Across different industries, there are always complex optimisation problems to be encountered, including routing vehicles, managing supply chains, assessing risks, optimising portfolios, operating power grids, and more.
Even with numerous complicated algorithms developed to efficiently handle select problems, a lot of real-world optimisation problems are still difficult to optimise. This is despite many advancements in both algorithms and the power needed to compute them. Traditional methods are struggling to solve these scenarios as they involve many variables and are computationally challenging to solve in the first place.
Quantum methods allow us to leverage the situation by cutting down the time consumed to solve such situations, and with significantly less work done. Emulating quantum systems has been very promising in creating breakthroughs for many industries and areas. Among them are MRI technologies, reducing traffic congestion, designing materials, and more.
Simulated annealing, parallel tempering Monte Carlo, and genetics algorithms have one thing in common, that being they mimic natural processes. As we understand quantum mechanics more, new optimisers are constantly developed. These use quantum mechanics through annealing to speed up optimisation and escape local minima in the cost function landscape.
Many new types of quantum solutions called quantum-inspired optimisation (QIO) algorithms have been developed following simulating these effects on classical computers. The algorithms allow us to use CMOS-based classical hardware to employ select advantages of modern quantum computing approaches. This helps speeds up the process compared to using traditional approaches only. With the possibility of a future having scaled and fault-tolerant quantum hardware, using quantum solutions on classical computers will help us prepare for when it arrives.
Azure Quantum allows users and customers to run optimisation programs on industry-scale classical computers on CPUs, GPUs, and FPGAs in Azure, with self-service solutions solving binary optimisation problems.
Microsoft has been expanding its portfolio of QIO algorithms and solvers and is pleased to announce Toshiba’s decision in joining the Microsoft Quantum Network as well as offering Toshiba’s Simulated Bifurcation Machine (SBM) in Azure Quantum. 1Qbit, Honeywell, IonQ, and QCI are the existing partners that Toshiba will be collaborating with. All of them will be providing services to the growing quantum ecosystem together.
Possessing cutting-edge techniques, Toshiba can rapidly obtain extremely accurate solutions for complicated large-scale combinational optimisation problems. It also demonstrated around ten times improvement over competing devices. The plethora of combinational optimisation problems includes dynamic portfolio management, risk management, and high-frequency trading. Some practical applications are optimising electrical transmission line routing with consideration of cost, safety, time, and environmental effects, as well as finding the shortest route between cities considering the times of day, traffic status, and driver schedules.
Every computational problem seen in practice is translatable to a specific type of binary optimisation problem, which is searching for the ground state of an Ising model. In general, it is too costly to map such a problem. However, rewriting it into this form is often easier, and problems native to this form (including planning, scheduling, and partitioning) will be solved with Toshiba’s powerful solution.
SBM is a practical and readily deployed Ising model solver emerging from Toshiba’s quantum computing research that can speedily solve large-scale combinatorial problems while harnessing the Azure cloud’s GPU resources.
Soon, Azure Quantum users will be able to use the Toshiba SBM’s quantum methods to experiment with highly accurate solutions for their unique scenarios.
Offered by Microsoft, Azure Quantum is an open quantum ecosystem comprising of various quantum partners and technologies. Azure has many efficient and effective tools that can run on classical and quantum hardware. Through collaborating with its partners, it produces resources for quantum computing.