We investigate the stochastic behavior of a non-Markovian version of an elementary Brownian gyrator. The model is defined by overdamped Langevin-like dynamics with a two-dimensional harmonic potential that presents distinct principal axes and is coupled to heat baths at different temperatures. The thermal noises are assumed to be Gaussian, and are related to friction forces through a dissipation memory kernel. The stationary states present rotational motion with non-trivial average torques due to harmonic, friction and fluctuating thermal forces. However, the Markovian limit of the system exhibits a zero average torque produced by fluctuating thermal forces. For the case of stochastic torque exerted by harmonic force, the cumulant-generating function is calculated exactly. We also study the average heat fluxes in the steady-state regime, where a memory-dependent behavior is observed.

我们调查基本布朗回旋器的非马尔可夫版本的随机行为。该模型由具有二维谐波电势的过阻尼兰格文式动力学定义，二维电势具有不同的主轴，并耦合到不同温度下的热浴。假定热噪声是高斯噪声，并且与通过耗散存储内核的摩擦力有关。由于谐波，摩擦和波动的热力，静止状态呈现出具有非平凡的平均转矩的旋转运动。但是，系统的马尔可夫极限显示出由波动的热力产生的零平均转矩。对于由谐波力施加的随机转矩，将精确计算出累积量生成函数。我们还研究了稳态下的平均热通量，