Network Simulations Research Articles

Offline memory replay in recurrent neuronal networks emerges from constraints on online dynamics.

During spatial exploration, neural circuits in the hippocampus store memories of sequences of sensory events encountered in the environment. When sensory information is absent during 'offline' resting periods, brief neuronal population bursts can 'replay' sequences of activity that resemble bouts of sensory experience. These sequences can occur in either forward or reverse order, and can even include spatial trajectories that have not been experienced, but are consistent with the topology of the environment. The neural circuit mechanisms underlying this variable and flexible sequence generation are unknown. Here we demonstrate in a recurrent spiking network model of hippocampal area CA3 that experimental constraints on network dynamics such as population sparsity, stimulus selectivity, rhythmicity and spike rate adaptation, as well as associative synaptic connectivity, enable additional emergent properties, including variable offline memory replay. In an online stimulus-driven state, we observed the emergence of neuronal sequences that swept from representations of past to future stimuli on the timescale of the theta rhythm. In an offline state driven only by noise, the network generated both forward and reverse neuronal sequences, and recapitulated the experimental observation that offline memory replay events tend to include salient locations like the site of a reward. These results demonstrate that biological constraints on the dynamics of recurrent neural circuits are sufficient to enable memories of sensory events stored in the strengths of synaptic connections to be flexibly read out during rest and sleep, which is thought to be important for memory consolidation and planning of future behaviour. KEY POINTS: A recurrent spiking network model of hippocampal area CA3 was optimized to recapitulate experimentally observed network dynamics during simulated spatial exploration. During simulated offline rest, the network exhibited the emergent property of generating flexible forward, reverse and mixed direction memory replay events. Network perturbations and analysis of model diversity and degeneracy identified associative synaptic connectivity and key features of network dynamics as important for offline sequence generation. Network simulations demonstrate that population over-representation of salient positions like the site of reward results in biased memory replay.

Training a Neural Network with Exciton-Polariton Optical Nonlinearity

In contrast to software simulations of neural networks, hardware implementations have limited or no tunability. While such networks promise great improvements in terms of speed and energy efficiency, their performance is limited by the difficulty of applying efficient training. We propose and realize experimentally an optical system where highly efficient backpropagation training can be applied through an array of highly nonlinear, nontunable nodes. The system includes exciton-polariton nodes realizing nonlinear activation functions. We demonstrate high classification accuracy in the MNIST handwritten digit benchmark in a single-hidden-layer system.

Multiplicative Shot-Noise: A New Route to Stability of Plastic Networks.

Fluctuations of synaptic weights, among many other physical, biological, and ecological quantities, are driven by coincident events of two "parent" processes. We propose a multiplicative shot-noise model that can capture the behaviors of a broad range of such natural phenomena, and analytically derive an approximation that accurately predicts its statistics. We apply our results to study the effects of a multiplicative synaptic plasticity rule that was recently extracted from measurements in physiological conditions. Using mean-field theory analysis and network simulations, we investigate how this rule shapes the connectivity and dynamics of recurrent spiking neural networks. The multiplicative plasticity rule is shown to support efficient learning of input stimuli, and it gives a stable, unimodal synaptic-weight distribution with a large fraction of strong synapses. The strong synapses remain stable over long times but do not "run away." Our results suggest that the multiplicative shot-noise offers a new route to understand the tradeoff between flexibility and stability in neural circuits and other dynamic networks.

