Optimizing bit-by-bit operations for energy-efficient computing

How can we make computing more energy-efficient when nature itself introduces randomness at every step? At the intersection of physics and information science, researchers from AMOLF and Imperial College London explore how thermal noise interferes with the physical switching of bits in computing devices. Their new work, published in Physical Review E, reveals that how we measure these state changes can make a surprising difference in designing the most efficient computational protocols. 

Thermal noise and a bit of a challenge 

Temperature is an indication of random movement on the molecular level. This random movement, or thermal noise, affects the way the molecular systems in materials change from one state to another. For example, for the physical bits in your computer to turn on and off, these systems need to change states. But because thermal noise pushes the bits around, they might go back-and-forth before reaching their final state. This can be hard to control in an energy-efficient way, a challenge which interests many scientists and developers in the field of computation.  

Measuring a journey in jumps 

In trying to optimize noisy bit transitions, researchers at AMOLF and Imperial College London looked at how the duration of a state transition is measured. You could measure the duration in terms of time, but you could also look at the process in terms of how many back-and-forth steps, or “jumps,” it takes to reach the final state. The total number of jumps is also called the total activity in the system.  

It turns out that, mathematically, measuring in jumps is much simpler. However, as Daan Mulder, Thomas E. Ouldridge and Pieter Rein ten Wolde show in their paper in Physical Review E, you get different results for the most efficient protocol for resetting a bit state when measuring in terms of these two different variables. To get the most energy-efficient protocol for a given duration, you still need to do the harder, time-based math. The activity-based approach can be a good approximation, but only in some cases. This finding contributes to our understanding of how to best manipulate materials for energy-efficient computation.  

Reference 

Daan Mulder, Thomas E. Ouldridge, Pieter Rein ten Wolde, “Bit reset protocols that obey activity-constrained speed limits do not minimize work for a given speed,” Physical Review E 111, 044141, 29 April (2025). 

Read the full article: https://doi.org/10.1103/PhysRevE.111.044141