Tunable Accuracy Research Articles

Using small database and energy descriptors to predict molecular thermodynamic energies through mediated learning models

Delta machine learning (DML) models have paved a new way to obtaining high fidelity ab initio simulation results of materials by using quantities with lower computational cost as learning materials. However, the low out-of-sample extrapolative ability and the requirement of large training sets have limited broader applications of conventional DML models. In this work, we proposed the concept of non-trivial electron energy, an intermediary energy quantity decoded from the electron total energy but exhibiting high Pearson's correlation with various thermodynamic energies, to build up mediated machine learning (MML) models. By hybridizing the intermediary non-trivial electron energy (N) with a bond descriptor (B) and a spatial matrix (S) of organic molecules, our integrated NBS descriptor shows excellent predictive power of thermodynamic energies with errors close to 1 kcal/mol for MML models when trained by a database with 100 entries and tested by a database with 500 entries. Moreover, adding supplemental sets with 10 ∼ 20 entries into the original training set could greatly improve the out-of-sample extendibility of NBS MML models, such as the molecules with obviously larger size, with disparate bond-type, and even with different elemental compositions. The method of mediated learning provides alternative ways to breakthrough limitations of traditional DML models and can be applied conveniently to study formation enthalpy, thermodynamic energy barriers, multi-dimensional Gibbs free energy surface, and other quantum chemical quantities related to materials' internal energy, enthalpy, and free energy under various conditions at tunable training cost, prediction efficiency, and accuracy.

Non‐parametric measure approximations for constrained multi‐objective optimisation under uncertainty

AbstractIn this article, we propose non‐parametric estimations of robustness and reliability measures approximation error, employed in the context of constrained multi‐objective optimisation under uncertainty (OUU). These approximations with tunable accuracy permit to capture the Pareto front in a parsimonious way, and can be exploited within an adaptive refinement strategy. First, we illustrate an efficient approach for obtaining joint representations of the robustness and reliability measures, allowing sharper discrimination of Pareto‐optimal designs. A specific surrogate model of these objectives and constraints is then proposed to accelerate the optimisation process. Secondly, we propose an adaptive refinement strategy, using these tunable accuracy approximations to drive the computational effort towards the computation of the optimal area. To this extent, an adapted Pareto dominance rule and Pareto optimal probability computation are formulated. The performance of the proposed strategy is assessed on several analytical test‐cases against classical approaches. We also illustrate the method on an engineering application, performing shape OUU of an Organic Rankine Cycle turbine.

Reachability Analysis of Neural Network Control Systems With Tunable Accuracy and Efficiency

Reachability Analysis of Neural Network Control Systems With Tunable Accuracy and Efficiency

Implementation of Ripple Carry Adder Design for Optimal VLSI Performance

In the province of wireless receivers, biomedical equipment, and portable/mobile devices, the demand for Very Large Scale Integration (VLSI) systems that optimize space, minimize power consumption, and deliver high performance is dominant. At the core of these systems, adders play a pivotal role as the primary component of arithmetic units. This paper proposes a ripple carry propagate adder with a high-speed area-efficient. The two’s complement representation is employed for adding two numbers, a common practice in various systems dealing with n-bit input sequences. Microprocessors and digital signal processing use these carry adders. The carry output of every finished adder stage is useful as a carry input to the one behind it. This procedure is continued until the last complete adder stage is reached. Each carry output bit in an entire adder is therefore rippled to the next stage. An exact (forward) carry adder be linked with the proposed structure to generate hybrid adders with tunable accuracy levels. The practical implementation of proposed model is carried out using Xilinx ISE Design Suite 13.4, showcasing its feasibility for real-world applications.

An Ultra-Low-Power Non-Uniform Derivative-Based Sampling Scheme With Tunable Accuracy

This paper presents an ultra-low-power non-uniform sampling scheme using a derivative-based algorithm that can maintain a comparable accuracy to other non-uniform sampling schemes but with less complexity and lower power consumption. In this method, the change in the derivative of the signal above certain threshold values is used to identify high signal activity for retention of the significant points of the signal. The scheme is implemented using simple building blocks that calculate and compare the change in approximate real-time derivative to a tunable threshed value that can be adjusted to obtain the desired Compression Factor (CF) and Post-Reconstruction Signal-to-Noise plus Distortion Ratio (PR-SNDR) for different signal types. Fabricated in TSMC’s 0.13 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu$</tex-math> </inline-formula> m CMOS technology and tested with real-world biomedical signals, the proposed Derivative Dependent Sampling (DDS) system consumes a maximum power of 155 nW while achieving a CF of more than 6 for an Electrocardiography (ECG) signal. By adding the proposed DDS block to a data acquisition and processing system, the non-uniform sampling can reduce the power dissipation of the entire system.