Design of V2X-Based Vehicular Contents Centric Networks for Autonomous Driving

Recent technical innovation has driven the evolution of autonomous vehicles. To improve safety as well as on-road vehicular experience, vehicles should be connected with each other or to vehicular networks. Some specification groups, e.g., IEEE and 3GPP, have studied and released vehicular communication requirements and architecture. IEEE’s Wireless Access in Vehicular Environment focuses on dedicated and short-range communication, while 3GPP’s New Radio V2X supports not only sidelink but also uplink communication. The 3GPP Release 16, which supports 5G New Radio, offers evolved functionalities such as network slice, Network Function Virtualization, and Software-Defined Networking. In this paper, we define and design a vehicular network architecture compliant with 5G core networks to enable and support autonomous driving. As a validation example, a high-definition map needs to contain the context of trajectory for localization and planning of autonomous driving vehicles. We also propose new methods by which autonomous vehicles can push and pull map content efficiently, without causing bottlenecks on the network core. We evaluate the performance of the proposed method via network simulations and our autonomous driving vehicle on the road. Experimental results indicate that the proposed method improves the performance of vehicular content delivery in real-world road environments.

Modeling and optimization of natural gas distribution networks for new supplier projects

• NG distribution networks involving suppliers with unspecified streams. • Cost minimization and specification assurance for the consumer market. • Real case study involving all the alternatives for receiving NG for distribution. • Feasibility and flexibility of applying the model in different configurations. • Reduction in the total cost of distribution of 22.65%. The opening of the market and the possibility of the participation of several suppliers in Natural Gas (NG) distribution networks has raised the level of complexity associated with the quality control of the final stream to be made available to the consumer market. There are suppliers that offer lower prices but supply a lower quality of fuel which can contribute to an out-of-spec final stream (after mixing the different suppliers). On the other hand, it is not always economically viable to build Processing Units and acquire analytical equipment for gas quality control. In general, simulations in natural gas distribution networks focus on the evaluation of physical criteria and energy balance. This work presents a dynamic optimization model, validated by real case studies, for the monitoring and quality control of a natural gas stream mixing process that involves several suppliers providing varying quality levels (including out of specification streams) and prices. The results show the feasibility of executing projects for new suppliers, as well as supplying natural gas to customers who use it as a raw material and who establish more restrictive specification limits than those established by the regulatory agency. Comparing with the conventional operating strategy, the results obtained by the proposed optimization model show a reduction in the cost of distribution equal to 13.5% and 22.6% in the two case studies analyzed, respectively.

PERENCANAAN SISTEM JARINGAN DISTRIBUSI SISTEM PENYEDIAAN AIR MINUM PEDESAAN (SPAMDES) DI DESA PEDAWA KECAMATAN BANJAR BULELENG

Water is a very important natural resource for mankind. Pedawa Village of Banjar Subdistrict located in the southern part of Buleleng Regency with an area of 16.68 km2 which has limited clean water services. Where the level of SPAMDES service which is only 29% or 1,672 people out of 5,733 people causes the fulfillment of clean water in Pedawa Village is still low, there are 4 springs used in SPAM Pedawa Village but not all communities get water, so it is necessary to plan an optimal rural drinking water supply network system (SPAMDES). The planning of this distribution network system is assisted by WaterNet where the software uses energy potential to see the extent to which water from the spring can flow through the distribution network.. The data required in this planning are elevation, discharge and coordinate point to calculate energy loss, pipe length and pump head and the number of residents of the last 6 years for projected population growth for 20 years. The results of projected water needs in the service area were obtained for Lambo Reservoir 54 m3, Lambo2 Reservoir 99 m3, Bangkiang Sidem1 Reservoir 45 m3, Bangkiang Sidem2 Reservoir 45 m3, Reservoir Gelunggang 75 m3, Asah Reservoir 15 m3, Insakan Reservoir 9 m3 and Munduk Waban Reservoir 1.5 m3. Based on the results of network simulations with galvanized pipes and flow systems with pumps and gravity obtained results for Constant and Extended flow types on WaterNet hydraulically the network can provide services as desired on each node. Where the pipe for the transmission network is obtained a pipe diameter of 10 and 5 cm, and for the distribution network obtained a pipe diameter of 15, 10 and 5 cm

On Stretching, Bending, Shearing, and Twisting of Actin Filaments I: Variational Models.

