Non-ideal Effects Research Articles

3-D computational fluid dynamics modeling of a hollow fiber membrane contactor ozonation process

Ozonation processes have been successfully used to remove organic micropollutants. Membrane contactors can transfer the ozone via a bubbleless process, making them promising devices for many applications thanks to a good control of the transferred ozone. Here we studied the local transfer of ozone into water through a single and a multi-fiber membrane contactor by developing a 3-D model in a Comsol Multiphysics® workflow. To our knowledge, this is the first model to combine 3-D analysis and a multi-fiber membrane contactor for an ozonation application. The model established here could serve as a startpoint for optimizing bubbleless processes and scaling up membrane contactor processes, including the ozonation of micropollutants. Dedicated experiments were performed to check the model. Using of a 3-D modeling approach makes it possible to account for the non-ideal effects that occur in hollow fiber membrane contactors. The simulations highlighted hydrodynamics as a very important factor to ensure good distribution of ozone and therefore good treatment of pollutants. Moreover, the transferred ozone does not increase proportionally with increasing membrane length or residence time. Finding the right compromise between contactor length and fibers number can help optimize the contactor geometry and thus make the process as effective as possible.

Accounting for non-ideal mixing effects in the hydrogen-helium equation of state

Accounting for non-ideal mixing effects in the hydrogen-helium equation of state

Electronically Tunable Current Mode Instrumentation Amplifier Employing Single DV-EXCCCII

In this paper, a new design of a current mode instrumentation amplifier (CMIA) is presented using a single differential voltage extra-X current controlled current conveyor (DV-EXCCCII) and a grounded resistor. Therefore, this circuit design is very simple and suitable for IC integration. The proposed circuit of instrumentation amplifier has high output impedance so it is easily cascadable with other current mode circuits without buffer requirement. Furthermore, the proposed circuit is electronically tunable with the bias current of the active element used. This circuit offers good frequency performance with wide differential gain and CMRR bandwidth. The non-idealities and parasitic effects of the proposed circuit are also investigated. To validate the proposed instrumentation amplifier, various simulations are performed using the CADENCE VIRTUOSO SPECTRE simulator using 0.18 µm CMOS technology parameters.

Achieving software-equivalent accuracy for hyperdimensional computing with ferroelectric-based in-memory computing

Hyperdimensional computing (HDC) is a brain-inspired computational framework that relies on long hypervectors (HVs) for learning. In HDC, computational operations consist of simple manipulations of hypervectors and can be incredibly memory-intensive. In-memory computing (IMC) can greatly improve the efficiency of HDC by reducing data movement in the system. Most existing IMC implementations of HDC are limited to binary precision which inhibits the ability to match software-equivalent accuracies. Moreover, memory arrays used in IMC are restricted in size and cannot immediately support the direct associative search of large binary HVs (a ubiquitous operation, often over 10,000+ dimensions) required to achieve acceptable accuracies. We present a multi-bit IMC system for HDC using ferroelectric field-effect transistors (FeFETs) that simultaneously achieves software-equivalent-accuracies, reduces the dimensionality of the HDC system, and improves energy consumption by 826x and latency by 30x when compared to a GPU baseline. Furthermore, for the first time, we experimentally demonstrate multi-bit, array-level content-addressable memory (CAM) operations with FeFETs. We also present a scalable and efficient architecture based on CAMs which supports the associative search of large HVs. Furthermore, we study the effects of device, circuit, and architectural-level non-idealities on application-level accuracy with HDC.

Spectral analysis of ultra-cool white dwarfs polluted by planetary debris

ABSTRACT We identify two ultra-cool (${T_\mathrm{eff}}\lt 4000$ K) metal-polluted (DZ) white dwarfs WD J2147−4035 and WD J1922+0233 as the coolest and second coolest DZ stars known to date with ${T_\mathrm{eff}}\approx 3050$ K and ${T_\mathrm{eff}}\approx 3340$ K, respectively. Strong atmospheric collision-induced absorption (CIA) causes the suppression of red optical and infrared flux in WD J1922+0233, resulting in an unusually blue colour given its low temperature. WD J2147−4035 has moderate infrared CIA yet has the reddest optical colours known for a DZ white dwarf. Microphysics improvements to the non-ideal effects and CIA opacities in our model atmosphere code yields reasonable solutions to observations of these ultra-cool stars. WD J2147−4035 has a cooling age of over 10 Gyr which is the largest known for a DZ white dwarf, whereas WD J1922+0233 is slightly younger with a cooling age of 9 Gyr. Galactic kinematics calculations from precise Gaia EDR3 astrometry reveal these ultra-cool DZ stars as likely members of the Galactic disc thus they could be pivotal objects in future studies constraining an upper age limit for the disc of the Milky Way. We present intermediate-resolution spectroscopy for both objects, which provides the first spectroscopic observations of WD J2147−4035. Detections of sodium and potassium are made in both white dwarfs, in addition to calcium in WD J1922+0233 and lithium in WD J2147−4035. We identify the magnetic nature of WD J2147−4035 from Zeeman splitting in the lithium line and also make a tentative detection of carbon, so we classify this star as DZQH. WD J1922+0233 likely accreted planetary crust debris, while the debris composition that polluted WD J2147−4035 remains unconstrained.

