Mechanical processor goes where no digital computer has gone before

Computing

Technological Innovation Website Editorial Team - September 8, 2025

Scanning electron microscope image showing a channel of the charge-density-wave component, used to create the coupled oscillator circuit. [Image: Jonas Olivier Brown et al. - 10.1103/zmlj-6nn7]

Combinatorial optimization

With all the advances in computer science, we still lack adequate tools to deal with a class of problems known as combinatorial optimization problems , which are common in real-world applications, from programming and telecommunications organization to travel and delivery logistics.

When it comes to these problems, supercomputers are quickly reaching their limits, training artificial intelligence models requires enormous amounts of energy, and quantum computers are not yet powerful enough.

Jonas Brown and colleagues at the University of California, Los Angeles, and the University of Riverside are pioneering alternative computing , and have now presented a new approach that overcomes the obstacles of combinatorial optimization using a kind of analog mechanical processor - they call the approach "physics-inspired computing."

The team designed a system that processes information using a network of oscillators, components that move back and forth at specific frequencies, rather than representing all the data digitally.

Physics-inspired computing

This type of computer architecture, called an Ising machine , has a special capacity for parallel computing, performing numerous complex calculations simultaneously. When the oscillators are finally synchronized, the optimization problem is solved.

This technology can operate with very low power consumption compared to electronic computers. And this particular prototype is compatible with conventional silicon technology.

"Any new physics-based hardware needs to be integrated with standard digital silicon CMOS technology to impact data processing systems," said team member Alexander Balandin. "The two-dimensional charge-density-wave material we selected for this demonstration has the potential for such integration."

Illustration of the maximum-cut optimization problem, showing the 6 × 6 connected graph, circuit representation of the six coupled oscillators using the weights described in the connectivity matrix, and values ​​of the phase sensitivity function. [Image: Jonas Olivier Brown et al. - 10.1103/zmlj-6nn7]

Mechanical processor

The team designed their mechanical processor to function as an analogue of certain quantum properties that connect electrical activity to vibrations (phonons) propagating through a material . However, unlike most current quantum computing applications, which require cryogenic temperatures to maintain their "quantumness," the team's prototype operates at room temperature.

To bridge the gap between quantum mechanics and the more familiar physics of everyday life, the team used a special material, tantalum sulfide (TaS ₂ ), a "quantum material" that allows the alternation between electrical and vibrational phases to be revealed.

"Our approach is physics-inspired computing, which has recently emerged as a promising method for solving complex optimization problems," Balandin explains. "It uses physical phenomena involving condensates of strongly correlated electrons and phonons to perform calculations directly through physical processes, thus achieving greater energy efficiency and speed."

The prototype demonstrated that the oscillators naturally evolve to a lower-energy ground state, where they then become synchronized, allowing the machine to solve combinatorial optimization problems. The next step will be to expand the processor by increasing the number of oscillators.

Other news about:

More topics