摘要：含能分子在军事和民用应用中都发挥着重要作用.传统上,确定含能分子的物理化学参数需要实验工作量且具有风险,而新兴的机器学习方法有望解决这一问题.在这项工作中,我们报道了一种分子描述符增强的图神经网络(MD增强的GNN)模型,该模型可以准确快速地预测含能分子的三个爆轰参数.该模型集成了基于序列的分子描述符和基于结构的图向量,提供了一个不需要自定义描述符的全面框架.因此,我们构建了一个包含18,991个CHNO含能分子的含能分子数据集,并将我们的模型与单一的分子指纹/描述符和GNN方法进行了比较.研究发现,我们提出的MD增强的GNN集成方法通过结合两个不同的互补特征,实现了卓越的精度,R2超过0.93,学习速度提高了20%以上,这突出了我们的模型在重塑含能分子设计格局方面的潜力,并有望在这一关键领域内大幅提高效率和有效性.

摘要：The design of electrocatalysts with enhanced adsorption and activation of nitrogen(N2)is critical for boosting the electrochemical N2reduction(ENR).Herein,we developed an efficient strategy to facilitate N2 adsorption and activation for N2 electroreduction into ammonia(NH3)by vacancy engineering of core@shell structured Au@Sn O2 nanoparticles(NPs).We found that the ultrathin amorphous SnO2 shell with enriched oxygen vacancies was conducive to adsorb N2as well as promoted the N2 activation,meanwhile the metallic Au core ensured the good electrical conductivity for accelerating electrons transport during the electrochemical N2 reduction reaction,synergistically boosting the N2 electroreduction *** confirmed by the15N-labeling and controlled experiments,the core@shell Au@amorphous SnO2 NPs with abundant oxygen vacancies show the best performance for N2 electroreduction with the NH3 yield rate of 21.9 lg h-1mg-1catand faradaic efficiency of 15.2%at-10.2 VRHE,which surpass the Au@crystalline SnO2 NPs,individual Au NPs and all reported Au-based catalysts for ENR.

摘要：Among all CO2 electroreduction products,methane(CH4)and ethylene(C_(2)H_(4))are two typical and valuable hydrocarbon products which are formed in two different pathways:hydrogenation and dimerization reactions of the same CO *** studies show that the adsorption configurations of CO intermediate determine the reaction pathways towards CH4/C_(2)H_(4).However,it is challenging to experimentally control the CO adsorption configurations at the catalyst surface,and thus the hydrocarbon selectivity is still ***,we seek to synthesize two well-defined copper nanocatalysts with controllable surface *** two model catalysts exhibit a high hydrocarbon selectivity toward either CH4(83%)or C_(2)H_(4)(93%)under identical reduction *** transmission electron microscopy and X-ray absorption spectroscopy characterizations reveal the low-coordination Cu^(0)sites and local Cu^(0)/Cu^(+)sites of the two catalysts,***-temperature programed desorption,in-situ attenuated total reflection Fourier transform infrared spectroscopy and density functional theory studies unveil that the bridge-adsorbed CO(CO_(B))on the low-coordination Cu^(0)sites is apt to be hydrogenated to CH4,whereas the bridge-adsorbed CO plus linear-adsorbed CO(CO_(B)+CO_(L))on the local Cu^(0)/Cu^(+)sites are apt to be coupled to C_(2)H_(4).Our findings pave a new way to design catalysts with controllable CO adsorption configurations for high hydrocarbon product selectivity.

摘要：随着信息技术的快速发展,各行业领域的数据大量涌现,以数据驱动的机器学习模型已广泛应用于人们的日常工作和生活中。准确性和泛化能力是机器学习模型的2个重要指标,这需要从大量有用的数据中进行学习。在一些机器学习应用场景中,各个数据来源方无法直接进行数据交换,严重影响了机器学习模型的能力提升。针对数据隐私泄露风险问题,本文设计并实现了一种基于联邦平均的差分隐私保护算法,兼顾学习训练效率和数据隐私保护能力。实验结果表明,本文算法在牺牲了一部分通信效率的情况下,可达到较好的数据隐私保护能力。