MN-ADC based on high-reliable 2T2R synapse structure

With the ever-growing demands for sampling rate, conversion resolution, as well as lower energy consumption, the memristor-based neuromorphic analog-to-digital converters (MN-ADC) becomes one of the most potential approaches to break the bottleneck for traditional ADCs. However, the online trainable MN-ADCs are not designed to be easily integrated into the 1T1R crossbar array, meanwhile suffering from the device non-idealities, which makes it difficult to realize high-speed and accurate conversion. To overcome these issues, this paper proposes a high-reliable 2T2R synaptic structure. And through the dedicated structure, we construct a 4-bit MN-ADC that allows for alternate conversions and online adjustments in a single clock period, which can significantly mitigate the effects of device non-idealities on dynamic performance. More importantly, this structure can be perfectly compatible with 1T1R crossbar arrays. Simulation results demonstrate the validity of the proposed MN-ADC, which achieves the ENOB of 3.77 bits, the INL of 0.16 LSB, and the DNL of 0.07 LSB.

An effective pseudo-potential lattice Boltzmann model with extremely large density ratio and adjustable surface tension

The pseudo-potential lattice Boltzmann (LB) model is versatile in modeling multiphase flows since the mesoscopic interaction potential enables it to directly describe the nonideal effect evading the tracking or integrating of phase interface. In this paper, we develop an effective pseudo-potential lattice Boltzmann model to simultaneously realize the thermodynamic consistency, the extremely large density ratio, and the adjustable surface tension. Decoupling the mesh space from the momentum space by a scale factor, denser lattice nodes depict the transition region more accurately. The high-precision explicit finite difference method (EFM) further enhances the calculation accuracy of interaction force. The present model is validated to satisfy thermodynamic even at very low temperature, where the liquid–gas density ratio exceeds 1010. The spurious current can be suppressed to a very low level (<0.0007) despite the density ratio reaching tens of thousands. A modified pressure tension is introduced to tune the surface tension free from the influence of the density ratio. The numerical stability of multiphase simulations is significantly improved, and the droplet splashing is successfully reproduced at Reynolds number 25 000, while the density ratio is more than 10 000.

Mitigating reverse emission in electroaerodynamic thrusters

Electroaerodynamic (EAD) thrusters have the potential to become a quieter and solid-state alternative to propellers in unmanned aircraft. An EAD thruster produces ions from atmospheric air and accelerates them across an electric field between two electrodes: an emitter, where ions are produced; and a collector, where they are neutralized. As ions drift from the emitter to the collector, they collide with neutral air molecules and transfer momentum to them, resulting in an ionic wind and a thrust force. Reverse emission from the collector is a major non-ideal effect in EAD systems that can reduce the efficiency of EAD thrusters: it increases the power draw of the device and eventually leads to electrical breakdown and sparking. Here, we perform a parametric study with a wire-to-cylinder electrode geometry using positive and negative corona discharges as well as wire-to-wire dielectric barrier discharges as ion sources. We show that reverse ion emission is primarily caused by a gas discharge at the tips of the cylindrical collectors, which increases current without contributing to thrust and produces audible noise. We identify three ways to mitigate reverse emission: increasing the collector diameter, increasing the collector span with respect to the emitter span, and attaching hemispherical end caps to the collector. By mitigating reverse emission, we show that the power draw of an EAD thruster can be significantly reduced (by up to 64% in our particular configuration) while still delivering the same thrust.

