arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2209.08855cond-mat.stat-mechcond-mat.soft
收藏

几何布朗信息引擎:实现最佳性能的要素

Geometric Brownian Information Engine: Essentials for the best performance

Rafna Rafeek, Syed Yunus Ali, Debasish Mondal

AI总结:

本文研究几何布朗信息引擎在熵约束下的最优功提取条件,发现当反馈位置为测量距离两倍且测量距离约为0.6倍标准差时功提取达到全局最大,并揭示了熵控制对效能和平均位移的影响规律。

摘要:

我们研究了一种几何布朗信息引擎(GBIE),该引擎在无误差反馈控制器存在的情况下,将收集到的关于被困在单瓣几何约束中的布朗粒子状态的信息转化为可提取的功。信息引擎的输出取决于参考测量距离 $x_m$、反馈位置 $x_f$ 和横向力 $G$。我们确定了在输出功中利用可用信息的基准,以及实现最佳功提取的最优运行条件。横向偏置力($G$)调节有效势中的熵贡献,从而调节平衡边际概率分布的标准差($σ$)。我们认识到,当 $x_f = 2x_m$ 且 $x_m \sim 0.6σ$ 时,提取的功量达到全局最大值,而与熵限制的程度无关。由于弛豫过程中信息损失更大,几何布朗信息引擎在熵系统中可实现的最佳功较低。反馈调节还承载着粒子的单向通过。平均位移随着熵控制的增强而增大,并在 $x_m \sim 0.81σ$ 时达到最大值。最后,我们探讨了信息引擎的效能,即调节所获取信息利用效率的量。当 $x_f=2x_m$ 时,最大效能随着熵控制的增强而降低,并表现出从 $2$ 到 $11/9$ 的交叉。我们发现,实现最佳效能的条件仅取决于沿反馈方向的约束长度尺度。更宽的边际概率分布归因于一个循环中平均位移的增加以及熵主导系统中效能的降低。

英文摘要:

We investigate a Geometric Brownian Information Engine (GBIE) in the presence of an error-free feedback controller that transforms the information gathered on the state of Brownian particles entrapped in monolobal geometric confinement into extractable work. Outcomes of the information engine depend on the reference measurement distance $x_m$, feedback site $x_f$ and the transverse force $G$. We determine the benchmarks for utilizing the available information in an output work and the optimum operating requisites for best work extraction. Transverse bias force ($G$) tunes the entropic contribution in the effective potential and hence the standard deviation ($σ$) of the equilibrium marginal probability distribution. We recognize that the amount of extracted work reaches a global maximum when $x_f = 2x_m$ with $x_m \sim 0.6σ$, irrespective of the extent of the entropic limitation. Because of the higher loss of information during the relaxation process, the best achievable work of a GBIE is lower in an entropic system. The feedback regulation also bears the unidirectional passage of particles. The average displacement increases with growing entropic control and is maximum when $x_m \sim 0.81σ$. Finally, we explore the efficacy of the information engine, a quantity that regulates the efficiency in utilizing the information acquired. With $x_f=2x_m$, the maximum efficacy reduces with increasing entropic control and shows a cross over from $2$ to $11/9$. We discover that the condition for the best efficacy depends only on the confinement length scale along the feedback direction. The broader marginal probability distribution accredits the increased average displacement in a cycle and the lower efficacy in an entropy-dominated system.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