Generalizing dynamic mode decomposition: Balancing accuracy and expressiveness in Koopman approximations

This paper tackles the data-driven approximation of unknown dynamical systems using Koopman-operator methods. Given a dictionary of functions, these methods approximate the projection of the action of the operator on the finite-dimensional subspace spanned by the dictionary. We propose the Tunable Symmetric Subspace Decomposition algorithm to refine the dictionary, balancing its expressiveness and accuracy. Expressiveness corresponds to the ability of the dictionary to describe the evolution of as many observables as possible and accuracy corresponds to the ability to correctly predict their evolution. Based on the observation that Koopman-invariant subspaces give rise to exact predictions, we reason that prediction accuracy is a function of the degree of invariance of the subspace generated by the dictionary and provide a data-driven measure to measure invariance proximity. The proposed algorithm iteratively prunes the initial function space to identify a refined dictionary of functions that satisfies the desired level of accuracy while retaining as much of the original expressiveness as possible. We provide a full characterization of the algorithm properties and show that it generalizes both Extended Dynamic Mode Decomposition and Symmetric Subspace Decomposition. Simulations on multiple systems show the effectiveness of the proposed methods in producing Koopman approximations of tunable accuracy that capture relevant information about the dynamical system.

FPGA implementation of fast square root algorithm with tunable accuracy

FPGA implementation of fast square root algorithm with tunable accuracy

FPGA implementation of fast square root algorithm with tunable accuracy

FPGA implementation of fast square root algorithm with tunable accuracy

Wavelength-tunable Q-switched erbium-doped fiber laser based on a digital micromirror device.

A wavelength-tunable Q-switched erbium-doped fiber laser based on a digital micromirror device (DMD) is experimentally demonstrated. The Q-switched pulses are generated by incorporating a saturable absorption device made of graphene oxide. Stable Q-switched pulses at 1.5 µm band are obtained at a low threshold of 20 mW, corresponding to the pulse width of 7.1 µs and the repetition rate of 43.3 kHz. The maximum output power and the maximum pulse energy of the Q-switched pulses are 260.1 µW and 3.97 nJ, respectively. By controlling the DMD, the center wavelength of the Q-switched pulses can be tuned from 1528.2 to 1559.3 nm, with a tuning range of about 31 nm. The fine tunable accuracy can reach 0.08 nm by precisely controlling the DMD. Combining the filtering characteristics of the DMD with the saturable absorption characteristics of nanomaterials, the Q-switched laser with tunable wavelength is realized, which, we believe, is reported for the first time and has broad application prospects.

Improving the accuracy and efficiency of quantum connected moments expansions * *This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC0500OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce

The still-maturing noisy intermediate-scale quantum technology faces strict limitations on the algorithms that can be implemented efficiently. In the realm of quantum chemistry, the variational quantum eigensolver (VQE) algorithm has become ubiquitous, with many variations. Alternatively, a promising new avenue has been unraveled by the quantum variants of techniques grounded on expansions of the moments of the Hamiltonian, notably the connected moments expansion (CMX) and the Peeters–Devreese–Soldatov (PDS) energy functional. Common to those approaches is that, upon preparing an approximate ground state used to compute the necessary moments, the accuracy of the estimated ground state energy depends on the degree of overlap between the prepared state and the true ground state. Thus, we use the ADAPT-VQE algorithm to test shallow circuit construction strategies for the purpose of increasing the overlap with the exact ground state, validated by the sizable accuracy improvement herein reported in the PDS and CMX ground state energies. We also show that we can take advantage of the fact that the terms to be measured are highly recurring in different moments, incurring a substantial reduction in the number of necessary measurements. By coupling this measurement caching with a threshold that determines whether a given term is to be measured based on its associated scalar coefficient, we observe a further reduction in the number of circuit implementations while allowing for tunable accuracy.

