With the demand for Internet connectivity in remote areas, the space‐air‐ground integrated network (SAGIN) was proposed to achieve ubiquitous coverage and enhance service capabilities of extant terrestrial networks. The paradigm of virtual network element placement (NEP) is applied into SAGIN. It can save energy and operating costs by placing specific network elements (NEs) as software instances. In addition, it helps to provide services in end‐to‐end networks with its ability to allocate and manage resources flexibly. However, NEP faces some challenges in SAGIN. The network topologies can be dynamic, and links such as the satellite‐to‐ground and air‐to‐ground ones are prone to fail. These will make NEP management more complicated. Moreover, the static NEP schemes are hard to accommodate the time‐varying traffic. In this context, this work explains the NEP problem in SAGIN from three aspects, i.e., the SAGIN radio access network (RAN), the SAGIN core network (CN), and the SAGIN barrier network (BN). First, the physical and networking architectures of SAGIN are introduced. Then the status of the network element placement and corresponding challenges are described from these two aspects. Finally, this paper discusses future research directions and key technical challenges.

Thinking space came into being with the emergence of human civilization. With the emergence and development of cyberspace, the interaction between those two spaces began to take place. In the collision of thinking and technology, new changes have taken place in both thinking space and cyberspace. To this end, this paper divides the current integration and development of thinking space and cyberspace into three stages, namely Internet of brain (IoB), Internet of thought (IoTh), and Internet of thinking (IoTk). At each stage, the contents and technologies to achieve convergence and connection of spaces are discussed. Besides, the Internet of creation (IoC) is proposed to represent the future development of thinking space and cyberspace. Finally, a series of open issues are raised, and they will become thorny factors in the development of the IoC stage.

Benefiting from the ultra‐wide bandwidth, terahertz (THz) communication is considered one of the core technologies of the 6G mobile communication. However, high path loss is a severe issue for THz band, which needs to be solved by the ultra‐massive multiple‐input multiple‐output (MIMO) technology. For the sake of reducing the hardware cost and power comsumption, the hybrid beamforming technology is required, which has been adopted in the 5G communication. In the hybrid beamforming for the THz wideband system, the enormous antenna and ultra‐wide bandwidth cause the spatial‐ and frequency‐wideband effect, resulting in acute beam squint. Eliminating beam squint is the key to the engineering realization of ultra massive MIMO for the THz band. To circumvent this issue, this paper presents a comprehensive overview of the hybrid beamforming technology including the channel model, the traditional phase shifter based and the true time delay based architectures, the 3D hybrid beamforming for the uniform planar array. Finally, multiple significant open issues and potential challenges are pointed out and elaborated.

Carrier behavior in halide perovskite is a critical factor impacting on the properties of material, and finally determines the performance of perovskite photovoltaic and luminescent devices. It is necessary to clarify the mechanism of carrier relaxation and migration at extremely microscopic time scale. Time‐resolved spectroscopy provides a powerful means for the detection of ultrafast processes, which has been an indispensable technique in research on perovskites. In this review, we will elaborate the basic principle and system implementation of time‐resolved spectroscopy, and introduce the applications for carrier dynamics in perovskite.

Deep forest is a tree‐based deep model made up of non‐differentiable modules that are trained without backpropagation. Despite the fact that deep forests have achieved considerable success in a variety of tasks, feature concatenation, as the ingredient for forest representation learning, still lacks theoretical understanding. In this paper, we aim to understand the influence of feature concatenation on predictive performance. To enable such theoretical studies, we present the first mathematical formula of feature concatenation based on the two‐stage structure, which regards the splits along new features and raw features as a region selector and a region classifier respectively. Furthermore, we prove a region‐based generalization bound for feature concatenation, which reveals the trade‐off between Rademacher complexities of the two‐stage structure and the fraction of instances that are correctly classified in the selected region. As a consequence, we show that compared with the prediction‐based feature concatenation (PFC), the advantage of interaction‐based feature concatenation (IFC) is that it obtains more abundant regions through distributed representation and alleviates the overfitting risk in local regions. Experiments confirm the correctness of our theoretical results.