Hyperparameter on-line learning of stochastic resonance based threshold networks

Aiming at training the feed-forward threshold neural network consisting of nondifferentiable activation functions, the approach of noise injection forms a stochastic resonance based threshold network that can be optimized by various gradient-based optimizers. The introduction of injected noise extends the noise level into the parameter space of the designed threshold network, but leads to a highly non-convex optimization landscape of the loss function. Thus, the hyperparameter on-line learning procedure with respective to network weights and noise levels becomes of challenge. It is shown that the Adam optimizer, as an adaptive variant of stochastic gradient descent, manifests its superior learning ability in training the stochastic resonance based threshold network effectively. Experimental results demonstrate the significant improvement of performance of the designed threshold network trained by the Adam optimizer for function approximation and image classification.     

Some Theorems on Leibniz n-Algebras from the Category Un(Lb)

We study the Leibniz n-algebra Un(∑),whose multiplication is defined viathe bracket of a Leibniz algebra ∑ as[x1,...,xn]=[x1,[...,[xn-2,[xn-1,xn]]...]].Weshow that Un(∑) is simple if and only if ∑ is a simple Lie algebra.An analog of Levi'stheorem for Leibniz algebras in Un(Lb) is established and it is proven that the Leibnizn-kernel of Un(Σ) for any semisimple Leibniz algebra Σ is the n-algebra Un(Σ).     

Numerical study of unsteady flow and heat transfer of circular tangential direction jets flowing over the inner cylinder surface in the annular chamber

In order to understand the dynamics of vortices on heat transfer, the unsteady flow field of tangential direction jets flowing in the annular chamber is numerically investigated by scale-adaptive simulation (SAS). The jet Reynolds number is 332,000 based on the jet’s diameter and inflow velocity for a specific geometric model. The analogy theory is used to obtain the convective heat transfer coefficient distribution on the hub surface. Spectral analysis via fast Fourier transform (FFT) is used to analyze frequency information that flows inside the chamber. The proper orthogonal decomposition (POD) method is performed on the velocity field in the chamber and the convective heat transfer coefficient on the hub surface using a snapshot method. The fast Fourier transform helps find the dominant frequency of the unsteady flow in the chamber. The time sequence of velocity fields on the radial plane shows the presence of cyclic flapping of the jet. The proper orthogonal decomposition analysis indicates that the unsteady periodic flow phenomenon in the chamber and unsteady heat transfer on the hub surface are mainly related to the dynamics of the counter-rotating vortices caused by the jet.     

基于CEEMD的重大事件对香港住宅价格影响的实证分析

应用互补集成经验模态分解（CEEMD）方法对香港1997―2018年的住宅价格月度数据进行了分解，将经过重构后的数据分成高频序列、低频序列与残差项。将BP多断点检测应用于低频序列，并结合样本时段内的重大事件进行实证分析。结果表明：1997年亚洲金融风暴对房价的影响大于2008年金融危机；外部经济体的救市政策间接地影响香港房价；在经济不景气的大环境下“孙九招”政策没有立即见效；资本投资者入境计划、住房供给调整与按揭贷款调整对房价的影响较为显著；税收调整对房价影响不显著、对交易量影响显著；SARS爆发使住宅价格下降约1%。     

Could chroman-4-one derivative be a better inhibitor of PTR1? – Reason for the identified disparity in its inhibitory potency in Trypanosoma brucei and Leishmania major

Most notable Kinetoplastids are of the genus Trypanosoma and Leishmania, affecting several millions of humans in Africa and Latin America. Current therapeutic options are limited by several drawbacks, hence the need to develop more efficacious inhibitors. An investigation to decipher the mechanism behind greater inhibitory potency of a chroman-4-one derivative (compound 1) in Trypanosoma brucei pteridine reductase 1 (TbPTR1) and Leishmania major pteridine reductase 1 (LmPTR1) was performed. Estimation of ΔG_bind revealed that compound 1 had a greater binding affinity in TbPTR1 with a ΔG_bind value of −49.0507 Kcal/mol than −29.2292 Kcal/mol in LmPTR1. The ΔGbind in TbPTR1 were predominantly contributed by “strong” electrostatic energy compared to the “weak” van der Waals in LmPTR1. In addition to this, the NADPH cofactor contributed significantly to the total energy of TbPTR1. A characteristic weak aromatic π interaction common in PTR1 was more prominent in TbPTR1 than LmPTR1. The consistent occurrence of high-affinity conventional hydrogen bond interactions as well as a steady interaction of crucial active site residues like Arg14/Arg17, Ser95/Ser111, Phe97/Phe113 in TbPTR1/LmPTR1 with chroman-4-one moiety equally revealed the important role the moiety played in the activity of compound 1. Overall, the structural and conformational analysis of the active site residues in TbPTR1 revealed them to be more rigid than LmPTR1. This could be the mechanism of interaction TbPTR1 employs in exerting a greater potency than LmPTR1. These findings will further give insight that will be assistive in modifying compound 1 for better potency and the design of novel inhibitors of PTR1.     

The spin-isospin decomposition of the nuclear symmetry energy from low to high density

The energy per particle BA in nuclear matter is calculated up to high baryon density in the whole isospin asymmetry range from symmetric matter to pure neutron matter.The results,obtained in the framework of the Brueckner-Hartree-Fock approximation with two-and three-body forces,confirm the well-known parabolic dependence on the asymmetry parameterβ=(N?Z)/A(β^2 law)that is valid in a wide density range.To investigate the extent to which this behavior can be traced back to the properties of the underlying interaction,aside from the mean field approximation,the spin-isospin decomposition of BA is performed.Theoretical indications suggest that theβ^2 law could be violated at higher densities as a consequence of the three-body forces.This raises the problem that the symmetry energy,calculated according to theβ^2 law as a difference between BA in pure neutron matter and symmetric nuclear matter,cannot be applied to neutron stars.One should return to the proper definition of the nuclear symmetry energy as a response of the nuclear system to small isospin imbalance from the Z=N nuclei and pure neutron matter.     

Properties of bounded stochastic processes employed in biophysics

Abstract

Realistic stochastic modeling is increasingly requiring the use of bounded noises. In this work, properties and relationships of commonly employed bounded stochastic processes are investigated within a solid mathematical ground. Four families are object of investigation: the Sine-Wiener (SW), the Doering–Cai–Lin (DCL), the Tsallis–Stariolo–Borland (TSB), and the Kessler–Sørensen (KS) families. We address mathematical questions on existence and uniqueness of the processes defined through Stochastic Differential Equations, which often conceal non-obvious behavior, and we explore the behavior of the solutions near the boundaries of the state space. The expression of the time-dependent probability density of the Sine-Wiener noise is provided in closed form, and a close connection with the Doering–Cai–Lin noise is shown. Further relationships among the different families are explored, pathwise and in distribution. Finally, we illustrate an analogy between the Kessler–Sørensen family and Bessel processes, which allows to relate the respective local times at the boundaries.