Min Of CPU Time Research Articles

A Low-Complexity Deep Learning Model for Predicting Targeted Sequencing Depth from Probe Sequence

Targeted sequencing has been widely utilized for genomic molecular diagnostics and the emerging DNA data storage paradigm. However, the probe sequences used to enrich regions of interest have different hybridization kinetic properties, resulting in poor sequencing uniformity and setting limitations for the large-scale application of the technology. Here, a low-complexity deep learning model is proposed for prediction of sequencing depth from probe sequences. To capture the representation of probe and target sequences, we utilized a sequence-encoding model that incorporates k-mer and word embedding techniques, providing a streamlined alternative to the intricate computations involved in biochemical feature analysis. We employed bidirectional long short-term memory (Bi-LSTM) to effectively capture both long-range and short-range interactions within the representation. Furthermore, the attention mechanism was adopted to identify pivotal regions in the sequences that significantly influence sequencing depth. The ratio of the predicted sequencing depth to the actual sequencing depth was in the interval of 1/3—3 as the evaluation metric of model accuracy. The prediction accuracy was 94.3% in the human single-nucleotide polymorphism (SNP) panel and 99.7% in the synthetic DNA information storage sequence (SynDNA) panel. Our model substantially reduced data processing time (from 334 min to 4 min of CPU time in the SNP panel) and model parameters (from 300 k to 70 k) compared with the baseline model.

Index tracking with controlled number of assets using a hybrid heuristic combining genetic algorithm and non-linear programming

In this paper, we discuss the index tracking strategy using mathematical programming. First, we use a non-linear programming formulation for the index tracking problem, considering a limited number of assets. Since the problem is difficult to be solved in reasonable time by commercial mathematical packages, we apply a hybrid solution approach, combining mathematical programming and genetic algorithm. We show the efficiency of the proposed approach comparing the results with optimal solutions, with previous developed methods, and from real-world market indexes. The computational experiments focus on Ibovespa (the most important Brazilian market index), but we also present results for consolidated markets such as S&P 100 (USA), FTSE 100 (UK) and DAX (Germany). The proposed framework shows its ability to obtain very good results (gaps from the optimal solution smaller than 5 % in 8 min of CPU time) even for a highly volatile index from a developing country.

Transistor-Level Synthesis of Pipeline Analog-to-Digital Converters Using a Design-Space Reduction Algorithm

A novel transistor-level synthesis procedure for pipeline ADCs is presented. This procedure is able to directly map high-level converter specifications onto transistor sizes and biasing conditions. It is based on the combination of behavioral models for performance evaluation, optimization routines to minimize the power and area consumption of the circuit solution, and an algorithm to efficiently constraint the converter design space. This algorithm precludes the cost of lengthy bottom-up verifications and speeds up the synthesis task. The approach is herein demonstrated via the design of a 0.13 μm CMOS 10 bits@60 MS/s pipeline ADC with energy consumption per conversion of only 0.54 pJ@1 MHz, making it one of the most energy-efficient 10-bit video-rate pipeline ADCs reported to date. The computational cost of this design is of only 25 min of CPU time, and includes the evaluation of 13 different pipeline architectures potentially feasible for the targeted specifications. The optimum design derived from the synthesis procedure has been fine tuned to support PVT variations, laid out together with other auxiliary blocks, and fabricated. The experimental results show a power consumption of 23 mW@1.2 V and an effective resolution of 9.47-bit@1 MHz. Bearing in mind that no specific power reduction strategy has been applied; the mentioned results confirm the reliability of the proposed approach.

Parasitic-Aware Optimization and Retargeting of Analog Layouts: A Symbolic-Template Approach

Layout parasitics can significantly affect the performance of analog integrated circuits (ICs). In this paper, a systematic method of optimizing an existing analog layout considering parasitics is presented for technology migration and performance retargeting. This method represents the locations of layout rectangle edges as variables and extracts circuit and layout integrity such as device symmetry, matching, and design rules as constraints. To ensure the desired circuit performance, bounds of layout parasitics are determined first. These bounds are used to constrain the layout geometries while retargeting existing high-quality layouts across technologies and specification sets. The problem is then solved by a graph-based algorithm combined with nonlinear optimization. The proposed method has been implemented in a parasitic-aware automatic layout optimization and retargeting tool (intellectual property reuse-based analog IC layout). Its efficiency and effectiveness are demonstrated by successfully retargeting operational amplifiers within 1 min of CPU time.

