Universality of deep convolutional neural networks

Deep learning has been widely applied and brought breakthroughs in speech recognition, computer vision, and many other domains. Deep neural network architectures and computational issues have been well studied in machine learning. But there lacks a theoretical foundation for understanding the approximation or generalization ability of deep learning methods generated by the network architectures such as deep convolutional neural networks. Here we show that a deep convolutional neural network (CNN) is universal, meaning that it can be used to approximate any continuous function to an arbitrary accuracy when the depth of the neural network is large enough. This answers an open question in learning theory. Our quantitative estimate, given tightly in terms of the number of free parameters to be computed, verifies the efficiency of deep CNNs in dealing with large dimensional data. Our study also demonstrates the role of convolutions in deep CNNs.     

Information percolation and cutoff for the random‐cluster model

We consider the random‐cluster model (RCM) on with parameters p∈(0,1) and q ≥ 1. This is a generalization of the standard bond percolation (with edges open independently with probability p) which is biased by a factor q raised to the number of connected components. We study the well‐known Fortuin‐Kasteleyn (FK)‐dynamics on this model where the update at an edge depends on the global geometry of the system unlike the Glauber heat‐bath dynamics for spin systems, and prove that for all small enough p (depending on the dimension) and any q>1, the FK‐dynamics exhibits the cutoff phenomenon at with a window size , where λ_∞ is the large n limit of the spectral gap of the process. Our proof extends the information percolation framework of Lubetzky and Sly to the RCM and also relies on the arguments of Blanca and Sinclair who proved a sharp mixing time bound for the planar version. A key aspect of our proof is the analysis of the effect of a sequence of dependent (across time) Bernoulli percolations extracted from the graphical construction of the dynamics, on how information propagates.     

密度泛函理论研究MN@H_2O (M=Ga,Ge, In,Sn, Sb; N=M,Al)团簇的特性

采用密度泛函理论的B3LYP, B3P86, B1B95, P3PW91和PBE1PBE方法结合SDD, LANL2DZ和CEP-121G基组计算了d~(10)组态二聚物MN(M=Ga, Ge, In, Sn和Sb; N=M和Al)的几何结构.采用B3P86/SDD进一步研究了MN@H_2O团簇的几何结构及吸附能.结果表明,水分子结合在二聚物M_2上时,对二聚物影响较大,对水分子自身影响较小.将M_2中Ga, Ge, In, Sn或Sb替换一个原子为Al时,水分子在GeAl和SnAl上的吸附能变化较大,而在GaAl, InAl和SbAl上吸附能变化较小.另外, H_2O吸附在Ga, Ge, In, Sn和Sb上时,与吸附在Al上时,吸附能的变化不大.     

A DFT study on mechanisms of CO2 coupling with propargylic alcohols using alkali carbonates

The mechanisms of CO₂ coupling with the propargylic alcohol using alkali carbonates M₂CO₃ (M = Li, Na, K, Cs) have been investigated by means of density functional theory calculations. The calculations reveal that the target product tetronic acid (TA) is yielded through two stages: (a) the formation of the α-alkylidene cyclic carbonate (αACC) intermediate via Cs₂CO₃-mediated carboxylative cyclization of the propargylic alcohol with CO₂, and (b) the conversion of the αACC intermediate with Cs₂CO₃ to produce the cesium salt of the TA. Since the overall kinetic barriers for the two stages are comparable and affordable, the excellent chemoselectivity to the TA should be primarily originated from the high thermodynamic stability of the cesium salt of the TA. Moreover, relative to the TA, the possibility to yield the by-product acyclic carbonate can be excluded due to the both kinetics and thermodynamic inferiority. This result is different from the organic base-mediated reaction. Alternatively, our calculations predict that CsHCO₃ together generated with the cesium salt of the TA might also be an available mediating reagent for the incorporation of CO₂ with the propargylic alcohol. Compared to other alkali carbonates M₂CO₃ (M = Li, Na, K), the stronger basicity of Cs₂CO₃ and the lower ionic potential of cesium ion can raise the effective concentration of the αACC intermediate, and thus the conversion of the αACC intermediate into the cesium salt of the TA can be achieved with high yield.     