Mechanochemical simulations of actomyosin networks are traditionally based on one-dimensional models of actin filaments having zero width. Here, and in the follow up paper (arXiv, DOI 10.48550/arXiv.2203.01284), approaches are presented for more efficient modeling that incorporates stretching, shearing, and twisting of actin filaments. Our modeling of a semiflexible filament with a small but finite width is based on the Cosserat theory of elastic rods, which allows for six degrees of freedom at every point on the filament's backbone. In the variational models presented in this paper, a small and discrete set of parameters is used to describe a smooth filament shape having all degrees of freedom allowed in the Cosserat theory. Two main approaches are introduced: one where polynomial spline functions describe the filament's configuration, and one in which geodesic curves in the space of the configurational degrees of freedom are used. We find that in the latter representation the strain energy function can be calculated without resorting to a small-angle expansion, so it can describe arbitrarily large filament deformations without systematic error. These approaches are validated by a dynamical model of a Cosserat filament, which can be further extended by using multiresolution methods to allow more detailed monomer-based resolution in certain parts of the actin filament, as introduced in the follow up paper. The presented framework is illustrated by showing how torsional compliance in a finite-width filament can induce broken chiral symmetry in the structure of a cross-linked bundle.

Ultrastrong and multifunctional aerogels with hyperconnective network of composite polymeric nanofibers

Three-dimensional (3D) microfibrillar network represents an important structural design for various natural tissues and synthetic aerogels. Despite extensive efforts, achieving high mechanical properties for synthetic 3D microfibrillar networks remains challenging. Here, we report ultrastrong polymeric aerogels involving self-assembled 3D networks of aramid nanofiber composites. The interactions between the nanoscale constituents lead to assembled networks with high nodal connectivity and strong crosslinking between fibrils. As revealed by theoretical simulations of 3D networks, these features at fibrillar joints may lead to an enhancement of macroscopic mechanical properties by orders of magnitude even with a constant level of solid content. Indeed, the polymeric aerogels achieved both high specific tensile modulus of ~625.3 MPa cm3 g−1 and fracture energy of ~4700 J m−2, which are advantageous for diverse structural applications. Furthermore, their simple processing techniques allow fabrication into various functional devices, such as wearable electronics, thermal stealth, and filtration membranes. The mechanistic insights and manufacturability provided by these robust microfibrillar aerogels may create further opportunities for materials design and technological innovation.

Input Parameters Comparison on NARX Neural Network to Increase the Accuracy of Stock Prediction

The trading of stocks is one of the activities carried out all over the world. To make the most profit, analysis is required, so the trader could determine whether to buy or sell stocks at the right moment and at the right price. Traditionally, technical analysis which is mathematically processed based on historical price data can be used. Parallel to technological development, the analysis of stock price and its forecasting can also be accomplished by using computer algorithms e.g. machine learning. In this study, Nonlinear Auto Regressive network with eXogenous inputs (NARX) neural network simulations were performed to predict the stock index prices. Experiments were implemented using various configurations of input parameters consisting of Open, High, Low, Closed prices in conjunction with several technical indicators for maximum accuracy. The simulations were carried out by using stock index data sets namely JKSE (Indonesia Jakarta index) and N225 (Japan Nikkei index). This work showed that the best input configurations can predict the future 13 days Close prices with 0.016 and 0.064 mean absolute error (MAE) for JKSE and N225 respectively.Â

Nonlinear lumped-parameter models for blood flow simulations in networks of vessels

To address the issue of computational efficiency related to the modelling of blood flow in complex networks, we derive a family of nonlinear lumped-parameter models for blood flow in compliant vessels departing from a well-established one-dimensional model. These 0D models must preserve important nonlinear properties of the original 1D model: the nonlinearity of the pressure-area relation and the pressure-dependent parameters characterizing the 0D models, the resistance R and the inductance L, defined in terms of a time-dependent cross-sectional area subject to pressure changes. We introduce suitable coupling conditions to join 0D vessels through 0D junctions and construct 0D networks preserving the original 1D network topology. The newly derived nonlinear 0D models are then applied to several arterial networks and the predicted results are compared against (i) the reference 1D results, to validate the models and assess their ability to reproduce good approximations of pressure and flow waveforms in all vessels at a much lower computational cost, measured in terms of CPU time, and (ii) the linear 0D results, to evaluate the improvement gained by including certain nonlinearities in the 0D models, in terms of agreement with the 1D results.

