Parallel implementations of the Complex-RF algorithm

Low-dimension derivative-free optimization problems are common in many engineering applications. Usefulness is often limited by long evaluation times due to large simulation models. For such problems, direct-search algorithms often outperform the naturally parallel population-based methods. While direct-search algorithms are more difficult to parallelize, there are many unexploited opportunities. Three methods for parallelizing the Complex-RF algorithm have been implemented and evaluated. Numerical analysis of the algorithm has been performed. This provides a basis for parametrization of the parallel methods. The methods are tested on two standard test functions with five variables and one real simulation model with eight variables. An entropy rate based performance index is used to compare the methods. Experiments show performance increases ranging from 3.9 to 6.4 depending on the model. The suggested methods outperform both a particle swarm and a differential evolution algorithm with up to 32 threads. When more threads are added, parallelization efficiency decreases.     

Alternative to the karatsuba algorithm for software implementations of GF(2n) multiplications

A new approach to subquadratic space complexity GF(2ⁿ) multipliers has been proposed recently. The corresponding algorithm for software implementations is developed. While its recursive implementation is as simple as that of the Karatsuba algorithm, it requires much less memory to store the look-up table. Therefore it is quite suitable for memory-constrained applications, for example smart cards.     

New implementations for the Simultaneous-FETI method

In this work, we present alternative implementations for the Simultaneous-FETI (S-FETI) method. Developed in recent years, this method has shown to be very robust for highly heterogeneous problems. However, the memory cost in S-FETI is greatly increased and can be a limitation to its use. Our main objective is to reduce this memory usage without losing significant time performance. The algorithm is based on the exploitation of the sparsity patterns found on the block of search directions, allowing to store less vectors per iteration in comparison to a full storage scheme. In addition, different variations for the S-FETI method are also proposed, including a new treatment for the possible dependencies between directions and the use of the Lumped preconditioner. Several tests are performed in order to establish the impact of the modifications presented in this work compared to the original S-FETI algorithm.     

Parallel unidirectional division algorithms and implementations parallel unidirectional division algorithms and implementations

Abstract

This paper describes the design of algorithms for unidirectional division. Rather than the conventional, restoring or nonrestoring, divisions which require both subtraction and addition operations, alternately, during the process of division, the proposed method needs only one direction, either subtraction or addition, but not both, during the process of the division. If both operands, dividend and divisor, are positive or negative, only subtraction is used. However, if one is positive and the other negative, only addition is applied. This method can skip zero bits in dividend, and consequently the number of additions/subtractions is expected to be less than conventional division, about (3/5)n compared to 3/2n (or n if MUXes are used) for restoring or to n for nonrestoring. In addition, unidirectional division can be processed in parallel or in semi‐parallel, if the bit length of the dividend is long enough. This method is also easily extended to apply to two's complement divisions. The conversion of the quotient to binary is not required for the proposed method, (in which nonrestoring division is necessarily converted back to binary code in order to match other operations, because the resulting quotient is expressed by signed‐digit code).     

State-of-the-art survey on digital twin implementations

Digital twin (DT) has garnered attention in both industry and academia. With advances in big data and internet of things (IoTs) technologies, the infrastructure for DT implementation is becoming more readily available. As an emerging technology, there are both potential and challenges. DT is a promising methodology to leverage the modern data explosion to aid engineers, managers, healthcare experts and politicians in managing production lines, patient health and smart cities by providing a comprehensive and high fidelity monitoring, prognostics and diagnostics tools. New research and surveys into the topic are published regularly, as interest in this technology is high although there is a lack of standardization to the definition of a DT. Due to the large amount of information present in a DT system and the dual cyber and physical nature of a DT, augmented reality (AR) is a suitable technology for data visualization and interaction with DTs. This paper seeks to classify different types of DT implementations that have been reported, highlights some researches that have used AR as data visualization tool in DT, and examines the more recent approaches to solve outstanding challenges in DT and the integration of DT and AR.The full text can be downloaded at https://link.springer.com/article/10.1007/s40436-021-00375-w     

Chaos computing: ideas and implementations

We review the concept of the 'chaos computing' paradigm, which exploits the controlled richness of nonlinear dynamics to obtain flexible reconfigurable hardware. We demonstrate the idea with specific schemes and verify the schemes through proof-of-principle experiments.     