CMQ: Crossbar-Aware Neural Network Mixed-Precision Quantization via Differentiable Architecture Search

The RRAM-based accelerators have become very popular candidates for neural network acceleration due to they perform matrix-vector multiplication in-memory with high storage density and low latency. Many related works have used fixed-precision quantization to achieve the model compression and enhance the tolerance of process variation, but these methods still suffer a large accuracy degradation and poor robustness to nonideal effects. In this work, we propose a crossbar-aware mixed-precision quantization scheme, which enables to search for the optimal precision of each part of the network as a way to improve the accuracy and robustness to noise. First, we introduce a group quantization strategy that can flexibly adjust the group size dynamically according to the crossbar size. Then, we propose a detailed mixed-precision search flow to search for the optimal precision set of the network. Finally, we give a noise injection adaption training method to enhance the tolerance of noise. Experimental results show that our proposed method can improve inference accuracy by at least 2.04% compared to the fixed-precision quantization under the same resource cost. The searched architecture with the highest accuracy most accurate (MA) could achieve an accuracy of 92.39% and a resource saving of 93.30% compared to the full precision model. The searched architecture with the biggest resource savings most efficient (ME) could achieve an accuracy of 91.11% and a resource saving of 95.57% compared to the full precision model. And, the average precision of the ME mixed-precision architecture is only 1.4 bits. Besides, the results show the mixed-precision network with noise adaption training is more robust to noise than the fixed-precision network with noise adaption training.

A Comparison of Global Magnetofrictional Simulations of the 2015 March 20 Solar Eclipse

The solar corona exhibits a wide range of phenomena, from highly non-potential objects such as solar filaments to near-potential structures such as the open magnetic flux. For any global model to be useful in space weather applications, the model must on a single day reproduce all of these phenomena in the same simulation, using the same set of coronal physics and parameters. The purpose of the present paper is to evaluate whether the evolving magnetofrictional model can achieve this goal. Twenty-eight separate simulations are analyzed, where each tries to reproduce both highly non-potential and near-potential phenomena observed in the solar corona on the same day. This day is chosen to be 2015 March 20, the date of the solar eclipse. The study evaluates how the cadence of bipole data, ideal or nonideal coronal physics, and the variety of helicity injection mechanisms affect the accuracy of the simulations. Results show that significantly better agreement arises when using Advective Flux Transport (AFT) synoptic maps to drive the simulations, as compared to 27-day Carrington rotation synoptic maps. Using the nonideal effect of hyperdiffusion leads to the worst agreement with all coronal phenomena. Alternatively, when running either ideal or ohmic diffusion simulations with helicity condensation or bipoles with a self-helicity, a good agreement with both on-disk and limb structures can be found. This suggests that future studies aiming to simulate the corona and reproduce multiple phenomena on a given day should use data products such as AFT and avoid using the nonideal physics of hyperdiffusion.

Shock tube and multi-species laser absorption measurements of rate constants for methanol pyrolysis

Quantitative and high-accuracy measurements of multiple rate constants regarding the reaction kinetics of methanol pyrolysis were performed in shock tube experiments over 1200 - 1610 K, 0.72 - 4.50 atm. The key reactions investigated include both the thermal dissociation of methanol and its major fragments (for example, CH2OH and CH3O), and their H-abstraction by H and CH3 radicals. Laser absorption time-histories of CH2O, CO and H2O were measured and analyzed with a Bayesian approach for systematic determination of the joint posterior distribution of their rate constants, using FFCM-1 as the base mechanism. Various types of uncertainty sources in the current measurements, including both the non-ideal effects in the shock tube and uncertainties in the mixture composition and the absorption model, were thoroughly analyzed. A comparative analysis using other reaction models was also briefly discussed.

Electronically Adjustable Multiphase Sinusoidal Oscillator with High-Output Impedance at Output Current Nodes Using VDCCs

This paper presents the high-output-impedance current-mode multiphase sinusoidal oscillators (MSO) using voltage differencing current conveyor (VDCC)-based lossy integrators, which consist of one VDCC, one grounded capacitor, and two grounded resistors. The proposed oscillator can provide an odd-phase and even-phase system without the use of an additional amplifier. The frequency of oscillation (FO) is electronically tuned via the bias current without affecting the condition of oscillation (CO). The proposed oscillator is designed to obtain three-phase sinusoidal waveforms (n = 3). The effect of non-idealities of VDCC on the lossy integrator section is also investigated. The validity of the proposed circuit is demonstrated by PSPICE simulation using 0.18 µm TSMC CMOS process parameters with ±0.9V power supply. The frequency of oscillation obtained from the simulation is 1.43 MHz. The total harmonic distortions of the sinusoidal output currents IO1, IO2, and IO2 are 1.22%, 1.18%, and 0.57%. The IO1–IO2 and IO2–IO3 phase differences are approximately 121 and 119 degrees, respectively. The feasibility of the proposed MSO is also verified with experimental results using the VDCC constructed from the commercially available ICs (LT1288 and AD844) with ±5 V power supply. The results of PSPICE simulations and experiments are closely consistent with the theoretical expectation.