Effect of architecture disorder on the elastic response of two-dimensional lattice materials.

We examine how disordering joint position influences the linear elastic behavior of lattice materials via numerical simulations in two-dimensional beam networks. Three distinct initial crystalline geometries are selected as representative of mechanically isotropic materials with low connectivity, mechanically isotropic materials with high connectivity, and mechanically anisotropic materials with intermediate connectivity. Introducing disorder generates spatial fluctuations in the elasticity tensor at the local (joint) scale. Proper coarse-graining reveals a well-defined continuum-level scale elasticity tensor. Increasing disorder aids in making initially anisotropic materials more isotropic. The disorder impact on the material stiffness depends on the lattice connectivity: Increasing the disorder softens lattices with high connectivity and stiffens those with low connectivity, without modifying the scaling between elastic modulus and density (linear scaling for high connectivity and cubic scaling for low connectivity). Introducing disorder in lattices with intermediate fixed connectivity reveals both scaling: the linear scaling occurs for low density, the cubic one at high density, and the crossover density increases with disorder. Contrary to classical formulations, this work demonstrates that connectivity is not the sole parameter governing elastic modulus scaling. It offers a promising route to access novel mechanical properties in lattice materials via disordering the architectures.

Effect of pore structure and injection pressure on waterflooding in tight oil sandstone cores using NMR technique and pore network simulation

As an economical and reusable method of enhancing oil recovery, waterflooding is widely used in the development of tight oil reservoirs, although it has often been found unsuccessful owing to the adverse properties of tight rocks. Accurate knowledge of the pore-scale distribution of oil in tight rocks during waterflooding is needed for devising strategies to maximize oil recovery. Here, we used a two-fold approach to identify the amount and spatial distribution of the residual oil as a function of injection pressure. We i) characterized the pore structure of two cores from a tight sandstone reservoir in the Ordos basin, China, and ii) conducted waterflooding experiments with different injection pressures using nuclear magnetic resonance measurements. We found that even small differences in the tight sandstones pore structures caused measurable variations of the amounts oil produced from different classes of pores. For example, in experiments using identical water injection pressures but samples with different permeabilities, a permeability increase by a factor of 3 corresponded to a 10% reduction in oil recovery. We also found that the increase in residual oil saturation caused by increasing the injection pressure was unfavorably dependent on the pore structure heterogeneity. This observation was supported by pore network simulations, which were additionally carried out. We therefore conclude that it is necessary to evaluate the pore structure heterogeneity of the reservoir rocks when attempting to enhance the oil recovery factor of a tight reservoir by means of the common strategy of raising the waterflood injection pressure.

Associative memories using complex-valued Hopfield networks based on spin-torque oscillator arrays

Simulations of complex-valued Hopfield networks based on spin-torque oscillators can recover phase-encoded images. Sequences of memristor-augmented inverters provide tunable delay elements that implement complex weights by phase shifting the oscillatory output of the oscillators. Pseudo-inverse training suffices to store at least 12 images in a set of 192 oscillators, representing 16 × 12 pixel images. The energy required to recover an image depends on the desired error level. For the oscillators and circuitry considered here, 5% root mean square deviations from the ideal image require approximately 5 μs and consume roughly 130 nJ. Simulations show that the network functions well when the resonant frequency of the oscillators can be tuned to have a fractional spread less than 10−3, depending on the strength of the feedback.

Rapid MR Scanner Independent B1 Field Measurement System for Phased Arrays.

