Quantum computing reports are still working up steam, but error correction remains elusive. Research into quantum computing has yielded many positive results, in quantum computing and also improvements in classical algorithms. In addition, it has inspired new hardware that potentially improves how businesses will optimize. Last year, SAP and Fujitsu collaborated to test out Fujitsu’s new quantum-inspired Digital Annealer Unit (DAU).
If you ask a quantum researcher what keeps them awake at night, error correction might be their likely answer. And so, while quantum computing still makes great strides, the quantum advantage can sometimes be overstated in the press. To mitigate this, however, a new type of quantum-inspired computing is coming of age. What started as simulated annealing has given rise to dedicated simulated annealing machines. These typically work on optimization problems, finding the best way to do something.
Harnessing Fujitsu’s Digital Annealer Unit for Efficient Optimization
Fujitsu’s Digital Annealer Unit (DAU) is a quantum-inspired machine. It works by finding the optimal combination when presented with a list of possible combinations, and it works fast. Last year, SAP ran a proof of concept with Fujitsu to test and benchmark its DAU – with some promising results.
As mentioned in a recent blog, the DAU was tested against some rather abstract but standardized research problems from the Library of Quadratic Programming Instances (QPLIB). We also tested the possibility to seamlessly integrate SAP solutions with the DAU, proving a simple execution to start. As a result, SAP was able to demonstrate that the DAU has real potential and marks an important milestone in the progress of quantum-inspired computing towards practical, real-world, standard business solutions. We are currently working with Fujitsu to extend this potential to other optimization problems, like creating more efficient schedules for manufacturing, vehicle space optimization, and ultimately an even broader use case application.
While still a proof of concept, these connections and benchmarking results demonstrate the potential for simulated annealing to deliver on real-world problems, such as balancing the load across all manufacturing tool sets, by using the DAU as a machine dedicated to the task of optimization, rather than a universally reprogrammable computer.
As a world-leading enterprise application provider, SAP continues to help innovate and bring the benefits of quantum-inspired computing to customers through our solutions.
Benchmarking Against Current Optimization Methods
Before committing more resources, SAP wanted to test the DAU against problems that serve as industry or academic standards for optimization. This is where the QPLIB comes in. It is essentially a huge repository of math problems, none of which are mapped to potential use cases but all of which can determine what kind of problems the DAU might excel at. The advantage of this approach is that it allows us to benchmark against current optimization methods, although the use case context has been removed. Take the example of a drug trial, where scientists test the new drug in test tubes against drugs already on the market. It may not show side effects, but it will show whether it’s worth risking trials on living subjects.
The DAU takes a Quadratic Unconstrained Binary Optimization (QUBO) problem as its input, which sets up the cost-benefit matrix. It then outputs a string of 1’s and 0’s, telling the user whether it is optimal to include, for example, a unit of stock, where the variance is to be minimized. The key is that it must either be faster or output a more optimal solution.
Most quantum computers run their quantum circuits at gate-level. Fujitsu’s approach was inspired by quantum annealing, a technique that specializes entirely in optimization. Although quantum annealing makes great progress in the number of qubits (quantum bits), it is subject to similar limitations of error. The advantage of Fujitsu’s quantum-inspired approach is that it relies on existing hardware, eliminating the need for expensive cryogenic cooling.
It is SAP’s sincere hope that this quantum-inspired optimization can help our customers bridge the optimization gaps until quantum hardware becomes more reliable. The incorporation of Fujitsu’s DAU into SAP technology helps establish a platform designed to tackle the impending gaps in the years ahead. This remarkable advancement is a key achievement in SAP’s journey towards the next era of enterprise software, aiming to enhance business operations and foster value creation.
Together, SAP and Fujitsu are pushing the boundaries and redefining what is possible, as the industry looks to a future transformed by the advancements in quantum-inspired computing.
Paul McElligott is a fellow in the Quantum eXplorers Group at SAP.