An Analytical Scatter Correction for Singles-Mode Transmission Data in PET

We present an analytical scatter correction, based upon the Klein-Nishina formula, for singles-mode transmission data in positron emission tomography (PET) and its implementation as part of an iterative image reconstruction algorithm. We compared our analytically-calculated scatter sinogram data with previously validated simulation data for a small animal PET scanner with 68 Ge (a positron emitter) and 57 Co (approximately 122-keV photon emitter) transmission sources using four different phantom configurations (three uniform water cylinders with radii of 25, 30, and 45 mm and a nonuniform phantom consisting of water, Teflon, and air). Our scatter calculation correctly predicts the contribution from single-scattered (one incoherent scatter interaction) photons to the simulated sinogram data and provides good agreement for the percent scatter fraction (SF) per sinogram for all phantoms and both transmission sources. We then applied our scatter correction as part of an iterative reconstruction algorithm for PET transmission data for simulated and experimental data using uniform and nonuniform phantoms. For both simulated and experimental data, the reconstructed linear attenuation coefficients (mu-values-values) agreed with expected values to within 4% when scatter corrections were applied, for both the 68 Ge and 57 Co transmission sources. We also tested our reconstruction and scatter correction procedure for two experimental rodent studies (a mouse and rat). For the rodent studies, we found that the average mu-values for soft-tissue regions of interest agreed with expected values to within 4%. Using a 2.2-GHz processor, each scatter correction iteration required between 6-27 min of CPU time (without any code optimization) depending on the phantom size and source used. This extra calculation time does not seem unreasonable considering that, without scatter corrections, errors in the reconstructed mu-values were between 18%-45% depending on the phantom size and transmission source used.

Practical database screening with docking tools.

As an increasing number of pharmaceutically relevant target 3D structures is becoming available, efficient techniques for exploiting the information contained in these structures gain importance over massive experimental screening (Bailey and Brown 2001). One of these techniques is virtual screening through docking calculations, i.e., the search for those compounds in a database of small organic compounds that display favorable steric as well as electrostatic interactions to the target binding site (Walters et al. 1998; Stahl 2000; Good 2001). Docking programs consist of two essential parts: an algorithm that searches the conformational, rotational, and translational space available to a candidate molecule within the binding site, and an objective function that is to be minimized during this process: a crude measure of binding affinity or receptor-ligand complementarity usually referred to as a scoring function. In order to be successful as a virtual screening tool, a docking program must be able to find optimum docking solutions for active molecules in accordance with experiment, it should be able to separate active compounds from inactive ones, and it should use no more than 2–3 min of CPU time per compound to be applicable to large libraries.

Inverse planning for interstitial gynecologic template brachytherapy: truly anatomy-based planning

Purpose : Commercially available optimization schemes generally result in an undesirable dose distribution, because of the particular shapes of tumors extending laterally from the tandem. Dose distribution is therefore manually obtained by adjusting relative dwell time values until an acceptable solution is found. The objective of this work is to present the clinical application of an inverse planning dose optimization tool for the automatic determination of source dwell time values in the treatment of interstitial gynecologic templates. Methods and Materials : In cases where the tumor extends beyond the range of the tandem-ovoid applicator, catheters as well as the tandem are inserted into the paravaginal and parametrial region in an attempt to cover the tumor volume. CT scans of these patients are then used for CT-based dose planning. Dose distribution is obtained manually by varying the relative dwell times until adequate dose coverage is achieved. This manual planning is performed by an experienced physician. In parallel, our in-house inverse planning based on simulated annealing is used to automatically determine which of all possible dwell positions will become active and to calculate the dwell time values needed to fulfill dose constraints applied to the tumor volume and to each organ at risk. To compare the results of these planning methods, dose-volume histograms and isodose distributions were generated for the target and each organ at risk. Results : This procedure has been applied for the dose planning of 12 consecutive interstitial gynecologic templates cases. For all cases, once the anatomy was contoured, the routine of inverse planning based on simulated annealing found the solution to the dose constraints within 1 min of CPU time. In comparison, manual planning took more than 45 min. The inverse planning-generated plans showed improved protection to organs at risk for the same coverage compared to manual planning. Conclusion : This inverse planning tool reduced the planning time significantly and produced improved plans with reduced dose to the organs at risk. Furthermore, the inverse planning approach improves the physician’s control over treatment. The focus becomes the physician’s prescription to the target and his or her compromise due to dose to normal structures.