This paper demonstrates a rapid B1 field benchtop measurement system that is independent of an MR scanner and network analyzer. This system can be used to obtain radiofrequency (B1 field) strength distribution plots of multiple 2D slices (with an extension to 3D) of a liquid cylindrical phantom for multi-element phased arrays used in MRI. The system can be used in three modes- element, phased array, and multiple fixed point pattern measurement. These modes are demonstrated for a 7T 1H eight-channel dipole array and a corn-syrup based phantom. The system can measure complex phase and amplitude measurements from up to 8 elements in the first mode one or 8 different phase settings in the second mode at a rate of approximately 37 positions per minute, allowing a full 2D B1 mapping for 1303 points in 33.05 minutes. The scan patterns obtained using this setup are compared to the ones obtained using an HP network analyzer and simulations. This work can be extended to measure the E field, SAR and upon increasing the speed of measurement, could be used for applications such as Transmit SENSE. Clinical Relevance- This work benefits a faster and more widely accessible measurement system for phased array antennas for MRI. As phased arrays are becoming very important in MRI, the ability to assess individual element performance more rapidly and B1 shimming performance is important to aid in their further development.

Quantifying the influence of stimulation protocols on neural network connectivity inference to optimize rapid network measurements.

Connectivity is key to understanding neural circuit computations. However, estimating in vivo connectivity using recording of activity alone is challenging. Issues include common input and bias errors in inference, and limited temporal resolution due to large data requirements. Perturbations (e.g. stimulation) can improve inference accuracy and accelerate estimation. However, optimal stimulation protocols for rapid network estimation are not yet established. Here, we use neural network simulations to identify stimulation protocols that minimize connectivity inference errors when using generalized linear model inference. We find that stimulation parameters that balance excitatory and inhibitory activity minimize inference error. We also show that pairing optimized stimulation with adaptive protocols that choose neurons to stimulate via Bayesian inference may ultimately enable rapid network inference.

On Snapshot-Based Model Reduction Under Compatibility Conditions for a Nonlinear Flow Problem on Networks

This paper is on the construction of structure-preserving, online-efficient reduced models for the barotropic Euler equations with a friction term on networks. The nonlinear flow problem finds broad application in the context of gas distribution networks. We propose a snapshot-based reduction approach that consists of a mixed variational Galerkin approximation combined with quadrature-type complexity reduction. Its main feature is that certain compatibility conditions are assured during the training phase, which make our approach structure-preserving. The resulting reduced models are locally mass conservative and inherit an energy bound and port-Hamiltonian structure. We also derive a wellposedness result for them. In the training phase, the compatibility conditions pose challenges, we face constrained data approximation problems as opposed to the unconstrained training problems in the conventional reduction methods. The training of our model order reduction consists of a principal component analysis under a compatibility constraint and, notably, yields reduced models that fulfill an optimality condition for the snapshot data. The training of our quadrature-type complexity reduction involves a semi-definite program with combinatorial aspects, which we approach by a greedy procedure. Efficient algorithmic implementations are presented. The robustness and good performance of our structure-preserving reduced models are showcased at the example of gas network simulations.

Channel capacity optimization scheme for safety applications in vehicular ad hoc networks

SummaryVehicular ad hoc networks (VANETs) have attracted significant attention due to their safety, traffic management, and infotainment applications. However, vehicular communication suffers from network congestion caused by high vehicle density, which raises numerous challenges for efficiently disseminating emergency messages in vehicular safety applications. This paper aims to solve the problem of capacity optimization of the control channel (CCH) during the delivery of emergency messages. Hence, we propose a novel channel capacity optimization scheme for safety application (CCOSSA). The performance of the CCOSSA scheme is evaluated by performing comprehensive traffic and network simulations regarding various parameters such as the probability of reception, channel busy time (CBT) ratio, and average delay. CCOSSA scheme is compared with the state‐of‐the‐art schemes under varying vehicle density and distance from the source in which the proposed scheme shows an improvement in the performance.

