National Science Review

Mặc dù đã có nhiều sự chú ý đến việc điều tra và kiểm soát ô nhiễm không khí tại Trung Quốc, nhưng xu hướng nồng độ chất ô nhiễm không khí ở quy mô quốc gia vẫn chưa rõ ràng. Ở đây, chúng tôi đã điều tra định lượng sự biến đổi của các chất ô nhiễm không khí tại Trung Quốc bằng cách sử dụng các tập dữ liệu tổng hợp dài hạn từ năm 2013 đến 2017, trong đó chính phủ Trung Quốc đã nỗ lực lớn để giảm phát thải con người ở các vùng ô nhiễm. Kết quả của chúng tôi cho thấy một xu hướng giảm đáng kể nồng độ PM2.5 ở các khu vực ô nhiễm nặng của miền đông Trung Quốc, với mức giảm hàng năm khoảng 7% so với các số đo năm 2013. Các nồng độ SO2, NO2 và CO (một chỉ số cho các hợp chất hữu cơ bay hơi do con người gây ra) đã giảm cũng giải thích một phần lớn cho sự giảm nồng độ PM2.5 ở các vùng khác nhau. Kết quả là, số ngày ô nhiễm nặng giảm đáng kể ở các vùng tương ứng. Nồng độ aerosol hữu cơ, nitrate, sulfate, ammonium và chloride đo được ở thành phố Bắc Kinh cho thấy một sự giảm đáng kể từ năm 2013 đến 2017, kết nối chặt chẽ sự giảm các tiền chất aerosol với các thành phần hóa học tương ứng. Tuy nhiên, nồng độ ozone bề mặt có xu hướng tăng ở hầu hết các trạm đô thị từ năm 2013 đến 2017, điều này cho thấy ô nhiễm quang hóa mạnh hơn. Chiều cao tầng biên ở các thành phố thủ đô của miền đông Trung Quốc không cho thấy xu hướng đáng kể nào trên các vùng Bắc Kinh - Thiên Tân - Hà Bắc, Đồng bằng sông Dương Tử và Đồng bằng sông Châu Giang từ năm 2013 đến 2017, điều này xác nhận sự giảm phát thải con người. Kết quả của chúng tôi đã chứng minh rằng chính phủ Trung Quốc đã thành công trong việc giảm bụi mịn ở các khu vực đô thị từ năm 2013 đến 2017, mặc dù nồng độ ozone đã tăng đáng kể, điều này gợi ý một cơ chế phức tạp hơn trong việc cải thiện chất lượng không khí tại Trung Quốc trong tương lai.

Many different phase structures have been discovered for silver iodides. The β and γ phases were found to be the most common ones at ambient conditions, while the rock-salt phase was found to be stable under pressures between 400 MPa and 11.3 GPa. Recently, the α phase was demonstrated to be stable under ambient conditions when the particle sizes were reduced to below 10 nm. However, no other phase has been reported to be stable for silver iodides under ambient conditions. Rock-salt and helix structures have been found to be stable under ambient conditions in this study. The structures have been characterized by elemental mapping, Raman scattering, and high-resolution transmission electron microscopy. The stabilities of these structures were also confirmed by molecular dynamics and density functional theory.

The frequent occurrences of the second-year surface cooling condition in the eastern equatorial Pacific, as observed in late 2021, are attributed to decadal changes in the thermocline depth, which determine the relative dominances of local cooling effect in the east and subsurface warming effect remotely from the west. Coupled models need to adequately represent these processes in a balanced way, thus being able to successfully predict the observed sea surface temperature evolution in late 2021.

