effectively solve complex problems.

Also Read about Pasqal’s Neutral-Atom QPU On Google Cloud Marketplace

Due to their inherent inability to guarantee performance in advance, heuristic algorithms make it challenging to forecast which issues they will be able to tackle successfully. Finding examples of quantum advantage necessitates considerable testing on real quantum hardware employing difficult, practically relevant issues, even if these systems are frequently able to produce answers that are “good enough” for real-world use cases.

A group of researchers from more than a dozen member organizations, including the Zuse Institute Berlin, Technische Universität Berlin, Purdue University, National University of Singapore, E.ON Digital Technology GmbH, Kipu Quantum GmbH, Forschungszentrum Jülich, University of Southern California, and IBM Quantum, recognized the need for a centralized resource and a collaborative environment, and they created QOBLIB.

It is impractical for any one researcher or organisation to carry out this thorough testing alone due to the sheer amount and complexity of optimization problems and solution approaches. Utilizing the combined knowledge of the working group and the larger optimization research community is the goal of QOBLIB.

Benchmarking Function in the Pursuit of Advantage

In computing, benchmarking typically has several uses, including as assessing how well a fixed algorithm performs on a particular system, figuring out ways to enhance algorithms and comprehend their scalability, or determining if a quantum or conventional approach is preferable for a given application.

The illustrate various forms of benchmarking, including applications, algorithms, and systems benchmarking. Applications benchmarking needs to be model-independent, although system and algorithm benchmarking can still be useful even if it is model-dependent and specialised to a certain model or approach. The most important factor in the quest for quantum advantage is thought to be applications benchmarking.

In the end, proving an advantage necessitates benchmarking that takes into account every possible method for defining and resolving an issue.

Previous optimization benchmarking work has mostly focused on system and method evaluation and has frequently been model-dependent. By offering model-independent benchmarks tailored for optimization applications, the new QOBLIB article and library aims to close this gap. The enormous diversity and complexity of optimization problem classes and their solution techniques make it difficult to create these model-agnostic benchmarks, but it is thought to be an essential obstacle to be overcome in the pursuit of quantum advantage.

Quantum Advantage

State that two essential requirements must be met by claims of quantum advantage in any application domain:

• All known classical approaches must have truly challenging challenges. There is no real use for quantum computers if classical computers can produce solutions that are good enough and reasonably priced.

• The capacity of quantum hardware and algorithms to solve the problem more correctly, efficiently, or economically than all known classical solutions must be proven.

Also Read about Utimaco Quantum Protect Uses Post-Quantum Cryptography

It is usually very difficult to find optimization problems that satisfy both of these requirements. Unsolved optimization problems that are both practically and scientifically relevant are rarely disclosed; instead, academics tend to simplify these problems in order to find “good enough” answers for certain use cases. However, the quality of the solution is sometimes sacrificed in the name of simplification. The practical impact of optimization could be greatly increased by being able to address more complicated issues with fewer simplifications.

The Intractable Decathlon is now available

The “intractable decathlon” a grouping of ten issue classes is the focal point of the new project. According to the working group, this collection is the first collection of optimization problems that are both scientifically and practically intriguing and that, even at relatively small problem sizes, become challenging for the most advanced classical solvers. These issues were also chosen because they are appropriate for investigation on near-term quantum devices, which still have qubit count and circuit depth restrictions despite advancements in technology.

Even while these particular challenges might not be the ones that ultimately yield quantum advantage in combinatorial optimization, they do offer a clear indication of possible areas for quantum advantage. The project offers precise, well-defined measures to make it easier to find an advantage and allow for equitable comparisons of all kinds of quantum and classical approaches.

To assist researchers in getting started and comparing performance, each problem class in the decathlon is provided with background data, a formal problem formulation, descriptions of particular problem instances that are available in QOBLIB, and traditional baseline findings. For certain issues, quantum baseline findings are also shown.

The QOBLIB Repository: A Platform for Collaboration

The Quantum Optimization Benchmarking Library is set up as a publicly available, open-source database. In order to cover typically tough situations, it includes problem instances for every problem class, varying in size and complexity. This makes it possible for researchers to monitor hardware and algorithmic developments leading to quantum advantage.

QOBLIB provides a submission template with explicit metrics to guarantee fair comparisons. The quality of the answer obtained, the overall amount of wall clock time, and the total amount of computational resources used both classical and quantum are some examples of these measures.
Reference models are also available in the repository. These comprise quadratic unconstrained binary optimization (QUBO) formulas as well as mixed-integer programming (MIP). For classical researchers, MIP frequently acts as an entry point, whereas QUBO does the same for quantum researchers. These models enable researchers to examine the performance of quantum algorithms in conjunction with the classical baseline data.

It is stressed that neither MIP nor QUBO are optimal; rather, they are just samples of how issues can be formulated. They are offered as places for researchers to start their investigation. Model complexity may rise as a result of mapping MIP to QUBO formulations, which may result in an increase in the number of variables, problem density, and coefficient ranges. Scholars are urged to draw inspiration from these in order to create whole new formulations that may be more appropriate for quantum processing or even allow for superior classical solutions.

Also Read About What are photonic qubits in quantum computing?

An Appeal for Action

Quantum advantage is thought to have very serious promise in optimisation. The unsolvable decathlon is a significant advancement, but achieving its full potential calls for cooperation. No one entity can finish this voyage by itself due to the enormous amount of challenges and algorithms to investigate.

This community effort specifically calls for and encourages participation from researchers and practitioners with expertise in quantum and classical optimization techniques. Researchers can directly contribute to a project that may result in the first demonstrations of quantum advantage by evaluating performance, testing new and existing algorithms against the QOBLIB challenges, and uploading results to the repository. It is also essential to continuously build new and enhanced models and algorithms.

In order to solve important problems that are now beyond the scope of classical methods alone, the goal is for everyone to work together to propel mathematical optimization into a new era of computation.