A new database and benchmark of the bond energies of noble-gas-containing molecules

We have developed a new database of structures and bond energies of 59 noble-gas-containing molecules. The structures were calculated by CCSD(T)/aug-cc-pVTZ methods and the bond energies were obtained using the CCSD(T)/complete basis set method. Many wavefunction-based and density functional theory methods have been benchmarked against the 59 accurate bond energies. Our results show that the MPW1B95, B2GP-PLYP, and DSD-BLYP functionals with the aug-cc-pVTZ basis set excel in predicting the bond energies of noble-gas molecules with mean unsigned errors (MUEs) of 2.0 to 2.1 kcal/mol. When combinations of Dunning's basis sets are used, the MPW1B95, B2GP-PLYP, DSD-BLYP, and BMK functionals give significantly lower MUEs of 1.6 to 1.9 kcal/mol. Doubly hybrid methods using B2GP-PLYP and DSD-BLYP functionals and MP2 calculation also provide satisfactory accuracy with MUEs of 1.4 to 1.5 kcal/mol. If the Ng bond energies and the total atomization energies of a group of 109 main-group molecules are considered at the same time, the MPW1B95/aug-cc-pVTZ single-level method (MUE = 2.7 kcal/mol) and the B2GP-PLYP and DSD-PLYP functionals with combinations of basis sets or using the doubly hybrid method (MUEs = 1.9-2.2 kcal/mol) give the overall best result.     

Designing spirobifullerene core based three-dimensional cross shape acceptor materials with promising photovoltaic properties for high-efficiency organic solar cells

The development of organic electron acceptor materials is one of the key factors for realizing high-performance organic solar cells (OSCs). Nonfullerene electron acceptors, compared to traditional fullerene acceptor materials, have gained much impetus owing to their better optoelectronic tunabilities and lower cost, as well as higher stability. Therefore, 5 three-dimensional (3D) cross-shaped acceptor materials having a spirobifullerene core flanked with 2,1,3-benzothiadiazole are designed from a recently synthesized highly efficient acceptor molecule SF(BR) ₄ and are investigated in detail with regard to their use as acceptor molecules in OSCs. The density functional theory (DFT) and time-dependent DFT (TDDFT) calculations have been performed for the estimation of frontier molecular orbital (FMO) analysis, density of states analysis, reorganization energies of electron and hole, dipole moment, open-circuit voltage, photo-physical characteristics, and transition density matrix analysis. In addition, the structure-property relationship is studied, and the influence of end-capped acceptor modifications on photovoltaic, photo-physical, and electronic properties of newly selected molecules ( H1-H5 ) is calculated and compared with reference ( R ) acceptor molecule SF(BR) ₄ . The structural tailoring at terminals was found to effectively tune the FMO band gap, energy levels, absorption spectra, open-circuit voltage, reorganization energy, and binding energy value in selected molecules H1 to H5 . The 3D cross-shaped molecules H1 to H5 suppress the intermolecular aggregation in PTB7-Th blend, which leads to high efficiency of acceptor material H1 to H5 in OSCs. Consequently, better optoelectronic properties are achieved from designed molecules H1 to H5 . It is proposed that the conceptualized molecules are superior than highly efficient spirobifullerene core-based SF(BR) ₄ acceptor molecules and, thus, are recommended to experiments for future developments of highly efficient solar cells.     

3d过渡金属掺杂M_2@Si_(20)(M=Sc-Zn)团簇的密度泛函理论研究

利用密度泛函理论(DFT)研究3d过渡金属掺杂硅团簇的几何结构和稳定性,计算了绝热电子亲和能和垂直电离能,内嵌双金属间距,自旋磁矩等.结果表明内嵌的Sc、Ti、V、Mn金属二聚体和十二面体硅笼构成了稳定的富勒烯结构,随着d电子数目的增加其内嵌的富勒烯构型有部分畸变,总体而言Si_(20)团簇掺杂双金属后稳定性得到了提高.