On reaction network implementations of neural networks

This paper is concerned with the utilization of deterministically modelled chemical reaction networks for the implementation of (feed-forward) neural networks. We develop a general mathematical framework and prove that the ordinary differential equations (ODEs) associated with certain reaction network implementations of neural networks have desirable properties including (i) existence of unique positive fixed points that are smooth in the parameters of the model (necessary for gradient descent) and (ii) fast convergence to the fixed point regardless of initial condition (necessary for efficient implementation). We do so by first making a connection between neural networks and fixed points for systems of ODEs, and then by constructing reaction networks with the correct associated set of ODEs. We demonstrate the theory by constructing a reaction network that implements a neural network with a smoothed ReLU activation function, though we also demonstrate how to generalize the construction to allow for other activation functions (each with the desirable properties listed previously). As there are multiple types of ‘networks’ used in this paper, we also give a careful introduction to both reaction networks and neural networks, in order to disambiguate the overlapping vocabulary in the two settings and to clearly highlight the role of each network’s properties.     

Design and efficient implementations of predictive FIR filters

Two new approaches to the design of predictive FIR filters are presented. First, we discuss the design of predictors and estimators for narrow-band signals based on the interpolated FIR (IFIR) filter approach. The transfer function of the IFIR predictor is of the form H(z)=P(z^L)G(z) where P(z) is a predictor and G(z) is an interpolating estimator. The general-purpose design procedure for efficient IFIR predictors is described, and demonstrated for polynomial predictors. The resulting predictors, optimized for white noise attenuation, have much lower computational complexity than the corresponding direct-form FIR predictors. Secondly, an IIR filter-based implementation of sinusoidal FIR predictors is presented. As an application, a zero-crossing detector for 50 Hz thyristor drives is designed     

Theory of constraints-based methodology for effective ERP implementations

The benefits as well as the turmoil that the implementation of enterprise resource planning (ERP) systems creates for multinational companies are well known. Several reports and evidence through case studies underline both the difficulties and the resulting benefits of ERPs. This paper addresses the reasoning behind long implementation times and organizational thunderstorms that tantalise the deployment of ERP systems. It focuses on two aspects of most implementation projects that generate the majority of technical and functional problems and constitute the projects’ bottleneck, i.e. code development within ERP systems due to key and unique requirements in a business environment, and localization and reporting needs that companies must adhere to or want to achieve. The approach proposes the classification of functional requirements into business critical and legally necessary, and the distribution of code development for system not fully supported processes among these two classes is discussed. Subsequently, the implementation times and deployment inefficiencies are coupled with the level of code development, and the difficulty of avoiding this for the two requirements’ classes, mainly due to user inflexibility and local environment peculiarities, is discussed. Using the Theory of Constraints, a coherent methodology for handling bottlenecks and effectively planning the code development effort is proposed, and trade-offs are derived between successful and on-time ERP implementations with managerial enforcement of best practices fully functional within major ERP systems. The approach is verified through field data from an actual SAP R/3 implementation at the largest manufacturer of packaging products and equipment in Europe.     

VLSI implementations of efficient isotropic flexible 2D convolvers

A new 2D isotropic convolver designed to operate on 256 grey-level images is presented. When realised by using 90 nm IV CMOS technology, the core of the proposed circuit exhibits a 1.25-GHz running frequency with an average power dissipation of only sime1 mW/MHz. The new convolver can also be efficiently realised using FPGAs.     

Comparing implementations of global and local indicators of spatial association

Functions to calculate measures of spatial association, especially measures of spatial autocorrelation, have been made available in many software applications. Measures may be global, applying to the whole data set under consideration, or local, applying to each observation in the data set. Methods of statistical inference may also be provided, but these will, like the measures themselves, depend on the support of the observations, chosen assumptions, and the way in which spatial association is represented; spatial weights are often used as a representational technique. In addition, assumptions may be made about the underlying mean model, and about error distributions. Different software implementations may choose to expose these choices to the analyst, but the sets of choices available may vary between these implementations, as may default settings. This comparison will consider the implementations of global Moran’s I, Getis–Ord G and Geary’s C, local \(I_i\) and \(G_i\), available in a range of software including Crimestat, GeoDa, ArcGIS, PySAL and R contributed packages.     

