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.     

Chemical bonding in representative astrophysically relevant neutral,cation, and anion HCnH chains

Most existing studies assign a polyynic and cumulenic character of chemical bonding in carbon-based chains relying on values of the bond lengths. Building on our recent work, in this paper we add further evidence on the limitations of such an analysis and demonstrate the significant insight gained via natural bond analysis. Presently reported results include atomic charges, natural bond order and valence indices obtained from ab initio computations for representative members of the astrophysically relevant neutral and charged HC_2k/2k+1H chain family. They unravel a series of counter-intuitive aspects and/or help naive intuition in properly understanding microscopic processes, e.g., electron removal from or electron attachment to a neutral chain. Demonstrating that the Wiberg indices adequately quantify the chemical bonding structure of the HC_2k/2k+1H chains—while the often heavily advertised Mayer indices do not—represents an important message conveyed by the present study.     

Generalized Jacobi spectral Galerkin method for fractional pantograph differential equation

This work is concerned with the extension of the Jacobi spectral Galerkin method to a class of nonlinear fractional pantograph differential equations. First, the fractional differential equation is converted to a nonlinear Volterra integral equation with weakly singular kernel. Second, we analyze the existence and uniqueness of solutions for the obtained integral equation. Then, the Galerkin method is used for solving the equivalent integral equation. The error estimates for the proposed method are also investigated. Finally, illustrative examples are presented to confirm our theoretical analysis.     

13C NMR investigations of molecular order of rod-like,bent-core,and thiophene mesogens

In this review, methods to obtain the orientational order of topologically variant molecular mesogens using by one- and two-dimensional (2D) solid-state ¹³C nuclear magnetic resonance (NMR) spectroscopy are described. Besides ¹³C chemical shifts, the ¹³C─¹H dipolar couplings measured from 2D-separated local field (SLF) technique are used for computing the order parameters of a variety of mesogens. The investigated molecules are composed of a variable number of rings in the core, that is, core ranging from simply one ring to five rings. Among the mesogens investigated, a special focus has been placed on mesogens with thiophene rings, which are gaining popularity as liquid crystalline organic semiconductors. The replacement of a phenyl ring by thiophene in the core has a dramatic influence on molecular topology, as observed from the measured order parameters. The review highlights the advantages of the 2D SLF method for understanding the local dynamics and for mapping the topology of mesogens through the measured order parameters. SLF NMR studies of as many as 24 molecular mesogens that vary in terms of the molecular structure as well as topology are covered in the review. Order parameters of the rings have been estimated from the ¹³C─¹H dipolar couplings in the nematic, smectic A, smectic C, and tilted hexatic phases as well as in B₁ and B₂ mesophases of various mesogens. It is anticipated that, in the years to come, the 2D SLF method would provide advanced molecular information on structurally complex mesogens that are emerging in liquid crystal science through the incessant efforts of synthetic chemists. The mini review covers the orientational order of topologically variant molecular mesogens determined by 1D and 2D solid-state ¹³C NMR spectroscopy. Accordingly, rod-like, bent-core, and thiophene mesogens were subjected to 2D SLF measurements to get the order parameters from which the topology was established. The replacement of phenyl ring by thiophene and its influence on order parameters as well as on molecular topology is also discussed.