Light is often characterized only by its classical properties, like intensity or coherence. When looking at its quantum properties, described by photon correlations, new information about the state of the matter generating the radiation can be revealed. In particular the difference between independent and entangled emitters, which is at the heart of quantum mechanics, can be made visible in the photon statistics of the emitted light. The well-studied phenomenon of superradiance occurs when quantum–mechanical correlations between the emitters are present. Notwithstanding, superradiance was previously demonstrated only in terms of classical light properties. Here, we provide the missing link between quantum correlations of the active material and photon correlations in the emitted radiation. We use the superradiance of quantum dots in a cavity-quantum electrodynamics laser to show a direct connection between superradiant pulse emission and distinctive changes in the photon correlation function. This directly demonstrates the importance of quantum–mechanical correlations and their transfer between carriers and photons in novel optoelectronic devices.

Abstract. This paper presents Recurrent Policy Gradients, a modelfree reinforcement learning (RL) method creating limited-memory stochastic policies for partially observable Markov decision problems (POMDPs) that require long-term memories of past observations. The approach involves approximating a policy gradient for a Recurrent Neural Network (RNN) by backpropagating return-weighted characteristic eligibilities through time. Using a "Long Short-Term Memory" architecture, we are able to outperform other RL methods on two important benchmark tasks. Furthermore, we show promising results on a complex car driving simulation task.

No abstract

We investigate correlations between orthogonally polarized cavity modes of a bimodal micropillar laser with a single layer of self-assembled quantum dots in the active region. While one emission mode of the microlaser demonstrates a characteristic S-shaped input-output curve, the output intensity of the second mode saturates and even decreases with increasing injection current above threshold. Measuring the photon autocorrelation function g (2) (τ ) of the light emission confirms the onset of lasing in the first mode with g (2) (0) approaching unity above threshold. In contrast, strong photon bunching associated with superthermal values of g (2) (0) is detected for the other mode for currents above threshold. This behavior is attributed to gain competition of the two modes induced by the common gain material, which is confirmed by photon cross-correlation measurements revealing a clear anticorrelation between emission events of the two modes. The experimental studies are in qualitative agreement with theoretical studies based on a microscopic semiconductor theory, which we extend to the case of two modes interacting with the common gain medium. Moreover, we treat the problem by a phenomenological birth-death model extended to two interacting modes, which reveals that the photon probability distribution of each mode has a double-peak structure, indicating switching behavior of the modes for pump rates around threshold.

No abstract