SBIR Phase I: Hybrid Computing Techniques for Quantum-inspired Ising Machines

Leviosa Technologies Llc, Woodbury MN

Investigators

Abstract

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project will result from the development of a high-speed, low-power solver based on standard semiconductor technologies that can directly solve combinatorial optimization problem (COP) problems. COP problems have broad applicability across manufacturing optimization, semiconductor wire routing, logistics planning and execution, and financial portfolio management. However, COP problems are notoriously difficult or even impossible to solve via classical computers on a scale suitable for commercial application. Current state-of-the-art COP solvers need tremendous computing power, are unsuitable for edge computing, rely on undeveloped technology, or use similar algorithms to classical computers. The goal of this project is to provide consumers with a drop-in replacement COP solver that is faster, more precise, more mobile, and more energy efficient than state-of-the-art classical computers and COP solvers. This Small Business Innovation Research (SBIR) Phase I project seeks to develop a complementary metal–oxide–semiconductor (CMOS) based parallel combinatorial optimization problem (COP) computing cluster accessible through a cloud interface. The proposed solution will directly solve COP problems, increasing the speed, precision, and power efficiency compared to classical computers or current quantum-based COP solvers. Additionally, the cluster uses all standard semiconductor technology allowing for near-term hardware manufacturing, unlike current quantum computing competitors. The device also works at room temperature, making it the only suitable edge device for directly solving COP problems. A hybrid computing algorithm combining the custom and classical methods will parallelize the COP solving across numerous custom chips. Many custom chips will be combined into a single cluster to maximize the speed and efficiency of the hybrid algorithms. An introductory cloud service interface will be incorporated with the custom computer cluster to facilitate outside access. The end goal of this project is to create a state-of-the-art scalable, accessible, and economical COP solver that overcomes the inherent disadvantages of quantum and classical computing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.