Syngnathids (seahorses, pipefishes and seadragons) exhibit an array of morphological innovations including loss of pelvic fins, a toothless tubular mouth and male pregnancy. They comprise two subfamilies: Syngnathinae and Nerophinae. Genomes of three Syngnathinae members have been analyzed previously. In this study, we have sequenced the genome of a Nerophinae member, the Manado pipefish (Microphis manadensis), which has a semi-enclosed brood pouch. Comparative genomic analysis revealed that the molecular evolutionary rate of the four syngnathids is higher than that of other teleosts. The loss of all but one P/Q-rich SCPP gene in the syngnathids suggests a role for the lost genes in dentin and enameloid formation in teleosts. Genome-wide comparison identified a set of 118 genes with parallel identical amino acid substitutions in syngnathids and placental mammals. Association of some of these genes with placental and embryonic development in mammals suggests a role for them in syngnathid pregnancy.

Conventional machine learning studies generally assume close-environment scenarios where important factors of the learning process hold invariant. With the great success of machine learning, nowadays, more and more practical tasks, particularly those involving open-environment scenarios where important factors are subject to change, called open-environment machine learning in this article, are present to the community. Evidently, it is a grand challenge for machine learning turning from close environment to open environment. It becomes even more challenging since, in various big data tasks, data are usually accumulated with time, like streams, while it is hard to train the machine learning model after collecting all data as in conventional studies. This article briefly introduces some advances in this line of research, focusing on techniques concerning emerging new classes, decremental/incremental features, changing data distributions and varied learning objectives, and discusses some theoretical issues.

The Tibetan Plateau and its surroundings are known as the Third Pole (TP). This region is noted for its high rates of glacier melt and the associated hydrological shifts that affect water supplies in Asia. Atmospheric pollutants contribute to climatic and cryospheric changes through their effects on solar radiation and the albedos of snow and ice surfaces; moreover, the behavior and fates within the cryosphere and environmental impacts of environmental pollutants are topics of increasing concern. In this review, we introduce a coordinated monitoring and research framework and network to link atmospheric pollution and cryospheric changes (APCC) within the TP region. We then provide an up-to-date summary of progress and achievements related to the APCC research framework, including aspects of atmospheric pollution's composition and concentration, spatial and temporal variations, trans-boundary transport pathways and mechanisms, and effects on the warming of atmosphere and changing in Indian monsoon, as well as melting of glacier and snow cover. We highlight that exogenous air pollutants can enter into the TP’s environments and cause great impacts on regional climatic and environmental changes. At last, we propose future research priorities and map out an extended program at the global scale. The ongoing monitoring activities and research facilitate comprehensive studies of atmosphere–cryosphere interactions, represent one of China's key research expeditions to the TP and the polar regions and contribute to the global perspective of earth system science.

Solar steam generation is emerging as promising solar-energy conversion technology for potential applications in desalination, sterilization and chemical purification. Despite the recent use of photon management and thermal insulation, achieving optimum solar steam efficiency requires simultaneous minimization of radiation, convection and conduction losses without compromising light absorption. Inspired by the natural transpiration process in plants, here we report a 3D artificial transpiration device with all three components of heat loss and angular dependence of light absorption minimized, which enables over 85% solar steam efficiency under one sun without external optical or thermal management. It is also demonstrated that this artificial transpiration device can provide a complementary path for waste-water treatment with a minimal carbon footprint, recycling valuable heavy metals and producing purified water directly from waste water contaminated with heavy metal ions.

As a promising area in machine learning, multi-task learning (MTL) aims to improve the performance of multiple related learning tasks by leveraging useful information among them. In this paper, we give an overview of MTL by first giving a definition of MTL. Then several different settings of MTL are introduced, including multi-task supervised learning, multi-task unsupervised learning, multi-task semi-supervised learning, multi-task active learning, multi-task reinforcement learning, multi-task online learning and multi-task multi-view learning. For each setting, representative MTL models are presented. In order to speed up the learning process, parallel and distributed MTL models are introduced. Many areas, including computer vision, bioinformatics, health informatics, speech, natural language processing, web applications and ubiquitous computing, use MTL to improve the performance of the applications involved and some representative works are reviewed. Finally, recent theoretical analyses for MTL are presented.