高通量计算与深度学习相结合的稠环含能化合物设计

含能化合物的设计效率决取于多方面因素,如筛选空间中潜在高性能样本的占比和关键性能的准确预测方法。本研究提出预筛选分子骨架提升虚拟筛选空间整体性能的方案,并将高通量计算与深度学习相结合用于含能化合物设计。研究发现,含能分子的晶体密度与其骨架密度之间存在中度的正相关性,通过预筛选高密度分子骨架可以有效提升虚拟筛选空间的整体密度。研究基于晶体学数据库CCDC提供的含能晶体密度数据集,采用深度学习方法获得含能晶体的密度预测模型,具有可靠的精度和泛化性。在此基础上,以稠环类含能化合物为研究对象,通过骨架预筛选获得高密度的稠环分子骨架,从而通过分子片段组装获得由潜在的高密度分子组成的虚拟筛选空间。研究采用量子化学计算和爆轰产物状态方程等方法实现了生成焓、爆轰性能和化学稳定性的预测,从而由性能排序筛选出能量水平优于RDX,稳定性优于TNT的新型含能分子6个。研究表明,分子骨架预筛选可以有效提升虚拟筛选空间的总体性能,在此基础上借助高通量计算与深度学习可实现含能分子的高效设计。

Physical Simulation for Large-Bore Copper Vapor Laser

On the basis of the electric circuit differential equations and the nonlinear partial differential equations of kinetics,a kinetic model for a series of large-bore copper-vapor lasers is carried out.Through the consistent computational results the radial kinetic parameters such as the stimulating electric Geld and the plasma electron temperature are presented.The factors,which evidently influence the output power and the beam quality of the large-bore copper-vapor laser,are the buffer gas pressure and the charging voltage.

A SUPER-BROADENED STIMULATED SCATTERING IN LIQUID-FILLED WAVEGUIDE WITH LONG GAIN LENGTH

Using an intense visible laser pulse to excite the liquid-filled waveguide with a long gain length,we have observed a new stimulated-scattering with super-broadened spectral width(500cm^(-1))in addition to the sharp lines of SRS of the benzene or carbon-disulfide which is used as a filling liquid.