No abstract

Reducing human reliance on energy-inefficient cooling methods such as air conditioning would have a large impact on the global energy landscape. By a process of complete delignification and densification of wood, we developed a structural material with a mechanical strength of 404.3 megapascals, more than eight times that of natural wood. The cellulose nanofibers in our engineered material backscatter solar radiation and emit strongly in mid-infrared wavelengths, resulting in continuous subambient cooling during both day and night. We model the potential impact of our cooling wood and find energy savings between 20 and 60%, which is most pronounced in hot and dry climates.

Federated Learning is a distributed learning paradigm with two key challenges that differentiate it from traditional distributed optimization: (1) significant variability in terms of the systems characteristics on each device in the network (systems heterogeneity), and (2) non-identically distributed data across the network (statistical heterogeneity). In this work, we introduce a framework, FedProx, to tackle heterogeneity in federated networks. FedProx can be viewed as a generalization and re-parametrization of FedAvg, the current state-of-the-art method for federated learning. While FedProx makes only minor algorithmic modifications to FedAvg, these modifications have important ramifications both in theory and in practice. Theoretically, we provide convergence guarantees for our framework when learning over data from non-identical distributions (statistical heterogeneity), and while adhering to device-level systems constraints by allowing each participating device to perform a variable amount of work (systems heterogeneity). Practically, we demonstrate that FedProx allows for more robust convergence than FedAvg across a suite of federated datasets. In particular, in highly heterogeneous settings, FedProx demonstrates significantly more stable and accurate convergence behavior relative to FedAvg-improving absolute test accuracy by 22% on average.1 Privacy is a third key challenge in the federated setting. While not the focus of this work, standard privacy-preserving approaches such as differential privacy and secure multiparty communication can naturally be combined with the methods proposed herein-particularly since our framework proposes only lightweight algorithmic modifications to prior work.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in Wuhan, China, at the end of 2019, and there are currently no specific antiviral treatments or vaccines available. SARS-CoV-2 has been shown to use the same cell entry receptor as SARS-CoV, angiotensin-converting enzyme 2 (ACE2). In this report, we generate a recombinant protein by connecting the extracellular domain of human ACE2 to the Fc region of the human immunoglobulin IgG1. A fusion protein containing an ACE2 mutant with low catalytic activity is also used in this study. The fusion proteins are then characterized. Both fusion proteins have a high binding affinity for the receptor-binding domains of SARS-CoV and SARS-CoV-2 and exhibit desirable pharmacological properties in mice. Moreover, the fusion proteins neutralize virus pseudotyped with SARS-CoV or SARS-CoV-2 spike proteins in vitro. As these fusion proteins exhibit cross-reactivity against coronaviruses, they have potential applications in the diagnosis, prophylaxis, and treatment of SARS-CoV-2.

All-component 3D-printed lithium-ion batteries are fabricated by printing graphene-oxide-based composite inks and solid-state gel polymer electrolyte. An entirely 3D-printed full cell features a high electrode mass loading of 18 mg cm(-2) , which is normalized to the overall area of the battery. This all-component printing can be extended to the fabrication of multidimensional/multiscale complex-structures of more energy-storage devices.

For the first time, two types of highly anisotropic, highly transparent wood composites are demonstrated by taking advantage of the macro-structures in original wood. These wood composites are highly transparent with a total transmittance up to 90% but exhibit dramatically different optical and mechanical properties.

We report a draft sequence for the genome of the domesticated silkworm (Bombyx mori), covering 90.9% of all known silkworm genes. Our estimated gene count is 18,510, which exceeds the 13,379 genes reported for Drosophila melanogaster. Comparative analyses to fruitfly, mosquito, spider, and butterfly reveal both similarities and differences in gene content.

No abstract