Global Energy Minimization of Small Molecules Combining Constraint Logic Programming and Molecular Mechanics

The paper presents an approach to molecular energy minimization using Constraint Logic Programming (CLP) as a preprocessor to a molecular mechanics minimization program such as CHARMM. The use of CLP enables the definition of simple constraints that sufficiently describe relations in a molecule, thus limiting the search space of the globally optimal solution. The approach was used on the problem of minimizing the energy of the alanine dipeptide and N-methylalanylacetamide. An approximate 3-D structure produced by a CLP system was used as an initial structure for the CHARMM program that considers all relations in such a structure and performs energy optimization. With this initial structure, computed in about 1 min of CPU time, the optimal solution (global minimum) was found in only a few seconds, while calculating the same solution with CHARMM alone took 15 h. All calculations were performed on HP/735.

A new computer method to quickly and accurately compute steady-state temperatures from ferromagnetic seed heating.

A new, very fast, yet accurate program, MGSEED, has been developed that computes steady-state temperatures from the three-dimensional bioheat transfer equation due to heating by a ferromagnetic seed. Seeds have a self-regulating power absorption characteristic such that their temperatures remain within a few degrees of their Curie transition point. The code is also very flexible, being able to model a seed of any orientation embedded in a tissue domain that can be inhomogeneous with respect to blood perfusion or thermal conductivity. MGSEED uses multigrid (or multilevel) programming techniques as well as a finite volume discretization that exploits knowledge of the approximate shape of the temperature solution very near to a seed. These techniques allow the code to sample the seed very coarsely, requiring only one or two nodes to cross the seed. With these coarse samplings MGSEED calculated very accurate temperatures in under 3 min of CPU time on a Sun Sparcstation 2. The accuracy of MGSEED is demonstrated at different levels of perfusion by comparing its solution in a perpendicular plane that bisects the seed with the known analytical solution. The speed of MGSEED is compared to other methods of solution and it is found that MGSEED performs 14 times faster than successive over relaxation and conjugate gradient methods, and 2.5 times faster than a preconditioned (modified block incomplete Cholesky) conjugate gradient method. It is concluded that the techniques for discretization and solution incorporated into MGSEED can greatly improve the flexibility and speed of hyperthermia treatment planning, which could ultimately lead to an increased level of control over treatment outcome.

Models and algorithms for three-dimensional topography simulation with SAMPLE-3D

Algorithms for general surface advancement, three-dimensional visibility, and convolution over a surface have been developed and coupled with physical models for pattern transfer. The resulting program, SAMPLE-3D, allows practical simulation of plasma etching and deposition processes on engineering workstations. The physical models are 3-D extensions of 2-D string and segment based models. The models include secondary effects, such as material density variations and damage enhanced etching. A general facet motion algorithm supports simple, isotropic, cosine-directional, and general surface orientation dependent processes. A 3-D grid of rectangular prismatic cells, which is updated by the advancing surface, contains an alternate topography representation for fast shadow and visibility calculation. The program is organized as a collection of modular functions for continued model and algorithm development. Guidelines for estimating CPU and memory requirements for various models and simulation cases are based on an analysis of the algorithms and data structures. Simple processes, such as lithography development, require 1-5 min of CPU time. Simulations involving integration over flux distributions, such as plasma etching and sputter deposition, require from 5-30 min for typical cases. Reflection or surface migration calculations require from 30-60 min. Physical memory of 4-32 megabytes is sufficient for many practical simulations. >

Efficient simulation of incompressible viscous flow over single and multielement airfoils

