Computers based on slime molds may be the next step for solving some of the most advanced and resource-heavy problems that tax the limits of conventional computing.
By studying the survival strategies of slime molds, researchers were inspired to create a new type of computer modeled on living organisms, as findings revealed by a team at Waseda University in a recent paper published in Physical Review Research describe how amoeba-like slime molds can be modeled in a new, simpler way to produce an ultra-efficient, low-energy computer.
These new computers are particularly suited to combinatorial optimization problems, which are of wide use in various industries but burn massive amounts of compute and energy.
Pushing Computers to Their Limit
The researchers’ new, simplified model can potentially be implemented using a wide selection of physical materials and phenomena to tackle combinatorial optimization problems. These types of problems involve finding the best possible combination of choices from a defined set of options. They are common in many industries where the challenge involves finding an optimal solution among numerous possibilities. This could involve discovering the ideal combination of ingredients for a new drug or, in materials science, designing a substance with highly specific properties.
Calculating every possible combination can become extremely expensive and taxing for traditional computers as the number of possibilities increases, leading computer scientists to explore new paradigms for solving these important optimization problems.
Slime Mold Computers
We often hear the human brain likened to a computer, but in this instance, researchers are casting a wider net. Instead of modeling human decision-making processes in ones and zeroes, slime mold computers look to the amoeba-like behavior of these organisms, particularly their survival strategies.
Surprisingly, these simple lifeforms regularly solve problems analogous to combinatorial optimization. This occurs through deformation dynamics, in which they alter their shapes to maximize nutrient intake while minimizing exposure to light. Such a balance of competing objectives closely resembles the kinds of optimization problems computers are tasked with solving.
However, implementing existing slime-mold-based models has been challenging, as they are often incompatible with the constraints of physical computing devices. In their new work, the Waseda researchers simplified an existing slime mold model to streamline how it handles information and make it easier to implement using physical systems.
“Our approach eliminates a major constraint, the volume-conservation law, making it possible to implement slime-mold computers using a much wider variety of materials and physical phenomena,” explained co-author assistant professor Yusuke Miyajima, of the Department of Applied Physics at Waseda University.
Amoebas and Combinatorial Optimization Problems
The traveling salesman problem (TSP) is one of the most classic and well-known combinatorial optimization problems. In it, a traveling salesman must identify the shortest route that visits every city on a list once before returning to the starting point.
The team started with the existing Amoeba TSP model, which uses slime mold dynamics to arrive at a solution. In physical implementations of this approach, researchers place an amoeba on a circular device with channels extending outward, representing cities in the TSP problem. As the amoeba extends branches along those channels in response to environmental stimuli, its behavior can be used to derive a solution.
With just five modifications to the Amoeba TSP model, the team enabled a much wider variety of physical materials and phenomena to be represented while also allowing the model’s parameters to be tuned. Tests showed that the new model produced feasible solutions in far fewer iterations than the original model, while increasing the number of cities it could handle from 100 to 180 and demonstrating mathematical equivalence to a recurrent neural network.
The researchers also identified alternative candidates for physical implementation, including spintronic devices and photonic circuits.
“The increased flexibility of our model can accelerate the development of energy-efficient slime-mold computers,” concluded co-author Dr. Masahito Mochizuki, a Waseda professor. “This decentralized mode of information processing could prove valuable for AI and large-scale combinatorial optimization, where conventional computers require significant power consumption.”
The paper, “Mathematical Model of the Amoeba-Inspired Combinatorial Optimization Machine for Physical Implementation and Its Equivalence to Recurrent Neural Networks,” appeared in Physical Review Research on June 9, 2026.
Ryan Whalen covers science and technology for The Debrief. He holds an MA in History and a Master of Library and Information Science with a certificate in Data Science. He can be contacted at ryan@thedebrief.org, and follow him on Twitter @mdntwvlf.