Exploration of nonideal effects in a beam model of a fuzzy structure

The presentation that preceded this one analyzed a model of a fuzzy structure as a beam with multiple suspended SDOFs. It examined the degree to which phenomena encountered in a model of the master as a rigid body are replicated when the base displacements of the attachments are unequal. The conclusion was that many response features are replicated in the frequency domain, whereas contributions of higher structure modes cause time domain responses to differ. The present paper examines whether the numerous idealizations that led to that conclusion are crucial. Parameters that are examined are tuning and bandwidth of the fixed base SDOF frequencies, and randomness of the spectral and spatial distribution of the fixed base SDOF natural frequencies. Responses extracted from an “exact” state space solution of the equations of motion show that minor randomness can significantly alter, and sometimes obscure, the phenomena observed in the ideal case. The results bring to the fore the question of whether phenomena observed in idealized fuzzy models are relevant to actual structures.

Energy-efficient neural network design using memristive MAC unit

Artificial intelligence applications implemented with neural networks require extensive arithmetic capabilities through multiply-accumulate (MAC) units. Traditional designs based on voltage-mode circuits feature complex logic chains for such purposes as carry processing. Additionally, as a separate memory block is used (e.g., in a von Neumann architecture), data movements incur on-chip communication bottlenecks. Furthermore, conventional multipliers have both operands encoded in the same physical quantity, which is either low cost to update or low cost to hold, but not both. This may be significant for low-energy edge operations. In this paper, we propose and present a mixed-signal multiply-accumulate unit design with in-memory computing to improve both latency and energy. This design is based on a single-bit multiplication cell consisting of a number of memristors and a single transistor switch (1TxM), arranged in a crossbar structure implementing the long-multiplication algorithm. The key innovation is that one of the operands is encoded in easy to update voltage and the other is encoded in non-volatile memristor conductance. This targets operations such as machine learning which feature asymmetric requirements for operand updates. Ohm’s Law and KCL take care of the multiplication in analog. When implemented as part of a NN, the MAC unit incorporates a current to digital stage to produce multi-bit voltage-mode output, in the same format as the input. The computation latency consists of memory writing and result encoding operations, with the Ohm’s Law and KCL operations contributing negligible delay. When compared with other memristor-based multipliers, the proposed work shows an order of magnitude of latency improvement in 4-bit implementations partly because of the Ohm’s Law and KCL time savings and partly because of the short writing operations for the frequently updated operand represented by voltages. In addition, the energy consumption per multiplication cycle of the proposed work is shown to improve by 74%–99% in corner cases. To investigate the usefulness of this MAC design in machine learning applications, its input/output relationships is characterized using multi-layer perceptrons to classify the well-known hand-writing digit dataset MNIST. This case study implements a quantization-aware training and includes the non-ideal effect of our MAC unit to allow the NN to learn and preserve its high accuracy. The simulation results show the NN using the proposed MAC unit yields an accuracy of 93%, which is only 1% lower than its baseline.

Solubility analysis of 18 active pharmaceutical ingredients and intermediates in the non-polar solvents dioxane, toluene and cyclopentyl methyl ether

The temperature dependent solubility of 18 organic compounds, of which some are active pharmaceutical ingredients, was measured in 1,4-dioxane, toluene and cyclopentyl methyl ether over a temperature range between 0 and 70 °C using a turbidity method. The solubility of the studied compounds in the three solvents increases with temperature in all binary solute-solvent systems, and all compounds exhibit a higher solubility in 1,4-dioxane than in toluene and cyclopentyl methyl ether. The non-ideal behaviour of the binary systems was evaluated in terms of the activity coefficient. All the compounds in toluene presented a positive deviation from ideal solution, while both positive and negative deviations were observed for the model compounds in the other two apolar solvents. From the solubility and the activity coefficient, a number of thermodynamic models including van’t Hoff, Apelblat, λh, Margules, van Laar, Wilson, and the Non-Randomness Two Liquids (NRTL) were used for the correlation of the experimental data. The NRTL and the van’t Hoff equations present closer overall correlations, while the Margules and the λh models show deviations >5 % for several compounds in the three solvent systems. The mixing thermodynamic properties of the several solute-solvent systems were further analyzed by the van ‘t Hoff and Wilson equations for a better understanding of the solution behaviour. The analysis revealed systems with apparent ideal behaviour where counteracting non-ideality effects cancel each other’s.

