Speaker
Description
Active work quantifies the amount of energy converted into the actual motion of active particles. In the long-time limit, the time-averaged active work quantifies how much energy is dissipated by active particles into the surrounding environment, i.e., it is a measure of the entropy production rate of active particles. Thus, by calculating the large deviation function (LDF) of the time-averaged active work in the long-time limit, we can gain information about (i) the likelihood of an atypical energy dissipation rate and (ii) what the system looks like when such atypical events occur. A previous study on two-dimensional active Brownian particles [1] showed that the LDF exhibits singularities associated with transitions from the collectively moving state via the phase-separated state to the phase-separated arrest state, in the order of decreasing entropy production rate. In our study, we focus on the case of one-dimensional active Ornstein-Uhlenbeck particles, which corresponds to the active particles moving in a single-file along a narrow one-dimensional channel. In contrast to the previous study, we find in this case that the collective motion can be induced by decreasing the entropy production rate, and that further decrease of the entropy production rate induces another transition to a crystal state, where active particles effectively repel each other. Our results are based on a recently proposed machine learning method that utilizes the stochastic optimal control theory [2].
References
[1] T. Nemoto et al., Phys. Rev. E 99, 022605 (2019).
[2] J. Yan, H. Touchette, and G. M. Rotskoff, Phys. Rev. E 105, 024115 (2022).
