Magnitude Reduction In Run Time Research Articles

Energy Efficiency Maximization in Large-Scale Cell-Free Massive MIMO: A Projected Gradient Approach

This paper considers the fundamental power allocation problem in cell-free massive mutiple-input and multiple-output (MIMO) systems which aims at maximizing the total energy efficiency (EE) under a sum power constraint at each access point (AP) and a quality-of-service (QoS) constraint at each user. Existing solutions for this optimization problem are based on solving a sequence of second-order cone programs (SOCPs), whose computational complexity scales dramatically with the network size. Therefore, they are not implementable for practical large-scale cell-free massive MIMO systems. To tackle this issue, we propose an iterative power control algorithm based on the frame work of an accelerated projected gradient (APG) method. In particular, each iteration of the proposed method is done by simple closed-form expressions, where a penalty method is applied to bring constraints into the objective in the form of penalty functions. Finally, the convergence of the proposed algorithm is analytically proved and numerically compared to the known solution based on SOCP. Simulations results demonstrate that our proposed power control algorithm can achieve the same EE as the existing SOCPs-based method, but more importantly, its run time is much lower (one to two orders of magnitude reduction in run time, compared to the SOCPs-based approaches).

MoDL-MUSSELS: Model-Based Deep Learning for Multishot Sensitivity-Encoded Diffusion MRI.

We introduce a model-based deep learning architecture termed MoDL-MUSSELS for the correction of phase errors in multishot diffusion-weighted echo-planar MR images. The proposed algorithm is a generalization of the existing MUSSELS algorithm with similar performance but significantly reduced computational complexity. In this work, we show that an iterative re-weighted least-squares implementation of MUSSELS alternates between a multichannel filter bank and the enforcement of data consistency. The multichannel filter bank projects the data to the signal subspace, thus exploiting the annihilation relations between shots. Due to the high computational complexity of the self-learned filter bank, we propose replacing it with a convolutional neural network (CNN) whose parameters are learned from exemplary data. The proposed CNN is a hybrid model involving a multichannel CNN in the k-space and another CNN in the image space. The k-space CNN exploits the annihilation relations between the shot images, while the image domain network is used to project the data to an image manifold. The experiments show that the proposed scheme can yield reconstructions that are comparable to state-of-the-art methods while offering several orders of magnitude reduction in run-time.

Ultra-wideband terahertz graphene absorber using circuit model

In this work, a novel solution for designing an ultra-wideband terahertz graphene-based absorber by exploiting the periodic arrays of graphene ribbons (PAGRs) and graphene continuous sheet (GCS) is presented. Here we develop and extend a transmission line method to describe the absorber structure as circuit elements. Regarding impedance-matching concept and using analytical circuit model of PAGRs and GCS which recently have been proposed, different PAGRs have been used to gain ultra-wide bandwidth. As a result, a bandwidth of 90% absorption is extended up to 130% and 146% of the central frequency with three and four layers of graphene, respectively. Excellent performance of the proposed method regarding computation time (showing more than five orders of magnitude reduction in runtime) and memory sources besides producing results with an acceptable agreement with the results of full-wave simulations (with an error less than 5%), nominates it as a promising approach to design and simulate other sub-wavelength graphene-based devices and; moreover, since the proposed structure is a multi-layer one, it can be manufactured by chemical vapor deposition.

Tunable ultra-wideband terahertz absorber based on graphene disks and ribbons

Terahertz frequency band has become one of the most interesting fields of research these days and the necessity of various modern applications in this area is THz absorbers. In this paper, we have proposed and investigated a novel ultra-wideband terahertz absorber by using periodic arrays of graphene disks and ribbons. Analytical circuit model of graphene arrays besides a developed transmission line theory is used to obtain analytical expressions for the input impedance of the proposed structure. The input impedance is adjusted to be closely matched to the free space impedance in a wide frequency range. Using three layers of patterned graphene, bandwidth of 90% absorption reached to 131% of central frequency. Regarding excellent performance of the proposed method in terms of computation time (showing more than 5 orders of magnitude reduction in runtime) and memory sources, this method with producing results with an acceptable agreement (with an error less than 5%) compared to the results of full-wave numerical modeling can be utilized in design of other graphene-based sub-wavelength devices.

Filtering Mixing Planes for Low Reduced Frequency Analysis of Turbomachines

A class of problems in turbomachinery is characterized by unsteady interactions at low reduced frequencies. These interactions are often the result of perturbations with length-scale on the order of the machine circumference and examples include axial compressors operating with inlet distortion, fans with downstream pylons, and turbine rotors downstream of midframe struts. Typically, this unsteadiness is accompanied by higher frequency fluctuations caused by perturbations with a length-scale on the order of a blade pitch. Conventional numerical analysis of this class of problem requires computations with a time step governed by the high-frequency content but a greatly reduced run time could be achieved if the time step was dictated solely by the low reduced frequency, long length-scale, interaction of interest. In this paper, a filtering mixing plane technique is proposed that removes unwanted short length-scale perturbations at the interfaces between blade rows. This approach gives the user control over the amount of mixing that occurs at these interfaces with the limits being fully mixed-out to pitchwise uniformity (conventional mixing plane) or no mixing (conventional sliding plane). By choosing to retain only enough harmonics to resolve the low reduced frequency interaction of interest, an order of magnitude reduction in run time can be achieved.

Circuit modeling of graphene absorber in terahertz band

Here we develop and extend a transmission line method (TLM) to analyze the performance of graphene assisted metamaterial (GM) devices working in the terahertz (THz) band. We demonstrate that a circuit model can be presented for different parts of the device including graphene and also the patterned metallic sheet by analyzing the distribution of surface induced current. In pursuit of evaluating the efficiency and accuracy of our proposed method, we compare its results, obtained from an easy to implement MATLAB code for a typical GM absorber with those obtained from full wave simulations. The excellence of the proposed method in terms of computation time (showing more than 3 orders of magnitude reduction in run time) and memory resource besides producing results with acceptable agreement with the results of full wave simulation (with an error less than 5%) versus incident angle, dielectric thickness and chemical potential, nominates it as a promising approach to simulate other graphene-based devices.

Multicore Parallelization of Kohn-Sham Theory.

A multicore parallelization of Kohn-Sham theory is described, using standard commodity multisocket and multisocket/multicore shared-memory processors. Near-linear scaling of the parallel parts of the code was observed up to the maximum of sixteen cores that were available for benchmarking, and an order of magnitude reduction in run time was achieved running using sixteen threads on a quad-socket quad-core Xeon system. The speed-ups achieved using multisocket/multicore processors were competitive with those achieved using numerical accelerator cards.