Â鶹ÒùÔº

An acoustic Ising machine: Novel system tackles hard combinatorial problems

Researchers at the University of Gothenburg have developed a novel Ising machine that utilizes surface acoustic waves as an effective carrier of dense information flow. This approach enables fast, energy-efficient solutions to complex optimization problems, offering a promising alternative to conventional computing methods based on von-Neumann architecture. The findings are in the journal Communications Â鶹ÒùÔºics.

Traditional computers can stumble when tackling combinatorial optimization problems—tasks of scheduling logistic operations, financial portfolio optimization and high frequency trading, optimizing communication channels in complex wireless networks, or predicting how proteins fold among countless structural possibilities.

In these cases, each added node—an additional logistic hub, network user, or molecular bond causes the number of possible configurations to explode exponentially. In contrast to linear or polynomial growth, an exponential increase in the number of possible solutions makes even the most powerful computers and algorithms lack the computational power and memory to evaluate every scenario in search of vanishingly small subsets representing satisfactorily optimal solutions.

Now, researchers at the University of Gothenburg have unveiled a new "surface acoustic wave" Ising machine (SAWIM) that could solve certain notoriously difficult computing tasks in fractions of the power and size demanded by typical computers. Inspired by coherent Ising machines (CIM) that use laser pulses, this acoustic-based design retains all-to-all connectivity but dramatically improves thermal stability by four to five orders of magnitude thanks to a much lower carrier frequency—presenting a commercially feasible platform for combinatorial problem accelerators.

Ising machines as fast and computationally effective combinatorial solvers

Ising machines are physics-inspired hardware and algorithmic solvers that tackle combinatorial problems by mapping them onto a classical model of magnetism—an Ising model. An Ising machine reimagines combinatorial problems by turning them into an "energy landscape" of tiny oscillators, each able to take an "up" or "down" state.

In physics, systems naturally settle into the lowest-energy arrangement—like an array of tiny magnets trying to align. If the connections among these oscillators represent the problem to be solved, then finding the overall lowest-energy state is performed by simply letting the system relax into its steady state, often representing a global or close to optimum solution. In essence, Ising machines implement a hardware shortcut that employs nature's own tendency toward minimal energy.

The Coherent Ising machine tackled the problem of using light pulses in an optical loop. While successful at scaling to hundreds of thousands of spins, they typically require precise temperature control or additional stabilization systems and operate stably only at microsecond to millisecond time scales.

"We took inspiration from coherent Ising machines and replaced light with radio-frequency acoustic pulses, significantly reducing the thermal-induced phase instabilities, allowing SAWIM to operate for hours without any frequency stabilization and thermal compensation system," said Dr. Artem Litvinenko, first and corresponding author of the new study.

The frequency factor

CIMs run at carrier frequencies of around 200 Terahertz—electromagnetic waves close to visible light spectrum that undergo enormous phase accumulation when they circulate through an optical loop.

Even tiny temperature shifts can cause phase drift, which disrupts phase binarization of equivalent Ising spins or requires the introduction of inefficient feedback correction systems. In contrast, SAWIM relies on a far lower frequency of roughly 300 Megahertz. As a result, the overall phase change is substantially reduced—by a factor of 100,000 or more—making the system far less susceptible to temperature fluctuations.

"In optical systems, a tiny change in temperature can distort the delicate phase-encoding of equivalent Ising spins represented with light pulses, effectively throwing off the entire calculation," said Professor Johan Ã…kerman, organizing and corresponding author on the work. "In SAWIM delay lines, acoustic waves don't build up anywhere near that amount of phase accumulation shift, so we don't need a special thermostat system of a phase-locked loop frequency stabilization system."

How acoustic waves compute

Inside SAWIM, microwave pulses create propagating surface acoustic wave packets that travel along a crystal surface storing the information about the phase and the amplitude of the equivalent Ising spins. Each wave packet (equivalent Ising spin) can settle into two stable phase states.

The system interconnects these packages virtually using a digital matrix multiplication block implemented with Field Programmable Gate Arrays (FPGA) so that they interact in a way that corresponds to the interconnection matrix in an arbitrary optimization problem. Once the machine is switched on, the acoustic wave packets self-organize into the configuration with the lowest energy, thus solving the combinatorial problem in a huge solution space.

At present, the SAWIM prototype supports up to 50 spins that can be interconnected all-to-all, already solving mid-size combinatorial tasks. Since it's built with standard, off-the-shelf electronics and has no bulky optical tables, the next step will be scaling up the number of spins to thousands or even tens of thousands, while maintaining the acoustic approach's inherent temperature stability and compact size.

A smaller, cooler path forward

Because of the low carrier frequencies and small operating amplitudes used in SAWIM, the system operates at room temperature with just a few watts of power—much lower than many optical systems. SAWIM exploits well-developed SAW technology since lithium niobate (the wave-propagation substrate) and the associated electronics are widely used in telecommunications.

"We're at the start of exploring stable, commercially feasible wave-based computing," said Dr. Litvinenko. "The SAW high thermal stability allows us to further develop computational complexity, exploring the opportunities to shift from binarized spins to multi-level spins, which hold tremendous potential for combinatorial computing and are difficult with optical Ising machines."