The Smooth Extension Embedding Method

A new approach to the solution of boundary value problems within the so-called fictitious domain method philosophy is proposed, which avoids well-known shortcomings of other methods of this type, including the need to generate extensions of the data. The salient feature of the novel method, which we refer to as SEEM (smooth extension embedding method), is that it reduces the whole boundary value problem to a linear constraint for an appropriate optimization problem formulated in a larger, simpler set containing the domain on which the boundary value problem is posed and which allows for the use of straightforward discretizations. It can also be viewed as a fully discrete meshfree method, which uses a novel class of basis functions and thus builds a bridge between fictitious domain and meshfree methods. The proposed method, in essence, computes a (discrete) extension of the solution to the boundary value problem by selecting it as a smooth element of the complete affine family of solutions of the original equations now yielding an underdetermined problem for an unknown defined in the whole fictitious domain. The actual regularity of this extension is determined by that of the analytic solution and by the choice of objective functional. Numerical experiments demonstrate that it is stable enough to efficiently solve boundary value problems on general geometries and that it produces solutions of tunable (and high) accuracy.

Traffic routing in stochastic network function virtualization networks

Network Function Virtualization (NFV) is emerging as an attractive solution, which can transform complex network functions from the dedicated hardware implementations into software instances running in a virtualized environment. In NFV, the requested service is implemented by a sequence of Virtual Network Functions (VNF) that can run on generic servers by leveraging the virtualization technology. These VNFs are placed in a predefined order, which is also known as the Service Function Chaining (SFC). While most of the existing work on traffic routing in NFV networks assume deterministic link delay and bandwidth, the real-life network usually behaves in a stochastic manner, due to e.g., inaccurate data, expired exchanged information, insufficient estimation to the network. Motivated by this, we consider the stochastic NFV networks, where the link bandwidth and delay are assumed to be random variables and their cumulative distribution functions are known. We first study how to calculate the delay and bandwidth value in an SFC such that their realizing probabilities are satisfied. Subsequently, we formally define the traffic routing problem in stochastic NFV networks and prove it is NP-hard. We present an exact solution and a tunable heuristic to solve this problem. The proposed heuristic is a sampling-based algorithm, and it leverages the Tunable Accuracy Multiple Constraints Routing Algorithm (TAMCRA) to find a multi-constraint path for each adjacent VNF pair. It dynamically adjusts the link weights as well as delay and bandwidth realizing probability constraint after finding the path for each VNF pair so that the cumulated probabilities will not violate the specified values. Finally, we evaluate the performance of the proposed algorithms via extensive simulations.

Sublaminate variable kinematics shell models for functionally graded sandwich panels: Bending and free vibration response

An advanced kinematic formulation is applied to multilayered cylindrical sandwich panels with continuously graded layers. Free vibration and bending problems are considered. The mean mechanical properties of the composite material are estimated by means of the extended rule of mixture or the Eshelby-Mori-Tanaka method. The displacement field is postulated by means of variable-kinematics sublaminate models, therefore the applicability is not restricted to monolithic panels, on the contrary, the approach is well suited for sandwich panels with marked thickness-wise heterogeneity. Due to the efficiency of the formulation, the effect of various design parameters, either geometrical or mechanical, can be easily explored. The validation is performed against benchmarks of increasing complexities, namely a single-layer square plate, a shell reinforced by randomly oriented nanotubes, sandwich panels with three distinct configurations. The importance of allowing kinematic descriptions of tunable accuracy within a unique framework is well demonstrated by the proposed assessments.

Confining Liquids inside Carbon Nanotubes: Accelerated Molecular Dynamics with Spliced, Soft-Core Potentials and Simulated Annealing.

Understanding emergent phenomena of fluids under physical confinement requires the development of advanced tools for rapid and accurate simulation of their physiochemical properties. Simulating liquid molecules commensurate in size with the nanoscale enclosures that confine them is a key challenge. We demonstrate an accelerated molecular dynamics simulation technique that combines soft-core potentials (SCP) and simulated annealing (SA) to analyze confined liquids. This integrated SCP/SA method relies on a new spliced soft-core potential (SSCP), which enables tunable accuracy with respect to the target hard-core potential (HCP). SCP/SA enables the packing of enclosures with bulk material in a controlled, thermodynamically consistent manner. The enhanced SSCP accuracy is a critical feature of SCP/SA, enabling a smooth transition between the SCP and the HCP at a desired SCP hardness. We applied SCP/SA to the problem of filling a carbon nanotube (CNT) in periodic boundary conditions with a popular ionic liquid (IL), 1-butyl-3-methylimidazolium hexafluorophosphate [BMIM+][PF6-]. We performed a series of triplicate simulations on systems with varying CNT diameter and charge to demonstrate SCP/SA's versatility. Beyond this IL/CNT system, the SCP/SA simulation framework has a broad range of potential applications, not limited to nanoscale enclosures and interfaces, including both solid-state and biological systems.