The incompressible, viscous, turbulent flow over single and multielement airfoils is numerically simulated in an efficient manner by solving the incompressible Navier-Stokes equations. The solution algorithm uses the method of pseudocompressibility with an upwind-differencing scheme for the convective fluxes, and an implicit line-relaxation scheme. The motivation for this work includes interest in studying high-lift takeoff and landing configurations of various aircraft. In particular, accurate computation of lift and drag at various angles of attack up to stall is desired. Two different turbulence models are tested in computing the flow over a NACA 4412 airfoil; an accurate prediction of stall is obtained. The approach used for multielement airfoils involves the use of multiple zones of structured grids fitted to each element. Two different approaches are compared: 1) a patched system of grids, and 2) an overlaid Chimera system of grids. Computational results are presented for two-element, three-element, and four-element airfoil configurations. Generally, good agreement with experimental surface pressure coefficients is seen. The code converges in less than 200 iterations, requiring on the order of 1 min of CPU time on a CRAY YMP per element in the airfoil configuration.

A fast large-signal model for coupled-cavity TWTs

A fast and accurate one-dimensional large-signal model for coupled-cavity TWTs has been developed that can be used interactively for the optimization of a TWT design. Different modeling techniques for enhancing the speed of the large-signal model with no loss in accuracy are discussed. It has been shown that the speed of the model can be increased by (1) suitable selection of the basic integration parameters such as the number of electron discs per RF cycle and number of integration steps per cavity, (2) choosing a suitable method of integration of the relativistic equation of motion, (3) optimizing the iteration processes through each cavity and each section, and (4) efficient calculation of the space-charge forces. The model has been tested for the low-space-charge high-efficiency NASA CTS 200-W tube and for a high-space-charge high-gain tube. The characteristics of a 58-cavity tube for a single value of input power and frequency can be computed in less than 1.5 min of CPU time on a VAX 11/785 computer. This is at least 5 to 6 times faster than the previous models with the same accuracy. >

Computation of viscous flows in turbomachinery cascades with a space-marching method

A space-marching method was used to predict the viscous flowfield in turbomachinery cascades. Departure solutions were suppressed by modifying the streamwise pressure gradient term in the momentum equations. For staggered cascades a nonperiodic grid system was employed, and appropriate approximations were used in place of the periodicity boundary conditions upstream and downstream of the cascade. The laminar flow through a flat-plate cascade at 45-deg stagger and through a symmetric cascade at zero incidence and stagger was tested first. The method was then used to predict the turbulent flow through compressor cascades composed of NACA 65-series blades. The predicted drag coefficient, turning angle, boundary-layer momentum thicknesses, and velocity profiles were compared with the experimental data; the agreement was good in most cases. The solution of the cases presented was obtained in less than 1 min of CPU time on an IBM 3081 computer.

A non-parametric clustering scheme for landsat

A 4-dimensional histogram is computed to reduce the large LANDSAT pixel data (up to 7.6 million pixels to a frame) to the much smaller number (6,000) of distinct vectors and their frequency of occurrence in the scene. The vectors are clustered by a recent non-parametric clustering algorithm (3) using the histogram count as a probability density estimate. The resultant clusters are unimodal m the 4-dimensional histogram and can possess arbitrary shapes. The algorithm is non-iterative and does not require specification of the number of clusters a priori. Hashing is used to generate the histogram and also subsequent table look-up classification of the individual pixels in the image after the histogram vectors are clustered. The resultant clustering scheme is very efficient and a 512 × 512 LANDSAT scene can be clustered in less than 2 min of CPU time on a PDP-10 computer. Results of the application of the clustering scheme on representative LANDSAT scenes are included.

A non-self-adjoint finite difference model of electron diffusion parallel to electric fields

The electron distribution function plays a key role in the quantitative study of various properties of electrons moving through a gas and is characterized as a solution of the Boltzmann transport equation. Simplication of this model results in a non-self-adjoint second-order elliptic partial differential equation with variable coefficients. The complexity of the coefficients precludes the use of a coordinate transformation to remove second-order cross-derivative terms, and the nonmodelness of the boundary value problem necessitates rather novel applications of various tools of numerical analysis for successful solution. Due to the non-self-adjointness, any consistent finite difference model is bound to be nonsymmetric; however, a finite difference scheme is devised which is “almost symmetric.” Iterative solution of the resulting difference equations is analyzed and an acceleration strategy for convergence is devised. Numerical results are included for a typical problem involving in excess of 5000 Mesh points. This problem was solved on the UNIVAC 1106 in less than 6 min of CPU time.