基于TorchMD的粗粒化分子动力模拟研究

粗粒化模型通过简化原子性质以及原子间的相互作用实现生物大分子长时间尺度的分子动力学模拟. 深度学习通过模拟人类的认知过程实现海量数据的准确分类和回归过程. 本论文将这两种技术进行融合,利用基于深度学习的粗粒化分子动力学模拟技术研究分子在不同状态之间的变化过程,并提出基于TorchMD的分子动力学模拟的分析框架. 在本工作中,MFDP聚类算法被用于在三维的CV变量空间中进行聚类,并确定分子的若干主要状态,在完成聚类的同时,给出各类中的代表分子构象,并给出类之间的分子构象. 这为后续利用String算法分析分子在不同状态间的转换路径打下基础. 通过String算法,迭代搜索得到分子在不同状态之间的变化路径以及对应的势能变化曲线. 通过与已有文献的结果进行对比,验证了基于TorchMD的粗粒化分子动力学模拟的理论框架可以在相对较短的时间尺度里研究分子的变化过程.     

CO2 emission control in new CM car-following model with feedback control of the optimal estimation of velocity difference under V2X environment

A new coupled map car-following model in this paper is proposed by considering the influence of the difference of the estimated optimal speed based on the coupled map (CM) car-following model under V2X environment. The stability of the new model is analyzed by applying the control theory, and the conditions are obtained for the stability of the traffic system. And the two scenes of vehicle stopping once and four times have been simulated. The simulation results show that the control term considered with optimal estimation of speed difference can effectively improve the stability of vehicle running and reduce CO₂ emissions in the CM car-following model.     

大气中气相和颗粒相三氟乙酸浓度测定

建立了我国大气中气相和颗粒相三氟乙酸(Trifluoroacetic acid,TFA)的采集和分析方法。采用环形扩散管-滤膜联用装置分离气相和颗粒相,利用环形扩散管的碱性涂层吸附气相TFA,石英滤膜吸附颗粒相物质。对气相和颗粒相样品分别处理,以2,4-二氟苯胺作为衍生剂,与TFA反应生成TFA的苯胺产物,采用GC/MS进行分析。本方法在0.31～4.91μg/L浓度范围内呈线性关系(R2=0.9991),检出限为66 ng/L。采样装置回收率为(101±3)%,当采样量为48 m3,TFA大气浓度检出限为31 pg/m3。于2012年4～10月在北京大学采样点采集大气,测得其中TFA总浓度在501～7447 pg/m3范围,TFA在气相中的浓度大于在颗粒相中的浓度,气固分配系数Kp随温度变化。     

微波加热和粉碎电磁波加热的区别

微波加热过程主要利用微波的波动性,在加热过程中伴随着趋肤效应.粉碎电磁波加热过程主要利用电磁波的粒子性,在加热过程中的趋肤效应变得很小,同时粒子穿透过程变得很强.在实验过程中还发现有电子浓度起伏的金属表面而会使周围空气温度降低的电子致冷现象.     

预脉冲在毛细管快放电软x射线激光中的作用

在一台毛细管快放电软x射线激光实验装置上，在相同主脉冲条件下（电流峰值18—30kA，半周期80ns），通过观测放电产生的软x射线辐射，研究了该装置固有的高幅值（2—5kA）和外加的低幅值（10—20A）两种预脉冲，对聚乙烯毛细管和高纯度陶瓷毛细管（99.9%）放电的管壁烧蚀及等离子体状态的影响.采用装置固有的几kA预脉冲和聚乙烯毛细管，放电 过程中产生了大量的管壁烧蚀，并且这种情况下的等离子体均匀性差，没有可能获得激光输 出.而采用20A的预脉冲和高纯度陶瓷毛细管，管壁烧蚀量大大减少，预电离等离子体的均匀 性好，在这种情况下，实验上利用x射线二极管观测到了激光尖峰信号. 关键词： 预脉冲 毛细管放电 软x射线激光     

单光子纠缠态的纠缠浓缩(英文)