Performance enhancement of coherent modulation imaging in the presence of missing data.

Coherent diffraction imaging (CDI) has become a powerful imaging modality in synchrotron x-ray imaging and electron microscopy communities. In the far-field geometry, image quality of CDI depends strongly on the performance of detector; specifically, the dynamic range, pixel size, and the absence of missing data. Coherent modulation imaging (CMI), an innovative variant of CDI, improves the algorithmic convergence by inserting a modulator upstream of the detector. Here, we explore the potential of CMI in eliminating nonideal effects of detector by modifying the modulus constraint to extrapolate the missing part of diffraction pattern. Nine folds of extrapolation in area of diffraction pattern have been shown feasible in experiment; while sixteen folds in simulation. For image quality measured by Structural Similarity (SSIM), our method shows a maximum of 32% improvement over the traditional method. Our method provides a way to alleviate the effects of beamstop, gaps between modules, limited dynamic range, and limited detector size for CMI.

Drastically reduced sensor hardware for solvable chaos‐based sonar

An acoustic ranging demonstration is presented in a reflective environment using solvable chaos with novel and practical hardware advances. Previously, solvable chaos was optimally detected using ideal alignment, random codes and an 11 element microphone array for environments free of multipath reflections. In this work, ranging experiments were performed with drastically reduced sensor hardware and measurements were extended to include non-ideal effects such as multipath and angle misalignment from reduced line of sight (LOS). Importantly, this new scheme not only reduces the size and power of these systems by over 83%, a novel triggering technique allows for triggering on targeted bit sequences and produces sidelobe levels that are independent of oscillator dynamics. Ranging results from a closed, reflective cavity obtained using a 10 kHz solvable chaotic oscillator are reported. Measured sidelobe levels were typically between −1.80 dB and −15.85 dB. The resulting demonstration shows successful ranging despite drastically reduced hardware and environmental imperfections.

A 14 μJ/Decision Keyword-Spotting Accelerator With In-SRAMComputing and On-Chip Learning for Customization

Keyword spotting has gained popularity as a natural way to interact with consumer devices in recent years. However, because of its always-on nature and the variety of speech, it necessitates a low-power design as well as user customization. This paper describes a low-power, energy-efficient keyword spotting accelerator with SRAM based in-memory computing (IMC) and on-chip learning for user customization. However, IMC is constrained by macro size, limited precision, and non-ideal effects. To address the issues mentioned above, this paper proposes bias compensation and fine-tuning using an IMC-aware model design. Furthermore, because learning with low-precision edge devices results in zero error and gradient values due to quantization, this paper proposes error scaling and small gradient accumulation to achieve the same accuracy as ideal model training. The simulation results show that with user customization, we can recover the accuracy loss from 51.08\% to 89.76\% with compensation and fine-tuning and further improve to 96.71\% with customization. The chip implementation can successfully run the model with only 14$uJ$ per decision. When compared to the state-of-the-art works, the presented design has higher energy efficiency with additional on-chip model customization capabilities for higher accuracy.

A Microring as a Reservoir Computing Node: Memory/Nonlinear Tasks and Effect of Input Non-Ideality

The nonlinear response of an optical microresonator is used in a time multiplexed reservoir computing neural network. Within a virtual node approach combined with an offline training through ridge regression, we solved linear and nonlinear logic operations. We analyzed the nonlinearity of the microresonator as a memory between bits and/or as a neural activation function. This is made possible by controlling both the distance between bits subject to the logical operation and the number of bits supplied to the ridge regression. We show that the optical microresonator exhibits up to two bits of memory in linear tasks and that it allows solving nonlinear tasks providing both memory and nonlinearity. Finally, we demonstrate that the virtual node approach always requires a comparison of the reservoirs performance with the results obtained by applying the same training process on the input signal.

Stability of hydrogen molecular ion in non-ideal classical plasmas

The effects of non-ideality (NI) of plasma on the energy levels of the hydrogen molecular ion () have been investigated by computing accurate binding energies of the ion under non-ideal plasma. The screened interaction potential in the plasma has been described by a pseudopotential which is derived from a solution of Bogolyubov’s hierarchy equations by ignoring the effects of quantum degeneracy. Changes in the energy of the ground state and the first excited state along with the expectations of some important quantities associated with the ground state of due to the NI of the plasma varying within a wide range have been studied in details. Special emphasis is made on the stability of the ion under varying NI of plasma. Moreover, appearance of the Borromean window and its changes with varying NI are also been discussed in details.