Fixstars develops development support environment for quantum computing

Fixstars Inc.
Fixstars develops development support environment for quantum computing Information on programming support middleware and simulation environment on the website

Fixstars Co., Ltd., a global leader in high-speed technology using multi-core CPU / GPU / FPGA (Head office: Shinagawa-ku, Tokyo, President: Atsushi Miki, hereinafter referred to as Fixstars) is a development support environment for quantum computing Publish information on the website. Fixstars has developed two development support environments: quantum annealing programming support middleware (hereinafter referred to as middleware) and a simulation environment that runs on the GPU. We will develop a development support
environment that will be useful for those who are going to work on quantum computing or who are already doing quantum computing, and will support the spread of quantum computing to the world. Information on the development support environment can be obtained from the Quantum Computing Solutions website (Quantum Computing Solutions (
https://quantum.fixstars.com/)

[Image 1d27784-13-621291-7.png
Quantum computing is attracting attention as a technology that can solve problems that cannot be solved by current computers. It is expected to be able to solve machine learning / AI calculations and combinatorial optimization problems in the financial and logistics fields in a short time. Quantum computers are being developed by companies around the world, including Japan. On the other hand, in order to solve problems using quantum computers, it is necessary to understand the limitations of each hardware in addition to the programming methods unique to quantum computers. At Fixstars, we wanted to create a world where many engineers in the world can work on quantum computing more easily. .
1) Middleware for quantum annealing programming support
Quantum computers currently being developed by various companies have different hardware specifications and restrictions depending on the model. Middleware reduces this burden and provides an environment where users can focus on application development. Middleware increases the efficiency of programming through three automations. Currently, it supports models from four companies, and plans to continue expanding the supported models in the future.
a) Automation of formulation
The process of defining combinatorial optimization problems and turning formulas and input models into data is complex, but automates these tasks. It supports various input formats and provides an interface that can be described in program code without converting mathematical expressions.
b) Input model data
Because there is a difference in the specifications of each model, it is necessary to convert to an input model that matches the hardware you want to use. Middleware automatically converts the formulated input model into an input-capable format that takes into account hardware limitations.
c) Deployment between multiple models
When executing the same program between different models, it is necessary to convert the program for each model. Using middleware, it is possible to switch the model to be operated by simply rewriting the command that indicates the model to be used in the program code. Since it is not necessary to formulate and re-implement each model, it is possible to select a model that has the performance suitable for your problem in a short period of time.
[Image 2d27784-13-426218-6.png
Figure 1. Quantum annealing programming procedures and areas supported by middleware
2) Simulation environment using GPU annealing machine
We are developing a GPU annealing machine as an environment for easily simulating quantum annealing methods. This is the same technique as the quantum annealing method built on the GPU, and you can experience quantum computing in a pseudo manner. Since it is implemented in software on a general-purpose GPU accelerator, after an optimization model has been created, customers such as developing a dedicated optimization solver specialized for the problem and porting it to an embedded system using the CUDA platform Flexible customization is possible according to the issues and applications.
The optimization model for this GPU annealing machine is QUBO (Quadratic Unconstrained Binary Optimization), which is a maximum combination of up to 8192 bits. We plan to expand to 100,000 bits in the future.
By combining this GPU annealing machine and middleware, it is possible to seamlessly transition to other models after development using the GPU annealing machine. In addition, since you can move smoothly between models, you can always use the latest model to accelerate development. Here are some examples of using this middleware and GPU annealing machine. This is a benchmark result of a typical traveling salesman problem in a combinatorial optimization problem. It solves the optimal solution of the traveling salesman problem between 51 cities called TSP LIB eil51 at the input of the optimization model in QUBO format in 1.1 seconds. Benchmark results for other cases are also posted on the website.
[Image 3d27784-13-420922-8.png
Figure 2. An example of solving a traveling salesman problem in 51 cities using middleware and a GPU annealing machine
In addition to the information on “Development Support Environment”, the website that describes the development status of the development support environment also includes “Technology Resources” that introduces the latest technology information on quantum computing and “Development Services” provided by Fixstars. It is released. “Technical resources” have been open since 2017 and are accessed by many engineers. Engineers who will continue to work on quantum computers will continue to provide technical information and information on trends.
The development support environment announced this time will be made available to the public in the future. The development status is updated on the website. If you wish, we will also send you an email. Applicants can apply from within the Quantum Computing Solutions website.
Fixstars aims to contribute to society by spreading and promoting the use of quantum computers.
About Fixstars
Fixstars is a technology company that advocates “Speed ​​up your Business” as its corporate message. Through the parallelization and optimization of software for efficient use of multi-core processors and the acceleration of applications utilizing new memory technology that realizes power saving and high-speed IO, various applications such as medical, manufacturing, finance, entertainment, etc. Accelerate the business of customers in the field and realize green IT.
Company Name Fixstars Co., Ltd. https://www.fixstars.com/jp/ Gate City Osaki West Tower 18th Floor 1-11-1 Osaki, Shinagawa-ku, Tokyo Established August 8, 2002
Capital 552.96 million yen (as of the end of September 2019) Representative Director and President Satoshi Miki



For more information about this release(Japanese):

https://prtimes.jp/main/html/rd/p/000000013.000027784.html




  • Comment ( 0 )

  • Trackbacks are closed.

  1. No comments yet.