Finite-Time Landauer Principle

September 11, 2020

A recently published paper in Physical Review Letters from members in the Bechhoefer group was picked to be an Editor's Suggestion!  

Finite-Time Landauer principle
by Karel Proesmans, Jannik Ehrich, and John Bechhoefer
Phys. Rev. Lett. 125, 100602 (2020)
Link to article:

Below is a short summary of the paper intended for a general physics audience:

The costs of fast computation

Around 5% of energy consumption is spent on computation. An unavoidable contribution to this cost is the erasure of information. While the minimum energy dissipation for information erasure was determined by Landauer in the 1960s, the need for fast computations means that real-world computation systems dissipate orders of magnitude more energy. Landauer’s thermodynamic bound assumes an infinitely slow erasure process. Previous studies have suggested that Landauer’s cost must be augmented by an additional term inversely proportional to the time for erasure: The quicker the process, the more energy is needed. In our Letter we generalize Landauer’s limit to finite time. We show how to erase a bit while dissipating the least amount of heat possible. Surprisingly, this cost can be sandwiched between simple bounds differing by a factor of four, irrespective of the detailed implementation of the computational bit. Roughly speaking, once one has full control over the computational bit, the erasure cost is given by Landauer’s cost plus a term proportional to the square of system size and inversely proportional to the erasure time.

The video above illustrates the optimal erasure process.