Algorithm 1019: A Task-based Multi-shift QR/QZ Algorithm with Aggressive Early Deflation

The QR algorithm is one of the three phases in the process of computing the eigenvalues and the eigenvectors of a dense nonsymmetric matrix. This paper describes a task-based QR algorithm for reducing an upper Hessenberg matrix to real Schur form. The task-based algorithm also supports generalized eigenvalue problems (QZ algorithm) but this paper concentrates on the standard case. The task-based algorithm adopts previous algorithmic improvements, such as tightly-coupled multi-shifts and Aggressive Early Deflation (AED) , and also incorporates several new ideas that significantly improve the performance. This includes, but is not limited to, the elimination of several synchronization points, the dynamic merging of previously separate computational steps, the shortening and the prioritization of the critical path, and experimental GPU support. The task-based implementation is demonstrated to be multiple times faster than multi-threaded LAPACK and ScaLAPACK in both single-node and multi-node configurations on two different machines based on Intel and AMD CPUs. The implementation is built on top of the StarPU runtime system and is part of the open-source StarNEig library.

Cooperative Spectrum Sensing Based on Multi-Features Combination Network in Cognitive Radio Network

cognitive radio, as a key technology to improve the utilization of radio spectrum, acquired much attention. Moreover, spectrum sensing has an irreplaceable position in the field of cognitive radio and was widely studied. The convolutional neural networks (CNNs) and the gate recurrent unit (GRU) are complementary in their modelling capabilities. In this paper, we introduce a CNN-GRU network to obtain the local information for single-node spectrum sensing, in which CNN is used to extract spatial feature and GRU is used to extract the temporal feature. Then, the combination network receives the features extracted by the CNN-GRU network to achieve multifeatures combination and obtains the final cooperation result. The cooperative spectrum sensing scheme based on Multifeatures Combination Network enhances the sensing reliability by fusing the local information from different sensing nodes. To accommodate the detection of multiple types of signals, we generated 8 kinds of modulation types to train the model. Theoretical analysis and simulation results show that the cooperative spectrum sensing algorithm proposed in this paper improved detection performance with no prior knowledge about the information of primary user or channel state. Our proposed method achieved competitive performance under the condition of large dynamic signal-to-noise ratio.

Front propagation and arrival times in networks with application to neurodegenerative diseases

Many physical, epidemiological, or physiological dynamical processes on networks support front-like propagation, where an initial localized perturbation grows and systematically invades all nodes in the network. A key question is then to extract estimates for the dynamics. In particular, if a single node is seeded at a small concentration, when will other nodes reach the same initial concentration? Here, motivated by the study of toxic protein propagation in neurodegenerative diseases, we present and compare three different estimates for the arrival time in order of increasing analytical complexity: the linear arrival time, obtained by linearizing the underlying system; the Lambert time, obtained by considering the interaction of two nodes; and the nonlinear arrival time, obtained by asymptotic techniques. We use the classic Fisher-Kolmogorov-Petrovsky-Piskunov equation as a paradigm for the dynamics and show that each method provides different insight and time estimates. Further, we show that the nonlinear asymptotic method also gives an approximate solution valid in the entire domain and the correct ordering of arrival regions over large regions of parameters and initial conditions.

Analysis of Supply Chain Super-Network Coordination Considering Risk Function in Emergencies

Economic globalization continues to expand the scope of the supply chain network structure, while increasing its own complexity, as well as the uncertainty of the network operating environment and the fragility of the operating system. An emergency on a single node or line in the supply chain network usually affects other nodes in the supply chain and brings significant risks to the enterprise. The impact of other nodes can cause the entire supply chain network to collapse, especially if the production and operation of a single-node enterprise in the supply chain may be interrupted or malfunctioned, especially in the event of an emergency. It also threatens development greatly, affecting the production and livelihoods of enterprises in the supply chain and people's lives, and has a major negative impact on social and economic development. These emergencies continue to affect the supply chain network, and the originally fragile companies face greater risks. This paper establishes a supply chain hyper-network model considering the risk function under emergencies. When an emergency occurs, the demand in the consumer market decreases or increases due to different emergencies. Therefore, revenue sharing contracts are used to coordinate, build a supply chain network model under emergencies, and solve them to obtain a model equilibrium Solution, that is, the new equilibrium state after the occurrence of an emergency.

