From today, Toshiba Digital Solutions Corporation (headquarters: Kawasaki City, Kanagawa Prefecture; President: Shunsuke Okada; hereinafter, “the Company”) launched the quantum-inspired1
optimization solution SQBM+, which is systematized as a solution based on the Simulated Bifurcation Machine (hereinafter, “SBM”)2
that is a combinatorial optimization solver utilizing the Simulated Bifurcation Algorithm (hereinafter, “SB Algorithm”) developed by Toshiba Corporation.
We offer a lineup of optimization solvers according to the intended application, and have adopted a new SB Algorithm3
that greatly improves speed, accuracy, and scale. Through this solution, we collaborate and co-create with partners who have expertise in each field; for example, selection of the optimal candidate substance for the urgent theme of COVID-19 therapeutics, and application to the creation of optimal work shifts for healthcare professionals. In this way, we will contribute to solving complicated social issues in various fields such as finance, drug discovery, genetic engineering, logistics, and AI.
For many social and industrial challenges, combinatorial optimization is essential for selecting the optimal items from an enormous range of choices; for example, optimizing financial transactions4, the movement of industrial robots, travel and transmission routes, and molecular design for drug discovery. When using existing computers, combinatorial optimization is difficult to solve at high speeds because the number of combination patterns increases exponentially as the scale of the problem grows. For this reason, specialized combinatorial optimization computers are being actively developed in Japan and overseas.
Toshiba Group is verifying the effectiveness of SBM. For example, we are starting joint verification for the effectiveness of applying pseudo-quantum computers to high-speed and high-frequency trading in the stock market5. This solution uses existing computers and utilizes SBM, which enables high-precision approximate solutions (good solutions) for complex and large-scale problems to be obtained in a short time. Additionally, the PoC (Proof-of-Concept) version of SBM was released to the world on the AWS Marketplace operated by Amazon Web Services Inc.6 We have been searching for real problems in order to solve social issues in a variety of fields by cooperating with research institutions such as universities, as well as with corporations aiming to solve social issues as actual problems in combinatorial optimization.
We will now begin providing systematized SQBM+ as a solution by utilizing the results of business co-creation efforts aimed at developing new markets and creating new solutions by using SBM technology7
, as well as the knowledge gained through various verification experiments in Japan and overseas.
SQBM is an acronym for “Simulated Quantum-inspired Bifurcation Machine,” which means that it is derived from SBM that implement the SB Algorithm invented in the research process for quantum computers at Toshiba's R&D centers. The “+” indicates that this solution includes various services and indicates our intention to continuously strengthen the solution.
Moving forward, in addition to providing execution modules on the current AWS Marketplace, we plan to provide cloud services for SQBM+ in Azure Quantum, which is a full-stack public cloud ecosystem for quantum solutions published by Microsoft Corporation. We also plan to provide on-premise versions for highly-confidential applications not suitable for cloud environments and applications that require ultra-low latency, and to embed those versions in partner applications or provide them on an OEM basis. Furthermore, we will provide professional services such as education and support for formulation.
Features of SQBM+
1. Adopts a new algorithm that significantly improves speed, accuracy, and scale
- Ballistic Simulated Bifurcation Algorithm (bSB): A high-speed algorithm for finding a good solution in a short time8
By devising ways to reduce errors occurring with conventional aSB, we realized a faster and more accurate simulated bifurcation machine.
- Discrete Simulated Bifurcation Algorithm (dSB): A high-accuracy algorithm for finding more accurate solutions at a calculation speed that surpasses that of other machines.9
The pseudo-quantum tunneling effect breaks the limits of classical mechanics to achieve even higher accuracy, and it is possible to obtain the optimum solution (estimated value of the exact solution) of the same 2,000-variable problem.
- Includes a function that automatically selects and uses one of the two algorithms listed above.
2. Provides optimization solvers according to the intended application
In addition to an Ising solver10
that can be used for general purposes, we also provide solvers for specific purposes. This makes it possible to easily and directly solve specific problems.
- Ising solver
A general-purpose solver. This is the basic solver of SQBM+ which uses the SB Algorithm to solve combinatorial optimization problems expressed in the format of quadratic unconstrained binary optimization (QUBO)..