Fast Simulation of a Multi-Area Spiking Network Model of Macaque Cortex on an MPI-GPU Cluster.

Spiking neural network models are increasingly establishing themselves as an effective tool for simulating the dynamics of neuronal populations and for understanding the relationship between these dynamics and brain function. Furthermore, the continuous development of parallel computing technologies and the growing availability of computational resources are leading to an era of large-scale simulations capable of describing regions of the brain of ever larger dimensions at increasing detail. Recently, the possibility to use MPI-based parallel codes on GPU-equipped clusters to run such complex simulations has emerged, opening up novel paths to further speed-ups. NEST GPU is a GPU library written in CUDA-C/C++ for large-scale simulations of spiking neural networks, which was recently extended with a novel algorithm for remote spike communication through MPI on a GPU cluster. In this work we evaluate its performance on the simulation of a multi-area model of macaque vision-related cortex, made up of about 4 million neurons and 24 billion synapses and representing 32 mm2 surface area of the macaque cortex. The outcome of the simulations is compared against that obtained using the well-known CPU-based spiking neural network simulator NEST on a high-performance computing cluster. The results show not only an optimal match with the NEST statistical measures of the neural activity in terms of three informative distributions, but also remarkable achievements in terms of simulation time per second of biological activity. Indeed, NEST GPU was able to simulate a second of biological time of the full-scale macaque cortex model in its metastable state 3.1× faster than NEST using 32 compute nodes equipped with an NVIDIA V100 GPU each. Using the same configuration, the ground state of the full-scale macaque cortex model was simulated 2.4× faster than NEST.

Balancing Quarantine and Self-Distancing Measures in Adaptive Epidemic Networks

We study the relative importance of two key control measures for epidemic spreading: endogenous social self-distancing and exogenous imposed quarantine. We use the framework of adaptive networks, moment-closure, and ordinary differential equations to introduce new model types of susceptible-infected-recovered (SIR) dynamics. First, we compare computationally expensive, adaptive network simulations with their corresponding computationally efficient ODE equivalents and find excellent agreement. Second, we discover that there exists a critical curve in parameter space for the epidemic threshold, which suggests a mutual compensation effect between the two mitigation strategies: as long as social distancing and quarantine measures are both sufficiently strong, large outbreaks are prevented. Third, we study the total number of infected and the maximum peak during large outbreaks using a combination of analytical estimates and numerical simulations. Also for large outbreaks we find a similar compensation mechanism as for the epidemic threshold. This means that if there is little incentive for social distancing in a population, drastic quarantining is required, and vice versa. Both pure scenarios are unrealistic in practice. The new models show that only a combination of measures is likely to succeed to control epidemic spreading. Fourth, we analytically compute an upper bound for the total number of infected on adaptive networks, using integral estimates in combination with a moment-closure approximation on the level of an observable. Our method allows us to elegantly and quickly check and cross-validate various conjectures about the relevance of different network control measures. In this sense it becomes possible to adapt also other models rapidly to new epidemic challenges.

Long-distance distribution of qubit-qubit entanglement using Gaussian-correlated photonic beams

We investigate the deterministic generation and distribution of entanglement in large quantum networks by driving distant qubits with the output fields of a nondegenerate parametric amplifier. In this setting, the amplifier produces a continuous Gaussian two-mode squeezed state, which acts as a quantum-correlated reservoir for the qubits and relaxes them into a highly entangled steady state. Here we are interested in the maximal amount of entanglement and the optimal entanglement generation rates that can be achieved with this scheme under realistic conditions taking, in particular, the finite amplifier bandwidth, waveguide losses, and propagation delays into account. By combining exact numerical simulations of the full network with approximate analytic results, we predict the optimal working point for the amplifier and the corresponding qubit-qubit entanglement under various conditions. Our findings show that this passive conversion of Gaussian into discrete-variable entanglement offers a robust and experimentally very attractive approach for operating large optical, microwave, or hybrid quantum networks, for which efficient parametric amplifiers are currently developed.