Monitoring the water balance of seepage lakes to track regional responses to an evolving climate

Understanding the causes of large fluctuations in lake water levels is important for adaptive resource management. The relatively simple water budgets of small seepage lakes make them potentially useful model systems, provided that key water balance components can be well constrained. Here, spatial variability in measured rates of evaporation (E) and precipitation (P) at the whole lake scale was investigated, and the effect on daily and seasonal water balance estimates was quantified. To estimate spatial variability, triplicate sensor platforms were deployed on and near an 18 ha seepage lake. Lake stage (S) was monitored at a single node in the lake. The water balance was closed by estimating net groundwater seepage (Gnet) analytically as Gnet = ∆S – (P – E). Instrumentation on a second seepage lake was maintained by citizen scientists to assess the potential for more widespread sensor deployments. Data were collected every 30-minutes for six months. The results indicate that low-cost sensor networks with single nodes to measure E, P and ∆S provide well-constrained water budgets at daily and seasonal time scales.

Application of Machine Learning Techniques to Estimate Unsoaked California Bearing Ratio in Ekiti Central Senatorial District

This paper investigates the relationship between soil physical properties and the Un-soaked California Bearing Ratio (USCBR) of soil found in Ekiti State Central Senatorial District (ESCSD), which includes Natural Moisture Content (NMC%) Percentage Fines, Specific Gravity (SG) and Consistency Limits (LL%, PL%, & PI %). The database was prepared in the laboratory by conducting tests on ninety-nine (99) soil samples which were obtained in a burrowed pit found in the Central Senatorial District of Ekiti State. An R version 4.0.5 and R studio version 1.2.5033 was used to analyze the Artificial Neural Networks (ANNs) and Least Square Regression (LSR) in order to develop a simplified CBR model. In both models, independent layer containing six nodes (soil physical properties) and the dependent layer containing a single node (i.e. CBR) were taken. The descriptive analysis for training and testing was performed; boxplots of the variables were plotted and; sensitivity analysis was carried out. The capacity of the developed equation was evaluated in terms of error metrics MSE and RMSE. The analysis showed that both ANN and MLR models predicted CBR close to the laboratory value. However, the model without the percentage passing sieve 200 (MIC) is the best, having Akaike Information Criterion and Bayesian Information Criterion values of 614.1707 and 627.5754 respectively, from the error metrics analysis, the results showed that PL and LL are the most influential variable that affects the developed CBR model's output. From the foregoing its concluded that the study has shown a relationship between the CBR value of Ekiti Central Senatorial District soil and its basic soils properties using machine learning techniques, also the developed CBR model will be useful tool to Civil engineers, geotechnical engineers and construction industry within the study area particularly in their preliminary stage of their project.

ANALYSING ASSOCIATION BETWEEN ANTHROPOMETRIC INDICATORS OF INSULIN RESISTANCE AND INSULIN-LIKE GROWTH FACTOR-1 IN PATIENTS WITH NODULAR GOITER