- TSP solver
A solver that can directly solve a type of problem called the traveling salesman problem without expressing it in QUBO.
- SHIFT solver
A solver that can directly solve shift scheduling problems such as assigning daily jobs to employees under various constraints without using QUBO.
3. Expansion of application range
We provide extended functions and solvers which make the Ising solver easier to use and to improve its performance.
- Parameter automatic adjustment function/solver
Automates the tuning of unique SBM parameters and quickly finds better solutions without the hassle of manual adjustment. There is an automatic adjustment function for each parameter and an Ising solver that eliminates the need to adjust any of the parameters.
- QP solver
A solver that can directly solve quadratic binary optimization problems with linear constraints. Compared to solving similar problems with the Ising solver, there is no need to incorporate linear constraints into QUBO and adjust penalty parameters, thus making it easier to obtain highly-accurate solutions.
- Enabling high-order terms and continuous variables
Real-life combinatorial optimization problems may contain cubic or higher terms and continuous variables. Solving such problems with the Ising solver requires conversion to quadratic expressions and binary variables, which is a factor in degrading the equation solution performance. SQBM+ will utilize the features of the SB Algorithm to support these higher-order terms and continuous variables, and to achieve higher solution performance for real-life combinatorial optimization problems.
The Company will continue to actively search for real problems where this solution is useful, develop functional enhancements to further expand the scope of application, and develop professional services to support introduction to customers. We aim to provide a service platform that meets the needs for all optimizations.
At the time of this announcement, we have received the following comments from our current collaboration partners.
“Unlike existing platforms that primarily offer access to hardware and programming options, we offer a true commercialization platform supporting a plug and play model with access to a ready and growing market through our partnerships with several leading IT services providers, research and educational institutions and other partners globally. With access to production grade SQBM＋™ and a 100 Qubit Gate Simulators supported via desktop access on Windows, Mac and Linux we offer a true cross platform experience. It also solves one of the biggest challenges for Quantum Start Ups – how to make money from their innovation.”
Vish Ramakrishnan, CEO CogniFrame.
・ About CogniFrame
CogniFrame, based in Toronto, Canada, solves NP Hard and other complex optimization, machine learning and simulation problems. It works with leading HPC and Quantum hardware providers to build and run proprietary algorithms and solutions that deliver immediate measurable value and help de-risk adoption of Near Quantum & Hybrid Quantum solutions by institutions globally. CogniFrame is a quantum pioneer and member of Quantum Industry Canada. Through its collaboration with Toshiba Digital Solutions Corporation, it is commercializing several of its existing Near Quantum solutions on Toshiba’s Simulated Bifurcation Machines (SBM). It has launched the FirstQ Store, the first of its kind aggregator plug and play commercialization store for near quantum and quantum ready applications. For more information email us: email@example.com URL: www.cogniframe.com
We sincerely welcome the launch of the quantum-inspired optimization solution SQBM+.
Groovenauts offers the cloud platform MAGELLAN BLOCKS, which is equipped with a unique combinatorial optimization model, in order to optimize customers' processes and work orders, logistics and delivery, loading, work shifts, and more. We expect that collaborating with SQBM+ will enable us to deal with even larger and more complex combinatorial optimization problems. Moving forward, we will continue to work on solving problems by utilizing SQBM+ compatible solutions in accordance with problems faced by our customers in areas such as manufacturing, distribution, finance, and government offices.
・ About Groovenauts, Inc.
Groovenauts has a vision of "contributing to the realization of a prosperous and humane society," and, in line with people's diverse values, it looks to support the perpetuation of society's future potential and prosperity through the use of technology. The issues facing society today may also be seen as a reflection of the accumulated issues faced by individual companies. That is why Groovenauts looks to unravel complex problems through the power of thought and technology in order to tackle real issues and solve problems in society to support the true potential of human beings.
・ About Groovenauts' MAGELLAN BLOCKS business: A quantum computer x AI cloud platform
- Providing MAGELLAN BLOCKS and consulting so that anyone can easily utilize quantum computers, AI, and big data
- For urban areas, provides “City as a Service” in order to (1) visualize and analyze conditions, (2) predict and simulation changes, and (3) use optimization to create city services which are comfortable and humane