On the effect of linear algebra implementations in real-time multibody system dynamics

This paper compares the efficiency of multibody system (MBS) dynamic simulation codes that rely on different implementations of linear algebra operations. The dynamics of an N-loop four-bar mechanism has been solved with an index-3 augmented Lagrangian formulation combined with the trapezoidal rule as numerical integrator. Different implementations for this method, both dense and sparse, have been developed, using a number of linear algebra software libraries (including sparse linear equation solvers) and optimized sparse matrix computation strategies. Numerical experiments have been performed in order to measure their performance, as a function of problem size and matrix filling. Results show that optimal implementations can increase the simulation efficiency in a factor of 2–3, compared with our starting classical implementations, and in some topics they disagree with widespread beliefs in MBS dynamics. Finally, advices are provided to select the implementation which delivers the best performance for a certain MBS dynamic simulation.     

A comparison of approaches and implementations for automating decision analysis

This paper describes and compares three formal methods and software implementations for solving decision analysis problems. The comparison includes decision trees, influence diagrams, and a new approach developed by the authors that combines features of trees and influence diagrams. The new approach introduces a language for expressing decision analysis problems and an efficient solution algorithm that exploits structural properties of decision analysis problems. These features allows decision analysis problems to be represented compactly and to be solved very efficiently.     

Mapping the barriers of AI implementations in the public distribution system: The Indian experience

With increasing digitization, private organizations have started adopting modern technology to improve the effectiveness and transparency of their system. In countries like India, where most public services are under government control, technology adoption is nascent due to various obstacles. The study considered artificial intelligence (AI) the most popular technology and identified 18 critical adoption barriers in India's public distribution systems (PDS). The study is further extended to find the contextual relationship among barriers using interpretative structural modeling (ISM) and prioritize them using the analytical network process (ANP) method. The study identifies lack of trust in technology, lack of AI literacy, and political issues as significant barriers to AI adoption in PDS. The hybrid methodology used in this study proposed five different strategies for effective and smooth implementation of AI in PDS, which would help the policymaker plan the same.     

Free-space optical shuffle implementations by use of birefringence-customized modular optics

A stacked integration technique based on polarization optics is studied for implementing shuffle-based interconnection networks with three-dimensional solid-state modules. A basic building block of the proposed scheme consists of a hole-patterned half-wave retarder for birefringence customizations and a calcite slab for subsequent beam deflections. On the basis of a cascade of such building blocks the submodules of various shuffle-family permutations can be implemented. To minimize channel cross talk, we incorporated a collimating-relaying imaging system. To help design birefringence customization, we developed algebraic formulations of folded shuffle operations using separable shuffles. Proof-of-concept experimental results, as well as system design, fabrication, and integration issues, are discussed.     

Alternative interpretation of the edge-diffraction phenomenon

An alternative interpretation of the phenomenon of edge diffraction is proposed according to a new separation of the Fresnel function. The subfields are investigated in the problem of diffraction of a plane wave by a perfectly conducting half-plane, and the results are compared numerically with other interpretations.     

Efficient and side-channel-aware implementations of elliptic curve cryptosystems over prime fields

Elliptic curve cryptosystems (ECCs) are utilised as an alternative to traditional public-key cryptosystems, and are more suitable for resource-limited environments because of smaller parameter size. In this study, the authors carry out a thorough investigation of side-channel attack aware ECC implementations over finite fields of prime characteristic including the recently introduced Edwards formulation of elliptic curves. The Edwards formulation of elliptic curves is promising in performance with built-in resiliency against simple side-channel attacks. To our knowledge the authors present the first hardware implementation for the Edwards formulation of elliptic curves. The authors also propose a technique to apply non-adjacent form (NAF) scalar multiplication algorithm with side-channel security using the Edwards formulation. In addition, the authors implement Joye's highly regular add-always scalar multiplication algorithm both with the Weierstrass and Edwards formulation of elliptic curves. Our results show that the Edwards formulation allows increased area-time performance with projective coordinates. However, the Weierstrass formulation with affine coordinates results in the simplest architecture, and therefore has the best area-time performance as long as an efficient modular divider is available.