Publications suggesting that thyroid nodule might be associated with insulin resistance (IR) and metabolic syndrome are quite interesting. In a very recent report, increased thyroid volume and nodule prevalence were also reported in patients with IR in an iodine-sufficient area []. The purpose of the work is to analyze the association between anthropometric indicators IR and IGF-1 in patients with nodular goiter. Materials and methods. During the study the authors examined 73 patients with euthyroid single-node (n = 34) and multinodular goiter (n = 39) aged 17 to 74 years (mean - (51.0 ± 10.6) years), determining WC, WC / HC, BMI, WHtR, ABSI, BFD, BRI, CI, AVI, BAI, IGF-1, TSH, fT4, fT3. Thyroid volume, its structure, number, size and location of foci was assessed by an ultrasonic complex Aloka SSD-1100 (Japan), using a linear sensor 7.5 MHz. Results and their discussion. In the total number of patients with nodular goiter IGF-1 is nonlinearly negatively associated with BMI (r = -0.30; P = 0.016), WC (r = -0.26; P = 0.036), WHtR (r = -0.30) ; P = 0.020), AVI (r = -0.27; P = 0.03), ABSI (r = -0.31; P = 0.015), nonlinear positive with BFD (r = 0.27; P = 0.033) ), BRI (r = 0.29; P = 0.02) and linearly positive with BAI (r = 0.36; P = 0.004); thyroid volume is linearly positively associated with age (r = 0.35; P = 0.009), nonlinearly positively with WC / HC (r = 0.43; P = 0.001), BFD (r = 0.26; P = 0.06 ) and CI (r = 0.31; P = 0.02). In patients with nodular goiter with BMI≥35 kg / m2 thyroid volume is linearly positively associated with BMI (r = 0.71; P = 0.049). In patients with nodular goiter with IRF-1 above the sex-age norm, thyroid volume is nonlinearly positively associated with WC / HC (r = 0.71; P = 0.01), BAI (r = 0.66; P = 0.03 ) and nonlinearly negative with BFD (r = -0.52; P = 0.01). It has been found that BAI explains 82.37% of the variance of IGF-1 in the general group and more than 90% of the variance of its level in groups of patients with nodular goiter with high IGF-1 with / without obesity. In patients with nodular goiter with high IGF-1 and obesity, the predictor of increased thyroid volume is BRI, which explains 81.14% of the variance of its volume. Conclusions: Patients with nodular goiter with IGF-1 level in blood above the sex-age norm have significantly higher values ​​of anthropometric indicators IR (WHtR, ABSI, BFD and BAI) compared with patients with a normal level of this indicator; in patients with nodular goiter with II degree obesity and above, thyroid volume is significantly associated with BMI; BAI (R2 = 82.37%) is a predictor of increased levels of IGF-1 in blood of patients with nodular goiter, regardless of the obesity; BRI (R2 = 81.14%) is a predictor of increased thyroid volume in patients with nodular goiter with IGF -1 high level and obesity. Key words: nodular goiter, anthropometric indicators, insulin resistance

Enhanced GAF protocol based on graph theory to optimize energy efficiency and lifetime in WSN technology

PurposeThe purpose of this paper is to design an enhanced routing protocol to minimize energy consumed and extend network lifetime in sensor network (WSN).Design/methodology/approachWith the use of appropriate routing protocols, data collected by sensor nodes reache the BS. The entire network lifetime can be extended well beyond that of its single nodes by putting the nodes in sleep state when they are not in use, and make active just a single node at a time within a given area of interest. So that, the lowest-cost routing arises by minimizing the communication cost. This paper proposes an enhanced adaptive geographic fidelity (E-GAF) routing protocol based on theory of graphs approach to improve the discovery phase, select the optimal path, reduce the energy used by nodes and therefore extend the network lifetime. Following the simulations established by varying the number of grids and tests, a comparison is made between the E-GAF and basic GAF (B-GAF) based on the number of dead nodes and energy consumption.FindingsThe results obtained show that E-GAF is better than the existing basic GAF protocol in terms of energy efficiency and network lifetime.Originality/valueThis paper adopts the latest optimization algorithm know as E-GAF, which is used to solve the problem of energy and improve the network lifetime in a WSN. This is the first work that utilizes network lifetime in WSN.

An analytic theory of shallow networks dynamics for hinge loss classification*

Neural networks have been shown to perform incredibly well in classification tasks over structured high-dimensional datasets. However, the learning dynamics of such networks is still poorly understood. In this paper we study in detail the training dynamics of a simple type of neural network: a single hidden layer trained to perform a classification task. We show that in a suitable mean-field limit this case maps to a single-node learning problem with a time-dependent dataset determined self-consistently from the average nodes population. We specialize our theory to the prototypical case of a linearly separable data and a linear hinge loss, for which the dynamics can be explicitly solved in the infinite dataset limit. This allows us to address in a simple setting several phenomena appearing in modern networks such as slowing down of training dynamics, crossover between rich and lazy learning, and overfitting. Finally, we assess the limitations of mean-field theory by studying the case of large but finite number of nodes and of training samples.