通过分析光学分束器对单光子态的作用特点,描述了一个利用单光子态和真空态制备纠缠单光子态的方法;提出了一个实现单光子纠缠态纠缠浓缩的方案。在这个方案中,两个单光子部分纠缠态被用来作为量子信道,通过利用光学分束器作用和单光子探测器探测完成了这个纠缠浓缩的过程。结果表明,对于单光子纠缠态,利用这种方法总有一定的几率可以从部分纠缠纯态中提纯出最大纠缠态。     

香水中邻苯二甲酸酯总量测定及其暴露评估

建立了快速测定香水中邻苯二甲酸酯(PAEs)的前处理方法。将香水中的PAEs水解为邻苯二甲酸(PA),再利用磷酸三丁酯萃取酸化液中的PA,使用高效液相色谱-二极管阵列检测器定量分析,并评估了PAEs的暴露量。实验考察并优化了影响PAEs水解率的因素,包括KOH的浓度和体积、乙醇的体积、水解时间和温度,优化后的水解条件:10 mL 4 mol/L KOH,1 mL乙醇,水解时间20 min,水解温度80℃。目标分析物的线性范围为3~240μmol/L(R2=0.999 1),LOD和LOQ分别为4.6μmol/kg和5.9μmol/kg,加标回收率为83.4%~92.7%,日内和日间精密度(RSD)不大于6.8%(n=5)。选择35种香水进行测定,PAEs总量范围在LOD~77.738 mmol/kg之间,成年女性经由香水途径接触PAEs的最大暴露水平为0.474 2μg/(kg·d)。该方法前处理过程简单,可靠性高,适用范围广,可以作为检测香水中PAEs的新选择。     

A new traffic model with the consideration of coupling effect with two lanes

This paper introduces a new two-lane high-order continuum model by embedding the two delay time scales continuum traffic model presented by Xue (2003 Phys. Rev. E 68 066123) into the multi-lane model proposed by Daganzo (1997 Transpn. Res. B 31 83) with the consideration of the coupling effect between the vehicles of two lanes in instantaneous traffic situation and lane-change effect. In the novel model, the coupling effect of two lanes and phenomena of lane change, which were not discussed in Daganzo's model and Xue's model, are taken into account. Numerical simulation shows that it is in accordance with real traffic flow. This obviously indicates that the new phenomenon and behaviour are analogous results as single lane presented by Xue, and the proposed model is more reasonable on two lanes. Furthermore, the generation rate between two lanes is also investigated. The results show that the formation and diffusion of traffic shock wave can be better simulated on two lanes.     

A viscous continuum traffic flow model with consideration of the coupling effect for two-lane freeways

In this paper, the viscous continuum traffic flow model for a single lane is extended to the traffic flow for two-lane freeways. The proposed model is a higher-order continuum model considering the coupling and lane changing effects of the vehicles on two adjacent lanes. It results from integrating the Taylor series expansion of the viscous continuum traffic flow model proposed by Ge (2006 Physica A 371 667) into the multi-lane model presented by Daganzo (1997 Transpn. Res. B 31 83). Our proposed model may be used to describe non-anisotropic behaviour because of lane changing in multi-lane traffic. A linear stability analysis is given and the neutral stability condition is obtained. Also, issues related to lane changing, shock waves and rarefaction waves, local clustering and phase transition are investigated through a simulation experiment. The simulation results show that the proposed model is capable of explaining some particular traffic phenomena commonly observable in real world traffic flow.     

Multiple car-following model of traffic flow and numerical simulation

On the basis of the full velocity difference (FVD) model, an improved multiple car-following (MCF) model is proposed by taking into account multiple information inputs from preceding vehicles. The linear stability condition of the model is obtained by using the linear stability theory. Through nonlinear analysis, a modified Korteweg-de Vries equation is constructed and solved. The traffic jam can thus be described by the kink--antikink soliton solution for the mKdV equation. The improvement of this new model over the previous ones lies in the fact that it not only theoretically retains many strong points of the previous ones, but also performs more realistically than others in the dynamical evolution of congestion. Furthermore, numerical simulation of traffic dynamics shows that the proposed model can avoid the disadvantage of negative velocity that occurs at small sensitivity coefficients λ in the FVD model by adjusting the information on the multiple leading vehicles. No collision occurs and no unrealistic deceleration appears in the improved model.