摘要：Recently,the newly booming metal halide perovskites have attracted extensive attention worldwide due to their outstanding optoelectronic performance,and are expected to be ideal candidates for photodetectors(PDs).However,there is still lack of perovskite PDs-based imaging devices coming into commercialization stage,due to some practical reasons including toxicity brought by lead-based perovskites and the large light current *** this paper,for the first time we fabricate a lead-free Cs3Bi2Br9 perovskite PD,and build a prototype of this perovskite PD-based imaging system with diffuse reflection imaging ***,we propose a new parameter F related to light current fluctuation to evaluate imaging performance of a PD especially for weak diffuse light condition,and prove its usability by comparison of unoptimized lead-free Cs3Bi2Br9 perovskite PD and atomic layer deposition(ALD)optimized Cs3Bi2 Br9 ***-optimization can improve the quality of perovskite film and suppress the dark current and current ***,we obtain satisfactory diffuse reflection images of 2D and 3D objects with wide dynamic ***,the ALD-optimized Cs3Bi2Br9PD has addressed two major concerns about perovskite PDs-based imaging devices,that may extend application of perovskite materials and improve imaging quality.

摘要：To solve the load balancing problem in a triplet-based hierarchical interconnection network（THIN） system, a dynamic load balancing （DLB）algorithm--THINDLBA, which adopts multicast tree （MT）technology to improve the efficiency of interchanging load information, is presented. To support the algorithm, a complete set of DLB messages and a schema of maintaining DLB information in each processing node are designed. The load migration request messages from the heavily loaded node （HLN）are spread along an MT whose root is the HLN. And the lightly loaded nodes（LLNs） covered by the MT are the candidate destinations of load migration; the load information interchanged between the LLNs and the HLN can be transmitted along the MT. So the HLN can migrate excess loads out as many as possible during a one time execution of the THINDLBA, and its load state can be improved as quickly as possible. To avoid wrongly transmitted or redundant DLB messages due to MT overlapping, the MT construction is restricted in the design of the THINDLBA. Through experiments, the effectiveness of four DLB algorithms are compared, and the results show that the THINDLBA can effectively decrease the time costs of THIN systems in dealing with large scale computeintensive tasks more than others.

摘要：使用Java语言设计了一个简便的实现药物化学实验虚拟化的虚拟实验设备装配引擎——VAEPCE。主要介绍了VAEPCE中药物化学类虚拟实验装置和器件的内部数据表示、虚拟装配过程关键技术和连接正确性检查的实现原理。使用VAEPCE可以方便地设计虚拟装置、定制虚拟器件、实现虚拟装配和标准装配流程展示等功能。

摘要：片上网络中路由算法的设计对芯片的性能有直接的影响.本文针对TriBA-Net网络提出一种新颖的最短路径.路由算法.首先,基于TriBA-Net网络设计了一种编码方法,该编码方法中所用到的文字1,3,2的集合与群论中的三文字集S_3群具有相同的含义.其次,设计了一种相隔节点间的通信模型,根据通信路径端点的可能状态,将通信划分为6种宏观数据流动模式.最后,利用S_3群的循环置换特性对通信模型进行简化,在XC6VLX550TL芯片上完成了SPR4T路由器的设计实现.实验结果表明,在27个节点的TriBA-Net网络性能测试中,在均衡负载模式下,与SPORT路由算法相比,SPR4T路由算法的饱和注入率提升7.5%,吞吐率提升7.7%,而且有效降低了硬件以及功耗的开销.

摘要：全息波导耦合元件是增强现实显示系统的关键光学元件,光束垂直入射至全息波导板中通过入耦合元件改变光束传播方向,在波导板中形成波导现象传输至出耦合元件处,通过出耦合元件再次改变传输方向从全息波导板出射至人眼。当入耦合元件与出耦合元件均为透射式或反射式耦合元件时,两次改变光束传播方向的角度值相等。根据全息波导的成像原理,结合k矢量圆理论与全息光学记录原理,设计了全息波导入耦合元件与全息波导出耦合元件的对称型周期结构。利用MATLAB软件模拟了全息波导耦合元件记录时的干涉光强模型,实验结果与所设计的耦合元件周期结构数据吻合;利用COMSOL软件模拟了全息波导耦合元件对光束传输角度的改变过程,实验结果与理论改变光束传播方向的角度值一致。实验证明了两个全息波导耦合元件以互为对称的结构排布在全息波导光学系统中,能够满足全息波导光学系统对光束传播方向的要求。