Convergence analysis of a highly accurate Nyström scheme for Fredholm integral equations

A stable and convergent Nyström scheme is proposed to solve Fredholm integral equations (FIEs). Our approximation is based on the barycentric rational interpolants. By introducing barycentric quadratures to the integral operator that appears in the FIE and modifying the standard Nyström scheme, we demonstrate that the new Nyström scheme is a viable option for the numerical solution of FIEs. Convergence rates of the method are proved taking into account the effect of grading the domain. The final convergence result shows clearly that one can choose an optimal domain grading. Numerical examples and comparisons with competitive methods of tunable accuracy are provided to support the theoretical analysis and illustrate the efficiency of the proposed numerical scheme.

Memristive Devices: Analog–Digital Hybrid Memristive Devices for Image Pattern Recognition with Tunable Learning Accuracy and Speed (Small Methods 10/2019)

Memristive Devices: Analog–Digital Hybrid Memristive Devices for Image Pattern Recognition with Tunable Learning Accuracy and Speed (Small Methods 10/2019)

A 37.5–45 GHz Superharmonic-Coupled QVCO With Tunable Phase Accuracy in 28 nm CMOS

This paper presents a new superharmonic coupling structure for millimeter-wave (mmW) quadrature voltage-controlled oscillators (QVCOs) with high phase accuracy and zero voltage headroom. The proposed coupling network can adjust the quadrature error with almost no phase noise penalty. A tail impedance theory for superharmonically coupled oscillators is introduced to analyze the locking mechanism, phase noise, phase accuracy, and phase tunability of the proposed superharmonic QVCO. For verification, a 40 GHz QVCO is fabricated in a 28 nm bulk CMOS process, which exhibits a phase noise of −94.32 dBc/Hz at 1 MHz offset and 0.18° quadrature error with 8.4 mW power consumption and 18.4% frequency tuning range.

Simple and Efficient Truncation of Virtual Spaces in Embedded Wave Functions via Concentric Localization.

We present a strategy to generate "concentrically local orbitals" for the purpose of decreasing the computational cost of wave function-in-density functional theory (WF-in-DFT) embedding. The concentric localization is performed for the virtual orbitals by first projecting the virtual space onto atomic orbitals centered on the embedded atoms. Using a one-particle operator, these projected orbitals are then taken as a starting point to iteratively span the virtual space, recursively creating virtual orbital "shells" with consecutively decreasing correlation energy recovery at each iteration. This process can be repeated to convergence, allowing for tunable accuracy. Assessment of the proposed scheme is performed by application to the potential energy diagram of the Menshutkin reaction of chloromethane and ammonia inside a segment of a carbon nanotube and the torsional potential of a simplified version of the retinal chromophore.

Characterizing Tolerable Disturbances for Transient-State Safety in Power Networks

This paper develops methods to compute the set of disturbances on a power network that do not tip the frequency of each bus and the power flow in each transmission line beyond pre-defined bounds. For a linearized AC power network model, we consider scenarios with varying degree of knowledge about the form of the disturbance. We propose a sampling method to provide inner and outer approximations with tunable accuracy of the set of tolerable disturbances. The complexity of computing such set approximations is a function of the number of sampling points. We introduce an error metric to measure the gap between the approximations and design an algorithm that finds, for fixed number of sampling points, the sampling sequence that minimizes its value. Simulations on the IEEE 39-bus power network illustrate our results.

Analog–Digital Hybrid Memristive Devices for Image Pattern Recognition with Tunable Learning Accuracy and Speed

AbstractBrain‐inspired memristive artificial neural networks (ANNs) have been identified as a promising technology for pattern recognition tasks. To optimize the performance of ANNs in various applications, a recognition system with tunable accuracy and speed is highly desirable. A single WO3−x‐based memristor is presented in which analog and digital resistive switching (A‐RS and D‐RS) coexist according to a selectively executed forming process. The A‐RS and D‐RS mechanisms can be attributed to the modulation of the Schottky barrier on the interface and the formation/rupture of conducting filaments inside the film, respectively. More importantly, a new analog–digital hybrid ANN is developed based on the coexistence of A‐RS and D‐RS in the WO3−x memristor, enabling tunable learning accuracy and speed in pattern recognition. The spike‐timing‐dependent plasticity learning rules, as a learning base for image pattern recognition, are demonstrated using A‐RS and D‐RS devices with obviously different fluctuations and rates of change. The learning accuracy/speed can be improved by increasing the proportion of A‐RS/D‐RS in the crossbar array. A convenient method is provided for selecting an optimized pattern recognition scheme to meet different application situations.